The Magic of Universal Analytics: Strategy, Tactics, Implementation Tips

PossibilitiesIn a Q&A after a keynote a couple of years ago, I was asked: "When will traditional business analysis subsume the web analytics silo?"

My reply: "All business will ultimately be digital, so, if anything, web analytics will subsume business analysis!"

That was a half-cheeky reply. But, if you reflect upon the developments in analytics over the last couple of years it is incredible to see that we, web analytics, have moved so quickly towards the aforementioned outcome.

In fact, even the term digital analytics is too stifling. It is all just business analysis – with digital being a dominant factor in influence (marketing, advertising, experiences, connections, relationships et. al.), digital plus real world owning outcomes (of the commerce type) and some facets of influence.

Business analysis. No digital. No web. No offline. No just this or that silo.

And, 15 years later I get to go back to my first job title after graduating from MBA school. Senior Business Analyst! : )

So, in our world, web analytics, what is helping us embrace to this change? Moving us away from our digital only silo? A little something, that Google Analytics calls, Universal Analytics.

It was announced to the world perhaps 24 months ago – in classic Google fashion, with a bold vision that was not fully baked. Gotta love those betas! The team at Google, thanks to that bold vision, has continued to invest time and people, and execute quickly. Universal analytics has been out, in proper fully baked production release, for a little while. It has exciting new features, an exciting cluster of new analyses you can do, and a lot that was impossible before. It allows you to be a full-blooded Business Analyst.

Yet, heart-breakingly, few people understand what Universal Analytics is and the power you have at your disposal. I want to try and fix today. Here's a run-down of what we'll cover:

And get this, even though by the end of this post you'll be crazy insane excited about what you can do today with Universal Analytics… I tend of think of the current stage as iPhone 3. There is so much more to come.

Universal Analytics: Why don't you love it more?

Three important reasons that we (vendors, practitioners, leaders) should internalize:

1. Universal Analytics is an extensible platform that allows you to literally send any data in, take any data out, and get a complete view of your business – inside Google Analytics or outside Google Analytics, your choice. When something is that open-ended and scalable, it is actually hard to understand.

2. There is no Universal Analytics folder in GA. No standard reports you can just go grab once you update your site code to analytics.js . This makes it a bit harder to wrap your head around all the possibilities. Big problem for Google to solve. For example, this Universal Analytics overview is about as exciting as an over-ripe banana.

There is a partial excuse for this. Because you can do anything with the platform's features, how does one show specific reports/possibilities? Still. Notice I said partial. :)

3. Universal Analytics is seeped in technicalese. If you are a regular reader of this blog, : ), then you are more than likely fall in the business side of the house rather than the technology side of the house. A lot of Universal Analytics is technical – see the over-ripe banana above. Not since we started parsing weblog files have we had to take these many code showers . This makes Universal Analytics less accessible.

So, if you've heard of Universal Analytics, and done nothing, don't blame just yourself. It is a tough topic to internalize. But that is what we'll address here.

And, my hope is to do that, by trying to reframe value in a business context, with as little technicalese as possible. Just layout, in simple business terms, what the essence is, and how you can massively benefit from it. My goal is to make you dangerous enough to be specific with the people who will help you (like a GACP:

At the end, I'll link to a few wonderful folks who will help you with deep technical solutions. You I'm not going to leave you dangling on the deep dives.

Universal Analytics: Why does your business need it?

Simple. You had a partial view of your business success in the past. Just what happened on your desktop website. For a while, that was ok.

Now, that partial view has become a partially partial view of digital behavior. And, this trend is massively accelerating. Ouch, ouch, ouch.

First, desktop became desktop and laptop. Still most people picked one. Then desktop and laptop got further fragmented to mobile behavior (via mobile phones, tablets and laptops). Then mobile behavior on mobile devices got further fragmented into m.sites and mobile apps.

The data you have in WebTrends, Google Analytics, Adobe, IBM is a terrible representation of reality. Take just one fact into consideration: For most commercial websites mobile traffic became a majority source this past Christmas, it will never look back. Scary, for when you log into your Adobe Analytics account.

One of the most minor implications: I'm now sixteen Unique Visitors as I engage with the company Adobe across it's digital presence. Regardless of if they are using Adobe Analytics or Google Analytics.

The other big reason for Universal Analytics is this, the column in the middle:

us retail commerce ecommerce sales

The data is from the Q4 2014 Quarterly Retail E-Commerce Sales from the US Census Bureau. They show, like any other report from anyone else you want to look at, that E-commerce is a small percentage of our commercial activity. And yet, study after study does show that a massive amount of time is being spent on digital channels and lots of influence is shifting away from traditional channels to online.

One of the most minor implications: Your digital behavior is driving out-sized multi-channel impact. You're not getting credit. You're not getting promoted!

There are many other reasons to invest in Universal Analytics (my favourite is to finally get a cohesive view of the consumer journey), but those two are the biggest ones.

Universal Analytics: What the heck is it?

The problem/benefit of being an extensible platform is that you get to be anything anyone can imagine. You basically get a collection of empty boxes. Different people will do different things with them.

But. This is a post about sucking up and being crystal clear. Ignoring the little things and some upgrades to current features, here are three things in simple business terms that Universal Analytics is:

1. It is the acceleration of moving from tracking Visits to tracking People to tracking Customers.

With current web analytics tools, you track cookies. That is why Adobe believes I'm sixteen people across it's digital existence (three different people just on my desktop!). In that scenario, we place an inordinate emphasis on Visits (and atrocious and awful decisions like defining Conversion Rate by using Visits). Not any more. We can understand one person, and that person's behavior uniquely as they become a loyal customer.

#cookiesareforeating #freethecookies!

consumer journey

2. It is the shift from tracking individual device silos to tracking cross-device behavior.

Each instance of your Adobe/Google/IBM code on various types of your digital presence (desktop site, mobile site, mobile app, or even your app on my phone and your app on my Glass) creates a new data silo. Not just because of your code treats that as a unique instance, but also because you might be using multiple tools.

Universal Analytics allows you to standardize on one vendor platform, across devices, and to unify customer behavior tracking across multiple silos.

3. It empowers, finally, the single-view of your online and offline datasets.

You can't compute true profit today. You can't truly do lifetime value calculations. You can now do both.

You can't really track where the person who just redeemed a coupon in your store came from. You can't really take your offline data about me (family person, xx age, loves to buy from your catalog) and optimize my online experience. You can now do both.

There is a lot more. But, the salient point is that you can get the single view of your customers (subject to your prioritization, of course).

To pull all this together with our earlier example: With Universal Analytics first you can shift to tracking people (so for Adobe, I go from sixteen to one!), you can track behavior across all devices (Adobe's multiple properties – desktop or mobile, and their myriad mobile apps etc.) and then you can shift to tracking customers (Adobe can now understand my behavior in the real world and, possibly, their multitude of real world partners!).

Did you just have a datagasm? #omg

It really is that exciting.

Universal Analytics: Talk techie to me!

These three things, Visits to People to Customers, Individual Device Silos to Cross-Device Unified, and Online Silos to Online plus Offline, are enabled by three core Universal Analytics features. For the rest of this post, I'll anchor the abilities of Universal Analytics to revolutionize your digital everything, by focusing on these three features.

They are:

1. User-ID Override – one world with people, not cookies!

2. Dimension Widening – hello sweet simple data from spreadsheets, data warehouses/CRM systems!

3. Measurement Protocol – all your data are belong to us!

For the rest of this post, we'll explore these three features from a business value perspective. We'll throw in just a pinch of technical details (remember, I want you to be dangerous enough).

Universal Analytics: T1: User-ID Override!

This is were cookies die. Perhaps, not for everyone who engages with you digitally. But as many people as you can encourage/guide.

The reason for this feature is simple enough. You'll get it right away.

Reusing our image from earlier in this post… this could represent on person's experience with your brand:

consumer journey

The very first time someone, let's call him Ola Hanø from Oslo, engaged with you, was via a DoubleClick delivered programmatic ad on Ola's tablet. That, at a later point, resulted in Ola visiting your website on his desktop computer. He also buys a product from you a few days later. A couple months later he revisits on his laptop to buy a service you are selling. Your products delight him, and he decides to follow you on twitter, download your app and make merry with you via his phone.

How many people is this in your digital analytics tool?

consumer journey cookie based

At least four. Assuming nothing happened on the desktop between Deep Dive and Purchase, in which case it will be five or six or seven.

You start with a Norwegian man and end up with a female swimmer.

I know, a bit melodramatic. But, trust me, the reality inside your digital analytics tool is worse.

But fear no more, me harties! (Did I just switch from a viking reference to pirates? I did! :))

You can take advantage of the Universal Analytics User-ID override feature to combine all the hits from these multiple visits/sessions and identify them with the same person.

With every interaction Ola Hanø has with you, you can send a unique ID that will override other identifiers (ex: Client-ID used in the cookies), and allow you to see this instead…

consumer journey user id based 2

How sweet is that! You can keep track of Ola throughout his journey.

Think about the deep understanding you can get of your customers, and the profound implications on marketing ROI and where you spend your budgets, content creation, multi-channel touch-points, and everything I'm forgetting.

So, how does this work?

As always, you first enable the feature inside your Google Analytics account:

user id setting

It is a three step process. Review policy, set up the User-ID, Create a User ID view.

If your web platform is using PHP, the snippet of code you would implement on the site would look like this:

google analytics user id code snippet

[I'm sorry, I've not yet figured out how to add a snippet of code to WordPress without it getting messed up. I know. Lame.]

Someone in your technical team (or a consultant you hire), or perhaps you, will set up the associated elements to empower GA to collect accurate User-ID across your digital presence.

Here's a detailed guide courtesy of the Google Analytics team:

A couple of important things to be aware of. The User-ID is assigned by you and not by Google Analytics. It is passed into Google Analytics via a User-ID parameter in the tracking code (this part's easy, above code). The User-ID cannot contain any Personally Identifiable Information (no PII!), thus falling into the standard GA TOS. You can use a 64 bit hash key as one of the simpler options for your User-ID.

Simply stated, you and only your company know who the actual person is.

What do you get if you go through the sweet joy of implementing User-ID? Here are some of what you will see:

User-ID reporting

The ability to look at users/visitors with User-ID, and those without. Overtime, in this case, the blue should grow allowing you to understand customer behavior more optimally. But even if this is all you see, you can use the people in the blue as proxy for other visitors to answer questions you simply can't answer today. I wonder how many people use both our mobile and desktop sites? I wonder if people really do research on their phones and then buy on desktop? I wonder if… other such questions.

If you are going to use User-ID you MUST figure out how to incentivize your website visitors to log into the site. It would be criminal not to. Even if you don't do ecommerce, even if your name is not Amazon. This is as much as marketing problem as anything else. You need to use human psychology. If you can't get people to login when they visit your digital presence, this is only and solely and wholly a reflection of your team's inability to be clever. Get help from outside the company if you are in this camp. Do not let this go.

One company that does this well is L'Oreal Paris. They have a wealth of amazing content to get you to come back again and again. Ding! Any article or product you find on the site has a Save button that works even if you don't open an account (but once you save, you want to access it on your mobile phone and don't really want to lose it, so more incentive to open an account). Ding! The products on the site also contain a button called Have, which allows you to get a deeply personalized experience (articles, offers, beauty picks customized to you, etc) which causes you to… login. Ding! And, more ding, ding, ding! Across mobile and desktop. Insanely clever.

Don't be the kind of company that just prays to the Gods to deliver vital information. Step up to the plate. Create reasons for your users to self-identify. In the name of all that is pure and holy, deliver real value to your users!

Back to our story… You can also get reports that answer your burning questions, like how many people experience you digitally across multiple devices:

User-ID device overlap reporting

Just move your mouse and get all your questions answered. In this case, I was surprised that the overlap was so little, most consumption was on a single device all by itself. Usually I see behavior with lots of overlap. But then again, that is why data beats "best practices"!

And, of course if you get the above two, understanding the sequences related to device usage can't be too far away:

User-ID device path reporting

This is the simplest view you can look at. You also have the ability to look performance by revenue and transaction rate. You can look at the sequences and understand where people start, what do they do there (content consumption, micro-outcomes) and then their behavior on subsequent devices.

The sky is the limit.

For folks like me who've been advocating / pining for the ability to track people, User-ID is truly an incredible advance. As web analysis moves to business analysis, the lynchpin of your success will be based on understanding people, actual single human beings. Regardless of the tool you use, go invest in this (and don't forget my guidance in the sidebar above).

[Bonus Guide: If you are a bit technically oriented, investing a little time in understanding Session Unification is totally worth it. A User might not immediately identify themselves, this allows you to stitch the earlier hits – with some conditions.]

Universal Analytics: T2: Dimension Widening!

It has a impressively complex name. Don't let that freak you out. It is the simple, but powerful, ability to send data into Google Analytics to enrich your understanding performance.

Your CRM data, information from your corporate data warehouse, or even data from your CMS (content management system), which typically will be in its own silo, can now be inside Google Analytics. You can send hit level data, primarily your refund data (if people return their orders). Or extended data to expand your ability to do interesting things with your user, campaign, product and other types of data.

Let me share a simple example. Remember Ola Hanø? In GA all you might know about him is his ID:

user id dimension widening

With Dimension Widening, you can extend, literally, the data that is available in GA. For example, you can add his gender:

user id dimension widening male

And, when you have more information, you can further extend the data that is available in GA for Ola (though, note, no PII, so you will never actually send in the name Ola, you'll use an anonymous hashkey):

user id dimension widening attributes

Makes sense? Extend the data inside Google Analytics. Widen the dimensions! :)

To get going, please log into the Admin section of your GA account, select the property, click Data Import and you'll be looking at your Data Sets page.

You can import data using the Data Import option inside the Admin section, or you can use the Google Analytics Management API. Get a developer to help you set the latter up. The former can be as simple as this:

data import csv

The reports that you'll see will depend on the the data you've sent into GA. But, if we use the above example, the report would look something like this one that the GA team kindly shared with me:

dimensioning widening report1

You can see the secondary dimension allows us to analyze performance on the site, across a whole cluster of our normal metrics, by using the Education Level.

Here's another report that might be used by a travel company who has sent in data into Google Analytics related to the type of traveler:

dimensioning widening report2

Imagine how much better both your website (mobile and desktop) content strategy will be with this insight, and the impact of this type of additional data on your online and offline owned, earned and paid marketing strategy!

The team at Google has created a wonderful user guide you can read (or share with your friendly neighborhood developer) to implement dimension widening for your own business:

I'll draw special attention to this page with the list of seven Data Import Examples . For each example, you'll learn implementation step by step using an actual example, and get guidance on how to get the report you'll need to analyze performance.

A lot of our best business data is typically is not a part of any standard, or a god-can-only-understand complex, implementation of Adobe/Google Analytics/IBM or any other web analytics tool. It is sitting in your core business systems. Get it out. Make love to it. Birth smarter insights!

Universal Analytics: T3: Measurement Protocol!

Another scary name. It is a friendly beast. You are going to grow to love it a lot.

Here's the simplest explanation: When offline interactions happen you can then send the event back to Google Analytics through the Measurement Protocol, where it will be combined with the visitor’s original activity.

Now, isn't that special!

Measurement Protocol is perhaps the most complex of the major Universal Analytics features. We've ended the above two UA features with reports you can use, in this case let me start by showing you the awesomeness that awaits you.

Here's one possible report you can create after you complete the necessary implementation using the GA Measurement Protocol:

measurement protocol

You'll recognize the first and the second column, the standard Per Visit Goal Value and Conversion Rate metrics. This is all you can usually see in GA.

But, in this case we've done offline integration with our CRM platform. Hence, we are able to get three more metrics that help us understand how the lead progressed through rest of the lead nurturing process. Offline Opportunity, Offline Closed Win and Offline Closed Lost.

Imagine any dimension you would like to for this report. Mobile, Desktop, Tablet. Or, Google, Yandex, Baidu. Or, Display, Email, Search. You can easily see how you get one picture from the first two columns, and a different picture of success from the other three columns.

When I said Universal Analytics is taking you closer to business analysis, this is what I mean more than anything else! Give you a deeper view of success and then make adjustments to both your online and offline strategies based on a report such as the one above.

[Bonus: I wrote this post to transform your understanding of your business success in May 2013: Excellent Analytics Tip #24: Obsess About Real Business Profitability You can finally do this in Google Analytics with Measurement Protocol! #profitabilityrocks]

There is little that you can't integrate using the Measurement Protocol. Offline point of sale systems. People walking into Coldplay concert. A kiosk inside your supermarket, or a train station. Every kind of lead gen online to offline life cycle (students applying to your university online to the most long and complex B2B lead to conversion cycle). And so much more.

If you are a tiny bit technically savvy, it is not all that hard to understand the basics. You are given a set of structured empty boxes. Put stuff in it. Send it back to Google Analytics.

Here's a piece of simple code from the GA team:

measurement protocol code snippet

You can see how simply the structure is laid out. Here's a brief explanation:

ec (category): Will group all offline data (use “ClientOfflineConv” as the value for all hits)
ea (action): The activity that was carried out. Will change as the user changes state in the db.
el (label): The name of the event. Will change as the user changes state in the database.
ev (value): Can be set when there is a value associated with the offline data.

As you scan the code above, you can see that it is not all that complicated to understand what's going on, and how you can quickly adapt it for your business environment.

Perhaps, it is prudent to also to scare you a tiny bit about how complex this really is. But, also excite you about all that's possible when you use Measurement Protocol.

Here's an example from my dear friend, and co-founder of Market Motive and co-teacher of the web analytics master certification course, John Marshall. In this case John's integrating offline sales by making a server-to-server call to Google Analytics.

Tell me you don't find this incredibly sexy….

measurement protocol crm integration 1
measurement protocol crm integration 2

Scared or excited? I'm including it in this post to inspire you by showing how simple and straight forward it can be to take islands of data inside your company and bring them all together in GA to give you a fantastic, cohesive, view of business performance.

Today, you are making your business decisions partly right by just relying on the standard implementation of Webtrekk or comScore or AT Internet or Google Analytics. You need to step outside your digital silo and make mostly right decisions. Leveraging the Measurement Protocol functionality in Google Analytics gets you very far along that path.

As in the other two cases above, here's a wonderful user guide to get you going:

This applies to the other two as well, but for Measurement Protocol you definitely need oodles of tech savvy. Find it inside your company, then become BFFs, or find it with an external certified consultant. You can figure this out yourself with enough time, but by then you would mostly be eating your competitor's dust. And, who wants that. Expend love and/or cash to get the help you need to get going yesterday.

Universal Analytics: T4: Bonus Features!

There are some things that Universal Analytics simplifies and makes a million times easier, for example Cross Domain Tracking and yummy things like Enhanced Ecommerce which provide a whole new set of deeper insights into your money-money making efforts. You can read the Dev Guides for all these delicious features: Introduction to analytics.js.

There are other things that Universal Analytics newly empowers you to do.

For example, modify session and campaign timeouts.:

universal analytics session settings

It sounds really fun, but unless you really, really know what you are doing, or are from Mars while the rest of us are from Venus, do not touch this. I know, it sounds harsh. But, really. Don't touch this. It has big implications on many metrics, it throws off any value you might find in things like benchmarking reports. Make sure you really, really know what you are doing. And, you are from Mars.

On the other hand, do leverage features like the ability to add custom dimensions. User Guide:

You can also do clever things like Referral Exclusions (eliminate the dreaded self-referrals, for example when you use third-party shopping carts), or not-so-clever things like Search Term Exclusions (don't do it, unless you are from Uranus!).

Universal Analytics: Helpful Deep Dives!

Let me close by sharing some super-valuable examples, if you want to dive deeper into this topic.

My peer, and author of numerous books on Google Analytics, Justin Cutroni has written extensively on this topic. Justin's Universal Analytics posts. Read everything.

Here's a lovely Universal Analytics Cheat Sheet by Nicolas Blexrud with handy references for configuring the code. You, or your developer co-worker/friend, will find this to be extremely handy. Save. Print. Breathe. Live.

The wonderful Simo Ahava writes about this topic frequently, and knows what he is talking about. Simo's Universal Analytics posts.

Speaking of integrating different platforms, Allaedin Ezzedin, has a detailed post on integrating Google Analytics and Salesforce data. Totally awesome.

When you feel you have digested this post, please see the work of our peers above. You'll be glad you did.

Closing Thoughts.

I could not be more excited about leaving the digital part of my job title behind. It is all just business analysis. In my case, primarily in service of earning customer love across See-Think-Do-Care .

The three features outlined above, User-ID Override, Dimension Widening, and Measurement Protocol, allow you unshackle the limitations of your systems to break free from siloed thinking and siloed decision making. Not day after tomorrow, not five months from now, not next year, but now. You can achiever that at this very moment.

I hope you'll seize the opportunity. Not only is there glory and success for your business, there is personal satisfaction and fame for you as well.

And, to think that we are still just at the iPhone 3 stage. So much more is yet to come. You are running out of time. Carpe Diem!

As always, it is your turn now.

Is your business currently using Universal Analytics, or the equivalent version of it from Adobe/Webtrekk/IBM? What barriers do you face inside your company, or clients? Which of the three major game-changing features above have you implemented already? If you've tried User-ID already, what were your challenges, what blew your mind when you received the data? Do you believe we are headed towards unified business analytics?

Please share your answers, feedback, critique, best practices, and helpful guidance via comments below.

Thank you.


  1. 1

    Great post!

    I think I am changing my job title now from Web/Digital Analyst to Universal Analyst !!

    • 2

      Lars: Ha, ha!

      I absolutely love it. It has a nice ring to it: Universal Analyst. Or, this sounds very Star Trek, Universal Data Analyst ("I translate Big Data into Big Freaking Insights!").


  2. 3

    Hi Avinash, thank you for great reminder about Universal Analytics.

    First of all we're implementing a lot of userID features right now, extending data for ecommerce and uploading custom data through measurement protocol, so it's not that bad for this time.

    I think that more often biggest problem is in mind of HIPPO peoples that didn't understand what kind of advantage Universal Analytics is giving. Hopefully it should change in the future ;)

    But still we're scarred by how many of GA installations are just naked/basic code pastes in website header…

    Best and have a nice day!

    • 4

      LM: You are right, there is an opportunity to convince the Hippos to invest in doing this upgrade. One thing you can do is that there are ideas in this post that you can use to quantify the impact on marketing and customer experience.

      Failing that, you can make a cost argument ("if we seriously invest in this, it will reduce the waste in our marketing, impact our CPA, truly reduce tech support on phone, etc. etc."). Or, you can try to make the ROI argument (increases in multi-channel outcomes, better and more effective sales, etc. etc.).

      It is not easy, but not all that hard either. [More ideas in this post about computer the Return on Analytics – ROA]


  3. 5

    I have been out of business for two years. But now, this… this will get me back in!

    Thank you Avinash, it's all I wanted to say :D

  4. 6

    This is a bit advanced for me (okay, it's actually FAR advanced for me), but I'm wondering if this would be useful for a site like ours – an Extension gardening site. We are simply sharing information – there's really no "conversion."

    We don't sell anything (or even give anything, besides the gardening/landscape/water conservation information). We use GA, but at what's probably the saddest, most basic level ("What page is getting the most views, and for how long?").

    • 7

      Jennifer: It is totally if it is a bit advanced (and I promise you, I tired to make this as simple as I possibly could!).

      I would say that you should for sure switch to analytics.js and move your pipeline to Universal Analytics data processing. (Find both in the Admin section of your account.) This will mean you are on the best code, and get access to new features not in the old code.

      Then, move from your current basic analysis to doing the ten things recommended here: Google Analytics Tips: 10 Data Analysis Strategies That Pay Off Big!

      Only after you have mastered that you should move to mastering User-ID, as your first step into Google Analytics. At the minimum, it will help you better understand your most loyal users (across devices) – this is absolutely precious for any information/content website.

      Good luck!


  5. 8

    Never thought about the analytic data beyond the common features mostly used in SEO sectors.

    But after the millions updates (GOD knows what is cooking inside Google from last six or more months!), I felt the real dig should be done inside the analytics. I'm on it now and your blog is the best start I can get.


  6. 9

    First off thank you for producing some great thinking points.

    My major question and problem I find is the implementation of UserID's. Lets be real honest about it. Why won't Google just create the UserID's for us versus us implementing a login or our own cookie that creates a hash? Google knows who most these people are already. OK ok privacy and all that but why won't they create the unique identifier… This is where we will really get to take data to the next level.

    The one thing I've been trying to solve is how we can connect all the session data from individual users into the CRM. Right now, it seems like most technology out there in the Marketing Automation and CRM space only cares about last interaction models because it's the source that creates the leads and what feels the MA/CRM based off a hidden field. We completely lose all the data before that last interaction that makes them a lead. Ok sure there are a few tools that have lead scores and try to do reverse ip look ups to match anonymous sessions to the ip of the known lead once they fill out a form but that's not good enough most of the time. Why can't Google just solve this multi screen problem with their own unique identifier for each session vs us trying to match hashes to userID's?

    I'd love to hear what you think is the best package or solution for tying all GA's users sessions into a CRM and being able to see how all the different marketing channels played a role in sales completed. It seems like we still rely only on the last interaction model when pushing leads into a CRM versus all the ones before hand.

    How sexy would an Assisted Conversions report with Custom MCF Channel Grouping look inside a CRM?!

    • 10

      John: You are raising good points. It is important to point out that I'm a Marketer and an Analyst, and not a lawyer. But, let me attempt to at least share a broad view of the challenge.

      "Google knows who most these people are already", is not really true.

      Google "knows" as in it can have a first party cookie on, say, this is useless once you leave

      Google "knows" as in a third party cookie that is used by, say the Doubleclick ads platform. But, these are anonymous cookies (and third party are very fragile), and any way these don't connect with your first-party Google Analytics cookies set on your behalf.

      Google "knows" as in Google+, for example, is not really "knows" for so many reasons.

      I'm not even touching on the fact that Google does not collect PII data, as you know in the Google TOS to use for non-Google purposes (so you can submit a lead to buy Google Apps accounts, and submit your name and all that, but it is not going into the ad tracking or any such system).

      So "Google knows" is a bit more complex than you imagine, and not a solution.

      I believe the best solution is that you know your customer by name, and all other attributes, and you only pass an anonymous hashkey to GA (or Adobe or IBM or Webtrekk). Good for privacy. Good for users (yours). A bit harder for tracking, but all for good causes.

      The latter part of your comment is sort of a solvable problem. Allaedin's post I link to at the end of my post above has one way to do it. If your scenario is a bit more complex, please hire him (he's a consultant!) or a different consultant you like.

      You are right though, it would be very sexy!


  7. 11


    Great job of taking the finer points I have been thinking about with Universal Analytics and not only clearly articulating them. But here they are amazingly organized in a way that will enable us to share these thoughts with others.

    My brain now feels as thought it has had dimension widening applied to it.

    Now that you have cleverly grabbed my user id, I look forward to what you come up with to enrich our experiences. :^)

  8. 12

    I think I'm missing something here from a technical perspective. How are visit 1 and visit 2 being linked?

    Are we assuming that Mr. Viking clicks through the ad and logs into our website both times?

    Is Google doing something behind the scenes to link Mr. Viking on his tablet to Mr. Viking on his desktop computer?

    Unless one of these two things is true, I don't understand how the users will be linked.

    • 13

      Or are we assuming that Mr. Viking logged into our website at some point on the various devices, so we know they are him even if he is not logged in?

    • 14

      Audrey: Thank you for asking the question.

      In the picture, Mr. Viking logs-in and is id'ed in the first visit, and then all subsequent sessions by him, across devices, can be connected back to him (as he is logging in and taking benefit of, say, our personalization features).

      There is another feature called Session Unification. This assigns the User-ID in some scenarios when initially the person might have not logged in, or assigned the User-ID. Please see the two pictures and examples here for more:


      • 15


      • 16

        Hi Avinash,

        Great article and very usefull comments! Here is my follow up question:

        If you use client id (without user id), you track all the interactions of the given browser and if the path until the conversion consists of several touches of the browser, you can then analyse the contributions of each campaign to this conversion. (Is it right?).

        Then you use user id and as it happens you start to collect data with user id not from the first touch (during the first touch no email was submitted to you). User provides you with email (sign up conversion) only during the second touch on the same browser and here is what happens

        – your client id reports show better data about the campaigns which made the sign up conversion compared to user id.

        – your user id reports show better data about what happens after the sign up conversion (activation conversion, purchase conversion).

        As a result in your user id report you don't see the source of sign up, however it exists somewhere in the depth of GA. Is there a way to unify those?

        Thank you!

        • 17

          Evgeny: The scenario you are describing falls under the lovely topic of Session Unification.

          Here's a page with details on what you can do in that context in Google Analytics (and what you can't):


          • 18

            Avinash, thank you for your answer — but it's not about session unification…

            With session unification ON — you will see in your user id view the hits made before login for the session, but not the information about the previous sessions… and the information about campaigns in the previous sessions.

  9. 19

    Great article (as usual), however I also think that Google could have done more in order to help users transition from stander to Universal analytics.

    I have to admit that change is necessary, but sometimes is just difficult to make your audience understand benefits from this change specially when you are relying on users authentication to assign an ID and identified "people" instead of "visits".

    • 20

      Oscar: Please see my detailed reply to John. Google doing it by itself, is a much more complex issue than you might imagine.

      I'm also rather pleased at how easy the team has made the transition to UA. You simply update the code on your site to analytics.js and press a button in your Admin area and you are migrated! And, you get upgraded features (like enhanced ecommerce and others) instantly.

      I absolutely take your criticism that perhaps the team at Google can do a lot more to make this topic simpler, to perhaps create interactive videos, to write simpler 5,000 word blog posts (wait, that's this one!), and other things. The team can, and should do more.


  10. 21
    Caroline says:

    Great Post !!!!!!!

    I feel like I understand Universal Analytics for the very first time. I have read all the posts on the official Google Analytics blog and was so confused about what to do.

  11. 22

    This is super amazing!


  12. 23

    Thank you for another insightful post, and thank you for the mention at the end of the post!

    I see Universal Analytics as a true evolution, and not just a feature-rich upgrade. When I think of (or when I'm asked) what is the logical next step for digital analytics (that's _digital_, not _web_ anymore), I like to imagine a conference presentation I'll be giving in 5 years from now which is titled "Remember Cookies? What Were We Thinking?" :-)

    UserID is not perfect. Like you emphasize, it requires an incentive to log in. Logging in carries such an enormous baggage with it that it transcends UX (to which its unpopularity is most often attributed) and becomes more a question of "How can we truly bring added value to our visitors without sounding like greedy spammers who'll just sell the login information for their own nefarious purposes?". On the plus side, it also transcends browser cookies as the tool of choice for implementing _state_ in a web browser, which is great (but introduces a whole new level of the privacy discussion).

    But the very idea that we're actively working towards negotiating the disconnect that browser cookies place on our analyses is comforting, and leads me to believe that there really is a solution to making our metrics more meaningful.

    The benefits of UA are very technological, as you write. That's why I think it's important that everyone should familiarize themselves with tag management solutions such as Google Tag Manager or Adobe DTM. A solid TMS takes a lot of the burden away from the user, and hides many of the technical requirements under a user interface and tag templates. It doesn't mean that the user should be complacent and have an "I don't care"-attitude towards web technologies, but it does help with the transition from being a casual user to being a power user. And Universal Analytics is truly a power user -tool, if there ever was one.

    Here are some of the things I use Universal Analytics for:

    – Fixing attribution problems with third-party payment platforms by excluding the portal domains in the Referral Exclusion List (with asynchronous GA you had to add code)
    – Adding Demographics and Audience reports with one button click in Google Analytics settings!
    – Linking offline campaign sales (in a store, for example) with the online channel that delivered the user to the business. This can be done by using a downloadable (or an app) coupon, which also holds the client ID of the user. When the user then goes to the store, the coupon is scanned, and the purchase is attributed to the same client ID using the Measurement Protocol!
    – Storing amazing, value-adding information in Custom Dimensions (only 20 in the free version of GA, unfortunately). Some examples: true hit timestamp, weather conditions at time of visit, login status, userID, personas, etc.
    – Using Enhanced Ecommerce for tracking content! Turn your blog posts into products, and look at how users interact with your articles across the whole session (and not just the checkout funnel).

    It's still not perfect, but it is an evolution. The technical improvements are plenty, but the whole idea of data import / dimension widening is, though prototypal now, a preview of things to come: true data integration across online/offline, mobile/desktop, web/Internet of things, etc.

    And this makes me very excited about the future :-)

    Thank you again for the great post, Avinash!

    • 24

      Simo: I'm so appreciative of your insightful comment.

      While it is fair to consider leveraging User-ID to be a pain in terms of getting people to log in, there is a part of me that believes that if a company can't figure out how to create that incentive then does it even deserve that valuable customer data? Or, even if they did, what could they do to benefit from it if they can't even come up with an incentive!

      I know. It is a little mean. But, come on!

      I love your list of five, it will inspire even more people. And, like you, I can't wait for UA to reach the iPhone 6 stage. :)


      • 25
        Doug Harrington says:

        Witnessing my two favorite superhero analytics ninja's sharing an Einstein/Eddington moment? Priceless.

        To think like Avinash, implement like Simo, and present like Fishkin — us digital marketers can transcend beyond what analytics currently means to modern business.

  13. 26

    Thank you very much Avinash for this mind blowing post!

    John's PHP script is great. I'm taking it for my personal uses :)

    If it interested you, here is the best open source library (by the pros from AnalyticsPros) I've found to send hits to GA with the measurement protocol:

    And for pasting scripts in WP without plugins you can read this post:

    Thanks :)

  14. 27
    Ravindra Reddy Chitla says:

    Hi Avinash,

    Fantastic post as usual.

    We know that User ID only works for logged in users…how about non logged in users?

    Are they still report as sessions? can't we see user level data for those visitor who are not logged into a website?

    • 28

      Ravindra: Yes, there is never any loss in tracking.

      The minimum you get is what already exists: Unique Visitors (or Users) and Visits (Sessions).

      If you implement User-Id, you get something even better.

      To understand what it looks like, please see the image of the Anonymous vs. User-ID Enabled Sessions report above.


      • 29

        Thank you Avinash. I appreciate your time

        I had a thought for long time, how we can get the user level data for those visitors who are not logged in. I know it's not a complete solution but at least it gives us 70 to 80% of the data.

        All we can do is pairing the logged in data of other applications like gmail & to the user who is using a particular website.

        Ex: Let's say I logged into my gmail account on a browser. My gmail is always logged in, so browser knows that what email id I am using. Now I opened a eCommerce website & browsing through the website without logging into it. So, browser or cookie can read my login information from gmail and tag my behavior on website with some random user id (a random helps us to differentiate the data from the real user-id')

        Same thing can happen on my phone. I am always logged into gmail or other email client on my phone, so the browser knows my email id. It can do the same by pairing my behavior with websites using a random id.

        Then we can sum up my behavior on different devices using that random id's

        Avg users per each device is coming down…as everyone has their own device these days. So, it gives us the confidence that we are only tracking single user behavior rather a multiple users

        I am not sure about the legal implications with this approach…but it's a possibility

        What do you think?

        • 30

          Ravindra: It is a interesting suggestion.

          But, it might not happen any time soon because this crosses corporations, sites, privacy policies, data storage/retention/sharing issues, government regulations, and so many other complex rights and legal challenges.

          At the moment, our only option is to use the feature-set in Universal Analytics to do this for our business ourselves.

          It will be an incomplete view, but as we've discussed in other comments here and in the post, if you solve that marketing/customer value exchange challenge, you will be set for many more things than just single person analysis.


  15. 31

    Per usual, your post is full of a lot of good information to digest.

    As much as I appreciate the new universe of data, I find many of the websites I work with don't have enough traffic for me to make great use of the new data available. Demographics for example, thinking from a statistically relevant perspective. Plus, I'm not sure I fully trust the info from Google at this point. I tend to trust the demographic data from social sources a little more.

    At what point do you start to feel there is enough data (traffic) to make good judgements?


    • 32

      Rosh: You might be conflating a couple of different threads.

      The demographic data you see in Google Analytics, in the Audience > Demographics folder, comes from third-party. Please read this page: Google Analytics Demographics Data.

      I'm quoting from there: "Demographics and Interests data comes from the third-party DoubleClick cookie (for web traffic) and from anonymous identifiers for mobile apps (i.e., Advertising ID for Android and IDFA for iOS). Neither analytics.js nor ga.js collects Demographics and Interests data."

      For very small websites, it is possible that this data can't provide a strong enough signal (they simply don't have enough visitors to sample from).

      But that has nothing to do with this post on Universal Analytics.

      The data you'll collect via Google Analytics in the User-ID will be "first-party" (as in you will collect it yourself based on your implementation), and the data you will pass via Dimension Widening will also be your data that you have collected from your signed-in users or your customers. In that case, it will be as good as what you've collected. It also implies that you can expend effort, even for a small site, to collect it accurately.

      I hope this helps!


  16. 34
    Sanchit Aneja says:

    I have been absolutely loyal to Adobe Analytics to drive all the Business decisions, but this article has forced me to think about switching back to Google Universal Analytics again.

    Awesome post Avinash!

    The timing could not have been better for this. Majority of the population here in India shop through mobile (some e-commerce firms get 90% traffic from smartphones), and it is essential to tie the customer journeys (across different devices) together to derive actionable insights.

  17. 35
    Shannon says:

    I just finished a 10 month contract architecting and managing a larger Enterprise GA Universal Premium implementation. Not surprised, but the bulk of my time at the beginning was really trying to sell it to ALL stakeholders and management. This was NOT an easy task. They wanted a standard implementation when I was hired (not Universal)…."what really?", but I was finally able to sell them on Universal. I had to break it all down, use case by use case and show them mock up reports.

    I do agree that Universal is a BIG sell and not many people (even Google themselves) are not selling it well.

    Universal has ridiculous potential but I don't think many organizations are ready for it and some HIPPOS still get in the way of others being able to see the light of Universal. I was blessed to be on a project where we implemented UserId tracking and some additional custom dimensions for user types, Internal promotions and enhanced commerce (checkout steps, add to cart and lists). These brought huge value to the reporting ability and the business analysis.

    But I am a huge advocate of importing wonderful data like persona's from the datawarehouse and refunds for those who did not make it past the 7 day trail ….I am still trying to sell it to the client though and perhaps I have to try harder in this area. I'm loosing ammunition and energy though and feel like I am talking to a wall. Perhaps the client's stakeholders are just overwhelmed with it all?

    • 36

      Shannon: First, you deserve a huge amount of credit for your perseverance, and despite the odds doing the right thing for the client. Well done!

      I'm afraid I've not had the privilege to meet you, and I don't know your client. So nothing below might be what is happening with your client, but let me share something nuanced but important from my experience.

      Most of the time when I see this challenge (such painful amount of effort with company leaders), I notices that the brave analysts are focused too much on DC, perhaps only a little DR, and almost no DA. This is the problem. We need to not be too obsessed with data capture, we need to get out of data reporting as soon as we can (with as little of it as we can get away with), and very fast move to data analysis.

      The reason is that the DA stage leads us to Insights, Actions, Business Impact (IABI) effort by the digital analyst. And when we deliver IABI (rather than Universal Analytics or Adobe Data Warehouse or whatever feature), no leader can resist spending money on these efforts – because they can clearly see the value.

      I even recommend: Implement something for four weeks, deliver value from it for eight weeks (and do no more implementation at that time). Then, if you delivered IABI successfully, do another month or a bit more of implementation, and then 2x or 3x time of analysis and IABI, and so on and so forth.

      As we deliver value at every step, we earn the right to ask for more money, and the management cannot refuse! And, there is never a scenario where they are overwhelmed, because we constantly pause and deliver value.

      Again, I hope to get to know you IRL, but I don't and I don't know who your client is and above is just my experience with different companies.

      All the best!


      • 37
        Shannon says:

        Thanks Avinash,

        I'm with you 110% on moving from DC to DA. I do have a challenge with this though because for a few projects within larger companies (especially the most recent one which I cannot name) I am brought in for implementation architecture only and not analysis as a contractor (I do stress to the client that I have had a fair bit of analysis in my previous roles in companies and that is my value on getting the right things tracked right from the get go by defining a good KPI/KPO framework).

        I have thrown in mini analysis (with deep dives) here and there of a months reports after DC implementation, but within very large organizations, I find you can be on deaf ears with the many stakeholders in the many silos who are stuck on reporting or just HIPPOs or I get a sense they feel threatened that I am an auditor calling out tactics that they may be doing wrong or may not want to change. I am usually on my way to another project after the 4-9 month implementation shortly after, so I don't have much time to convince them over time like someone or an evangelist who would be in-house.

        I realize that the implementation side is in demand (DC) and bigger sites are a mess and need it done right, and that being said, I think I need to be more persistent and add on mandatory analysis into the contract (even if padded within the implementation side) and try to show them the way. I feel like that is my responsibility but feel like I hit a wall on not breaking the barrier for showing them analysis. Or maybe I should try to get smaller clients who don't have as much silo'd departments and ones that are willing to be taught analysis more!!

        Not sure, this is a challenge that I do have to figure out a better strategy for.

        BTW- I was one of your Market Motive students a few years ago… would love to meet you in person.

        • 38

          Shannon: I cannot mean this enough, you are doing everything you should and then some. I'm glad that where possible, you are throwing some DA in. I hope that will open some minds.

          From the situation you describe, it looks like a corrosive environment. People doing something wrong, bosses covering their butts related to analytics by just funding implementation, and only hiring someone with such good skills as yourself for a very short time.

          In such scenarios, do the very best you can by providing some DA, force them, if you can, to create a Digital Marketing and Measurement Model (it will clarify what's actually important). But, over the long term… When possible, and it might not always be, look for better companies. And ideally, get the contract from the person who is responsible for making changes to the site (they will appreciate DA and IABI).

          Every single day on Planet Earth a whole new cluster of companies and Sr. Executives discover the staggering stupidity of only focusing on DC, and not massively funding DC (with an eye to asking for IABI). So, your client pool, my client pool, is getter bigger. :)


  18. 39

    Hi Avinash.

    Great read as every. It would though help to understand where some of the charts are located in UA. For example the user ID vs none User ID. Also the device diagram (those who use what device).

    In relation to user ID, noted you have placed it differently than what google says
    " ga('create', {{GA Tracking Code}}, { 'userId': {{datalayer userid}}});"

    Any reason for that?


    • 40

      Found them Avinash. So ignore my comment on that bit.

      UA has certainly opened the door to "real" user information and away from the aggregated nonsense that senior management crave.

  19. 41

    Hi Mr. Kaushik, I'm grabbing front-end text label after users logged in and save it as the USERID with GTM for my WordPress site.

    Create a macro of type DOM Element and use the DOM Element id to locate the userid.

    This requires no or minimum backend change.

    Hope it helps.

  20. 42
    Supriya says:

    Hi Avinash,

    I am quite encouraged to read your first few lines which says that Digital Analysis is like business analysis with more emphasis on 'Digital'.

    Currently my conventional Market Analyst job has nothing to do with digital analysis, but I am quite enthusiastic about entering into this space but didn't know where to start off. So your videos, interviews and blogs are such a big inspiration to me as you can make this complex looking digital world so simple and understandable.

    Thanks so much for sharing your insights and thoughts :)

  21. 43
    Foundaspace says:

    I love the way you explain complex ideas in your article. It is easy for me to understand.

    Thanks for sharing.

  22. 44

    I'm pretty happy to live in Germany, where this technique isn't allowed. We respect privacy here! Sounds terrible being spyied across devices in order to get lot of spam ad's or "last time you thought about buying this and that". Would you like to have a shadow writing down every little thing you make during your shopping tour in a mall?

    • 45

      Frank: Thank you for sharing your perspective with us. I respect it. Please allow me to clarify a couple of thoughts…

      It is important to point out that you have to opt into Universal Analytics, and so do your Users. Please see my replies to John and Ravindra (the second one). You'll see that you have to know this as a company, you have to get your users to opt-in by logging-in.

      And, there is no PII (Personally Identifiable Information) allowed.

      I also assure you that German websites can implement User-ID, Dimension Widening, and Measurement Protocol. It is not against the law.

      One last thing. This is a bit technical. You are right. If you view a product y on site x, that site can use a web advertising platform to show you an ad later on site z for product y. This process uses a third-party cookies. (Which by the way you can easily block in your computer's browser like Chrome or IE or Firefox.) Two thoughts on this: 1. Nothing in this blog post has anything to do with this. 2. It is quite ok to do this in your country.

      As always check with your local lawyer if you need more help with government regulations.

      Nothing above is to change your mind about what you believe, my goal is to simply provide you with information you might find to be of value.


  23. 46

    Very interesting post.

    Thanks for the practical tips, presented so simply.

  24. 47

    another great post! thanks for sharing this.

    quick follow up question.

    if you set internal links (i.e. one blog post linking to another blog post) to open in a new tab (i.e. target="_blank"), does this count as a new session if a user clicks the link?

    i like to set internal links to have this property because i write very lengthy blog posts and i like visitors to be able to click a link to another article and still be able to read their current article without losing their place. this is how i use the internet so to me, it provides the best ux.

    looking at bounce rates on the site, they appear VERY high (mid 80s) yet time on site is high (5:00+). it appears that if a user clicks a link and it opens in a new tab, it starts a new session. when they then close the article they were reading, it triggers a bounced visit (assuming that was the entry page).

    is this correct?

  25. 49
    Mahesh Dobhal says:

    Good post indeed about universal analytics.

    I plan on aggressively leveraging your recommendations.

  26. 50
    Arvind Dabas says:

    I have reached here from blog hubspot while searching marketing analytics resources, and i found this is great website and people getting big advantage like you have discussed about bounce rate problem and demographic data.

    Thanks for giving so much input.

  27. 51

    Hi Avinash,

    You have really nailed the pseudo analytics professionalism of ours. Really, I fear to learn analysis and say very little about its capabilities to clients. It is so advanced that I have to close my established shop and go to school to understand and practice it.

    Hope this article will inspire me to chew the basic and advance google analytics.


  28. 52

    Hi Avinash,

    Hope you're doing well!

    Could you please help me with setting up scope for custom dimensions?

    I had set Page Level scope for my custom variables in Classic GA earlier, so now which scope from User/Session/Hit should I use to match that?

    Your words will be of great help me to move to Universal Analytics, thanks :).

    Kind Regards,

  29. 55
    Vidhya Sriram says:

    Wow! You are a lifesaver. T

    his post is clear heading in a lot of ways – But a lot to think about and the comments raise the bar! I need to re-read this.

    You do Amazing work Avinash!

  30. 56
    Muhammad Talha says:

    Hello Avinash,

    This is an awesome post. I had the datagasm when I saw what dimension widening could actually look like. Just too hot to…

    I have never implemented universal analytics but I surely will. Not in the near future but for sure. I have a few challenges and maybe you can shed some light in the dark tunnel i am walking in right now.

    i work in a big company that operates businesses in nunerous verticals (Food, medicine, constructuon, tractors etxetc). The country all the businesses operate in is quite far behind from the rest of the world. And all sales happen offline.

    so two questions. how to collect and unify user level data when every macro conversion (sale) happen offline? The scale would be huge.

    Question 2) do you think it is even wise to try unify data across different businesses types?

    Again Avinash… awesome post.

    Enjoyed every word of it.

    Peace .

    • 57

      Muhammad: The answer to question two is likely no. Simply because it does not seem for these unique businesses digital is a strong part of their current strategy or consumer experience. If it is not, we are likely better off just giving them standard GA insights and motivating them to create a stronger digital strategy.

      To answer the first question… Two things are critical. First, there has to be a way to incentivize people to log-in (self-identify). Second, a more advanced implementation where we can use the API and Custom Dimensions to send data into GA (for the micro or macro outcomes). This is totally doable if there is a will. (Which you can create by doing the above part first! :))


  31. 58

    Wow what a great post Avinash! Thank you for taking the time to put this article together. You raise some really excellent points especially about user id – we have got to check this out!

    I will certainly look forward to reading more of your articles in the future!

  32. 59
    Camilo Forero says:

    Hi Avinash,

    Hope everything is going well.

    This is an amazing and very comprehensive post. I have been following your posts, and find them very entertaining and informative. I can’t imagine a better way to address this kind of topics, especially Universal Analytics. Thanks for this.

    This is the first time I comment on one of your posts.

    The company, that I work for, has implemented Universal Analytics, passing a non-identifiable User ID generated by our CRM. We are enjoying the benefits in terms of reports and insights that are generated in Google Analytics.

    Couple of days ago, a question arose from the marketing team regarding the possibility to extract data behaviour from Google Analytics using the User ID that we have passed to Universal Analytics. I have seen comments around this across Internet, saying that it is possible to query the Universal Analytics API for specific data using the User ID as a key, but I haven’t been able to find examples of this or even confirming that this is actually true.

    If you don’t mind, I wanted to check with you if retrieving information from Universal Analytics using the API and the User ID as a key to retrieve user’s behaviour from Google Analytics is actually possible? I have seen examples on passing the User ID in one of the custom dimensions and then use this custom dimension as a key to query the Universal Analytics API, but this will imply passing twice the User ID to Universal Analytics and more hits, which I don’t see it as the best approach.

    Any insights on this would be greatly appreciated.


    • 60

      Camilo: User ID is not exposed in Google Analytics reporting.

      The simple way around this is to pass the User ID in the tracking code as a User-level custom dimension (in addition to setting the User ID field).

      Then you can query for this custom dimension.

      Good luck!


  33. 62

    Very informative post, thank you so much for sharing it with everyone!

  34. 63
    Michael Leedy says:

    OK… so in the realm of Analytics… I have a question regarding when should we start paying attention to the data that's being formed (when we're starting off a site from scratch)?

    Why I ask, is because this is EXACTLY where I'm at, and I was thinking… well, let's get some content into the blog & start playing with the syndication strategies that I've learned… so that I get let some results build up (giving something to look at… versus getting lost in a rabbit hole of trying to fix something that may not be broken… but since it's so new… i didn't know).

    Does this question make sense?

    Thank you for sharing this post on universal analytics. Much appreciated.

    • 64

      Michael: You should start paying attention the day after you implement analytics!

      The type of decisions you'll make will change with time, and more data, and their scope will also expand.

      So, right from Day One you can track the performance of your campaigns (search, social, email, and more). You are spending money, you are getting data, see what's working and what's not.

      From Week One you can track your structured experiences, like cart/checkouts and lead gen forms. These many you money. Even if you have small amounts of traffic you can find interesting patterns.

      From Month One you have enough data to start looking at content and what's working and what's not.

      In the initial month, or two, don't make earth shattering changes, you don't have enough data. Start with small decisions and then expand over time.

      For a bit more on how to go from start to glory, see the Analytics Ladder of Awesomeness in this blog post:

      Digital Marketing And Analytics: Two Ladders For Magnificent Success


  35. 65
    Antonio says:

    Dear Avinash,

    I hadn't found much information about tracking Users (User-ID) between web (desktop and mobile) and native apps.

    AFAIK, the way of doing it is by creating a Web or Mobile property in GA and then using that property both in your web and native app. Is that right?

    Do you plan to create other type of report for easily know which events and actions are being generated in you app and in your web version?


    • 66

      Antonio: There are a few subtle nuances that come into play here that make an answer just based on your comment a bit dangerous.

      The best option for complex GA requests is to hire a GAAC to go through the requirements and validate and recommend the right path. You'll find a list here:

      Let me add that they are quite affordable (and of course can also set up your initial reports to your requirements).


  36. 67
    Oremo Ochillo says:


    First of all I don't know how you possibly put this much time into writing a blog and still have two full time jobs and a part time job as a speaker.

    Secondly as for Universal Analytics I really love the simple visual showing how a single customer can be treated as multiple customers during various visits accessing your website from multiple platforms. UA certainly solves this problem.

    Perhaps I would say the more important issue it solves is being able to pull in offline data like from point of sale systems and integrate it with online data from you website and mobile apps.

    This really helps companies that have large physical presence, and helps to show how online behavior influences offline purchases.

    I'm sure you could probably right an entire blog post or an entire chapter in your next book about that.

  37. 68

    The requirement for people to login is where this is going to fail for many businesses.

    The excuse that marketers need to be clever enough to make people login is too trite. We know from years of research that the one thing that annoys users the most is having to create an account to use or access a site.

    So you're asking marketers to do one of the things users hate the most.

    Most marketers would tell you to simply not do that EG: "don't annoy the customer"

    • 69

      Mark: Let me push back a bit.

      First, you don't need everyone to self-identify. Even if you can get 25%, you start to make more informed decisions. Today we have zero.

      Second, open your history in the browser. Look at sites from the last 24 hours. How many are you persistently logged into? For me, I see around 80 distinct sites in the last 24 hours and I'm logged into 60% of them permanently. Ecommerce, News, Business sites, Blogs I read often (logged in as I comment), and more. In none of those cases as a customer am I annoyed, because they exchange my login for value. It would be irritating if they just wanted me to login because it only helped them!

      So, I appreciate your feedback that marketer's can't be clever. But, clearly that is not what you'll see in your own browser history. It does work! :)

      You don't need to shoot for perfection, you don't need get the world's greatest value exchange. Do something. Start small. Build from there.


      • 70


        sorry, I missed the bit about them already being persistently logged in. Even so, what happens if they log out? Do they need to log back in again before they can be tracked again?

        Also with growing concerns about privacy and security many users are actively preventing tracking and persistent logging in.

        Also, it's not true to say today we have zero info on our users. Tools like Marketo and Hubspot are already doing user ID tracking and have been for awhile.

  38. 71
    Marcin R. says:

    Hello Avinash!

    That was a great read – comprehensive and well-written :) We'd appreciate it if you could check out our article about migration from Classic Analytics to Universal Analytics.

    Here's the link:


  39. 72
    Michael says:

    Maybe its my novice-ness showing but why would you integrate your other data into GA instead of pulling the web data and User-ID into a data warehouse or even just a db linked to your CRM data?

    • 73

      Michael: Not your novice-ness!

      I've frequently stated a point of view on this blog that states…

      If you want to make broad business performance decisions, and digital's role in those outcomes, then it is best to take the relatively small subset of data you need from your web analytics tools into your corporate data warehouses.

      If you want to make broad digital performance decisions, and the role of other business activities/outcomes, then it is beneficial to take the relatively small subset of business marketing and offline outcomes data you need into your web analytics tool.

      So… It depends. :)


  40. 74

    Hi Avinash,

    your blog is damn good one for analytic people like myself – open the mind and give a whole new perspective

    i know my company submit the User-Id (from the DWH) to any user from the first hit, no matter if he logged in or not and also keep it with cookie at the user, that way when he come back again the user-id is first checked against the cookie

    now i am not sure but i also think the if a user after logged in then the user Id is overwritten to the one associated to the login details

    there are some lose holes in this i am sure, (few different user from same computer,, cross device etc.)

    but i am sure that the benefits of catching much larger population is worth it

    what you think on the process? is there right way to do it?

    i am not familiar with all the technical related component of the process but i can figure them out for the benefit of discussion


    • 75

      Nir: This is a very interesting strategy, to use an anonymous user-id across sessions until the person logs in. But it might not be all that different from anonymous cookies. Let me explain.

      Under your strategy this works per browser for all the behavior until the person logs in. Hence the person, until they are not logged in, will be different IDs in different browsers (and of course machines). Say #23 on Chrome, #24 on Tablet, #25 on Safari, etc.

      So say the person logs in on Chrome, now you know #23 is Nir. But how will you know #24 is Nir, #25 is Nir? That person would have to use the same login in those places and then you have to go make changes to your database to ensure you identify that 23, 24, 25 are all Nir.

      Essentially, this means you will be back to the current system where cookies are used as user-id. You don't have to create your own user-id.


      • 76

        thanks for the replay Avinash, exactly what i think before posting my replay, i must go find out what the exact process of issuing the user-ID when and by why, will be back to update and hear your thought

        saying that, we can assume that the vast majority of users in our company are using same login details from different environments for login (they have money under that account)
        more then that, we have related model running in DWH so we can connect different users with same details (email, address etc.) and put them under one unique account id –> maybe this what need to be used as user id?

        any way i must understand the flow much better to understand where the weak spots in what we doing today (i feel that what you said above is what happen, i see almost the exact amounts of users/session on same account with user id and without)

        • 77

          Nir: You definitely have a bunch of options that can work. If you need help, get in touch with a local authorized consultant here:

          I'll just caution you a bit on using the individual's unique account id. Don't use that because usually it is very sensitive information. In your DWH create a 64 bit anonymous hash key, use that for your id. It is much safer. A consultant can help you set it up.


  41. 78
    Reece McDonald says:

    I found this to be an informative and interesting article, thanks so much for sharing it Avinash.

    I'm going to accelerate the process of our company shifting to User ID.

  42. 79
    Alex Wegner says:

    Great… helpful information!

    My company has not truly appreciated the value of Universal Analytics, just the impact of User-ID Override is transformational. Thank you for arming me with exactly what I need to drive a new and effective strategy with our analytics implementation.


  1. […]
    The Magic of Universal Analytics: Strategy, Tactics, Implementation Tips –

  2. […]
    The Magic of Universal Analytics: Strategy, Tactics, Implementation Tips – Occam's Razor

  3. […] A visual representation of the information presented in Avinash Kaushik's blog post The Magic of Universal Analytics: Strategy Tactics, Implementation Tips. […]

  4. […]
    Anyone wanting to get their head wrapped round what Universal Analytics is all about, needs to check out this amazing post – The Magic Of Universal Analytics – from web analytics guru Avinash Kaushik.

Add your Perspective