The world of digital analytics seems to be insanely complicated.
And, yes, some of it is. Third-party or first-party cookies anyone? And, are we tracking people, devices, web browsers or whoknowswhat?
But it is a lot less complicated than you might believe. No. Really. A lot less complicated.
I led a discussion the other day with a collection of people who were brand new to the space and some who were jaded long-term residents of Camp Web Analytics. When someone played the omg, it is all so complicated (!!) card, I took the opportunity to sketch a picture of the entire ecosystem to highlight that it really was not all that complicated. The process involved slowly laying out each piece of the puzzle and how it fit the piece next to it.
By the end of the exercise there was a lovingly simple picture, and a path to glory. In this blogpost I want to share that with you.
Regardless of your experience in the space, I believe you'll find it to be of value. Even if you are in the super-jaded category, this will help you present something to your boss's boss that will get them to finally understand what you do!
Our journey to understanding, dare I say nirvana, follow these steps:
Doesn't it sound absolutely exciting? It is. And along the way you'll find helpful tips, links for deep dives, and a ravishing amount of new insights.
Ready? Let's go!
At the core of everything you will do in digital analytics is the concept of metrics. How do you define a metric: It is simply a number. That is as simple as it gets.
Your digital analytics tools are full of metrics. Averages this. Total that. Percentage that other thing.
A very special type of metric is designated to be a Key Performance Indicator (KPI). A KPI is a metric that helps you understand how you are doing against your objectives.
This implies you cannot have a KPI identified unless you know what your objectives are. For eCommerce site X, Conversion Rate might be a KPI because their current objectives are tied to reversing key business trends. At the same time for site Y, it could be Average Order Value. The key is knowing what your business objectives are.
Remember… If you don't know where you are going, any road will get you there. And you'll be miserable.
While there are no such things as blessed KPIs everyone must follow – because everyone is not executing the exact same strategy -, some metrics can never be KPIs. A good example of this is Bounce Rate. It will never be a KPI. Simply because even if your bounce rate goes from 100% to 10%, it might not have any impact on your business. Bounce rate improvement means you just got one more page view. That is great. But hardly earth shattering.
Please remember this important caveat as you pick KPIs.
Now you have your foundation, metrics and KPIs. The next layer is called dimensions . Definition? A dimension is, typically, an attribute of the Visitor to your website .
As we develop more sophistication to our measurement, they'll morph from being attributes of a Visitor to those of a Person.
Traffic sources, like keywords, referring sites, campaigns, countries, are examples of dimensions. As are names of pages or videos and devices people use to consume the content. It might seem a bit odd but the number of visits someone makes to your site is a dimension, as well as, if you are tracking it, visitors who make donations or if they are a subscriber or not.
This is not always true, but in general dimensions appear in rows in a table (in your analytics tool) and metrics and KPIs appear in columns.
We now have the key ingredients. We can start to create some lovely music, and it will come via the use of custom reports – one of my favourite features in any digital analytics tool.
Notice I did not say standard reports, those barely cross the bar of data puking. I said custom reports. And the primary reason is that I want to stress the difference between web reporting and web analysis . When you begin to use custom reports, you'll be forced to stare at a blank slate and figure out what goes on it. In order to paint your symphony, you'll be forced to talk to your leaders, to your peers, to your customers and understand what the real questions are that need answering. You, then, won't go on random fishing expeditions. You'll go on focused journeys and find the secrets you are looking for.
By this stage your immediate leadership is happy, and it is quite likely you are really kicking some serious butt when it comes to identifying the initial cluster of valuable insights. This will last between four to six months. Enjoy it (I genuinely mean that).
At the end of that time period, you'll focus on the singular thing that separates the kids from the adults. Advanced segmentation! There is nothing, and I mean nothing, more important for you to master to truly become an analysis ninja. The reason is simple. All data in aggregate is crap.
In order for you to really truly understand what is happening to your business, your customers and the yummy business outcomes, you need to be able to segment the data. You need to slice! You need to dice! You need to drill! Repeat after me: Slice, dice, drill!!
Advanced segmentation moves you from looking at the entire universe to focusing on micro-clusters in your quest for actionable insights. You can use the segmentation selector framework that outlines three broad buckets: Acquisition, Behavior and Outcomes. Each bucket has a host of specific segments for you. From there you move on to mastering Visitor segmentation, Cohort-analysis, Sequence segments, and you will feel a high at work that can only otherwise be achieved via illegal drugs.
If you are just getting started, download three of my favourite segments and go from there.
Oh, and if you mate your custom reports with advanced segments… prepare for your mind to be completely blown!
Now your core is set. Metrics, KPIs, Dimensions, Custom Reports and Advanced Segments. There is nothing else to add. All the other elements of the ecosystem will surround this core.
There are three clusters of inputs, let's start with the ones on the left.
As mentioned above, for you to move from your metrics to your KPIs (and indeed to identify valuable dimensions and advanced segments to focus on) you are going to need to know your business priorities. This one is quite straightforward.
Get these from the highest person in your company who'll talk to you. If you have to call in lots of favors, or even your family connections, to get through to the CMO or CEO, do it! It will totally be worth it. A lot of data analysis goes directly into the shredder because the Analyst was more obsessed about what they thought was interesting rather than the boss or the boss's boss's boss. Don't make this mistake.
Then comes something really interesting, regardless of the size of your business.
The next set of input will be your competitive reality. Competitors you know about and the ones you don't know about. Start by asking your CMO and CEO who your competitors are, who keeps them up at night. Then go to Google (or Yandex or Baidu or Seznam) and type in your top category queries, see who shows up in organic and paid search listings. These are your competitors (whether you like them or not).
What are their key strengths? What are they doing differently then you – better or worse? Where do they get their traffic? What do their visitor trends look like? This competitive intelligence analysis will be absolutely critical input in ensuring your business priorities are more informed, you pick the right KPIs and advanced segments, which will showcase important dimensions in your custom reports.
This won't happen every day, but the last piece of your left-side inputs will be new opportunities that spring up. Say Google goes bankrupt. Ha! Brand new landscape. You need to take advantage of it.
New opportunities could take your business in a completely different direction. At the minimum it will have a major influence on the analysis you will do. Both at a tactical and strategic level. Be prepared for it. No. Proactively hunt for it and seek it out. It will make you a much more savvy digital business practitioner.
Those are the three on our left.
The top and bottom inputs for our core are the ones you expect.
Analysts/Big Brains go on top. (They do, don't they? :))
Tools go at the bottom. Ok, not bottom as in bottom. Bottom just as the bottom part of the schematic we are creating.
From the person, me, who created the 10/90 rule all the way back in May 2006 (#omg), it should not be surprising that the importance of the tool is a bit smaller than that of the Big Brains.
As the 10/90 rule for Magnificent Web Analytics Success states: If you have $100 to invest in making smart decisions with data, invest $10 in the tool and consulting required for implementation and invest $90 in Analysts/Big Brains .
No matter how big your data, no matter how powerful your tool, you stray too far from the 10/90 rule and the promise of big data will never materialize.
People matter. Smart people matter even more. Tools just help them. Not the other way around.
That completes three sides of the our picture. You might be worried that thus far all you have see are inputs. Yes, very true. But it takes a lot to get to the good stuff!
No matter how much you and I might wish otherwise, the first set of outputs will be pure and unadulterated data pukes. Let's just embrace that fact and just bathe in its slightly muddy glory.
Data pukes! Hurray! Hurray! Big data pukes!!! : )
It is a part of the natural evolution. You are excited about having data, you can't wait to puke it out. People might not have seen anything, they are thrilled to see % Exit Rate (perhaps the worst metric created in web analytics) or the Reverse Goal Path report (perhaps the most useless report in digital analytics). Get the data, send it out.
You'll get over it quickly, and your company will follow soon enough. Worry though if they just ask you to "produce more reports" and don't come back and start asking questions about deeper insights you can find for them. Because it might be your first big red warning sign that you are at the wrong company and your career is going to stink.
Usually, in a month or two, people will realize the data pukes are not useful and move to asking you for just the data puke they need. This is a great sign. We move to the custom data puking (CDPs) stage.
This is a good stage. The Paid Search team will ask for just the stuff they need. The Content team will ask for Page Value. The landing page optimization team will demand regular reports of all entry points to the site/app. And so on and so forth.
They will ask for a mix of not useful metrics and some really interesting ones. They will still ask for data. All ok. Where you can, throw in metrics and dimensions you think they need (in addition to what they want). And just give them data.
You want them to understand that they are not Analysts. They don't know how to do on the fly advanced segmentation, they don't know how to dive deeper to understand root causes, and they sure as heck don't know how to create the kinds of custom analysis required to answer the really hard questions.
But they will figure that out soon enough. Give them a couple months.
At the end of that period, they'll ask you for the very last output of our picture: Insights, Actions and Business Impact.
When this last piece is in place, you will know that you have arrived. That you are working at a company where there will be constant happiness due to all the orgasmic feelings from data!
Insights (I) are findings from the data. They typically look like this: "Data suggests that x happened. When we dig deeper we identified the causal reasons y and z." Most Analysts stop at the first sentence, because it is all they can see from the graph or table in front of them. The best start there, go deeper because you need to find the causal factors (y and z).
Actions (A) are steps that the business should take. They typically look like this: "Triple the investment in Paid Search for this list of keywords." "Focus exclusively on products abc, def, ghi in Florida and products klm in Amsterdam." "Invest in creating video content because of reason 453." And so on and so forth. Actions are something specific the business should do. It is your job as an Analyst to identify them (though not if you are just a report writer).
Finally, Business Impact (BI) computations are simply you quantifying what will happen should your company take the action. They typically look like this: "Tripling the investment in Paid Search for this list of keywords will increase revenue by $893k per week." "Focusing on these key products in these key geos will increase profitability by 657%."
Recently I had the opportunity to cover the IABI in a significant amount of detail in my blog post on creating strategic dashboards. You'll find it here: Strategic & Tactical Dashboards: Best Practices, Tips, Examples.
And that's the complete picture!
Not all that complicated right? Five core elements surrounded by three sets of inputs and one set of outputs. All leading up to nirvana!
If you are in a leadership role, or want to appear to be in, the next two sections will be quite valuable. We'll look at the phases in which you should set your execution strategy, and close with which teams should own what part of this ecosystem for optimal success.
Too many people try to go for a revolution when it comes to their digital strategy. They fail. On the web, evolution works. An evolutionary strategy allows not just one part of the company to get better overnight (revolutions can do that), rather they allow everyone to get better together so that the sum of the parts can be bigger than the individual parts.
This is why I love the ladder of awesomeness thinking. Do one thing, get good at it, get everyone good at it. Move to the next one. Rinse. Repeat.
If you are just starting on your digital analytics ecosystem, you'll find value in knowing what the optimal order is to execute your strategy below. If your company is already in the middle of this, you can figure out why things are so messy or you've not made more progress.
The phases in which you execute will vary a little bit on your company, country, skills available, digital sophistication and other factors. But I want to offer an optimal starting point from my years of experience from working in many different companies and countries.
I believe usually people execute in three phases as they go up the analytics ladder of awesomeness .
Phase one is all about data capture. Putting tools in place and identifying the first set of metrics. This will quickly be followed by an effort to understand business priorities.
That will takes us promptly to executing based on a core set of KPIs and dimensions. Steps one through five.
Take a breath. This is a nice chunk of work. It puts the foundation in place. All quality control. Checking data collection quality etc. happens here.
Phase two is all about data reporting. It begins with the pure data pukes, which will help the company realize tools don't do diddly squat and cause them to immediately prioritize an investment in Big Brains. [Remember to pay well for the big brains. If you throw peanuts you only get monkeys.]
This now sets you up beautifully to move into custom reports and CDPs (responding effectively to business needs!). We close out this phase by getting good at advanced segmentation (the most sophisticated kind, not the silly new vs. returning visits – perhaps the worst standard segment in any tool).
Take an even deeper breath. You are now on the cusp of glory. Few people get to the next phase.
Phase three is all about rocking the data analysis universe. It consists fulfilling the dreams you dreamed of as a little child when you looked at the stars and wanted to grow up to be a true blue business analyst.
Start by getting good at identifying insights. Try to not send CDPs or data of any kind. Just send an email with bullet items in English that describe what the data says and why it is saying that. Then move to competitive intelligence, this will bring a whole new set of bright lights into play. The illumination will be fantastic for your business strategy and data analysis.
Computing business impact is non-trivial. You have to get good at some predictive analysis and forecasting (basic kinds), being able to talk to other teams, working with Finance in particular, understanding macro business trends etc. Good stuff, great outcomes. And finally you'll only have to figure out how to put in place a process to identify, evaluate and monetize the relevant new business opportunities.
Simply put…. Phase one is about getting really good at Data Capture. Phase two is all about Data Reporting. Phase three is all about Data Analysis (identifying insights that lead to very specific actions the business should take presented with impact from aforementioned actions).
Timing is also dependent on many variables that will be unique to you. But allow me to share the wisdom I've had the good fortune to accumulate during my professional experience.
If you are just getting started you'll spend the first six months , perhaps a little less in Phase One. And that includes your initial analytics tool implementation (you don't need every gosh darn thing implemented right away – any alternative to that strategy is one sign your internal team or agency or tool vendor is trying to con you).
You spend month six through month twelve in Phase Two. Your leadership team will really start to value data during this time. They will give you more money to invest in ancillary tools, more savvy technical strategies etc.
There is no end to Phase Three. But, it will take you between nine to twelve months to get to a state where people inside your company will start to recognize that this is a completely distinct phase and that you are adding unique value. Data, at this end of this time period, will be a indispensable part of business decision making inside your company.
Two years to get to the end of Phase Three – a phase that never really ends. You just get better and better at it.
I hope this helps you understand the entire digital analytics ecosystem, the phases in which you can execute your strategy and the approximate timing expectations you should put in place for success.
There is a lot to be done. It requires hard work and perseverance (and Big Brains!). But, it is not hard to understand, it is easy to realize if you are on the optimal track, and it is straight-forward to measure if your strategy is taking too long to come to fruition.
As always, it is your turn now.
I was worried about missing something from the core five elements, would you add anything there? Do you agree with the three inputs on the left? Is the order of execution of the digital analytics ecosystem reflective of the order in the 15 steps above? Is there an element, input or output that your company struggles with in particular? Would you have categorized the execution strategy into more or fewer phases? How about the buckets of time? Does phase two take longer than six months?
Please share your wisdom, critique, ideas, tips, lessons and battle scars via comments.