Multichannel Analytics- Tracking Online Impact Of Offline Campaigns

Outstanding Admit it, you secretly live in the fear of your Senior Management finding out that your online greatness is less a result of your online campaigns and more a result of the tons and tons your company has invested in the real world.

The real world. "Offline" to you and me. :)

We tend to often overlook the pesky offline real world. Sooooo booorrring ! (Say that with a Paris Hilton'ish brush off. :)

I think most of it is not malicious.

For one thing it is really hard to measure. For another thing your "online" presence is probably three geeks (says the proud geek!) living in Seattle and your "offline" presence is 15,786 people in the company with power concentrated in New York and Atlanta. Hard to coordinate and get "them" to listen to us and pay attention to us.

But the world is not online or offline, it is nonline (hat tip to David Hughes for that magnificent term). One of these day everyone will get that.

Meanwhile you are the smart one, know that it is important that our "online analytics" morphs into true Multichannel Analytics, i.e. non-line analytics.

Recently we had covered how to measure offline impact of online activities.

In this post, dedicated to Sandra Dowker, we'll cover the reverse: how to measure the online impact of offline campaigns / activities.

Then you'll have a blue print for doing nonline analytics: Offline -> Online -> Offline .

Why should care about offline at all?

For most companies radio, tv, newspapers, magazines, catalogs, retail stores, call centers, etc still form a bulk of their business advertising and marketing expenditure. All these touch points have a online impact (intended or not).

The online channel often, though not always, can be a lower cost channel. Showing exactly how to use it (which campaigns, mediums, products) and for whom (visitor personas, geographies) and why (recession baby!) will ensure your job security and the channel's security as well.

Know that pesky "direct" (or in some web analytics tools: "unknown") bucket that you never seem to be able to understand? Many of those visitors are certainly "url typers", but depending on your business a good chunk of them could be visiting as a result of your offline activities. Give them due credit.

Culturally there is no better way to get your company (HSN or NFL or Dell or B&H Photo & Video or 1-800-Flowers or….) to pay attention to "lowly and under appreciated" web analytics than show you can quantify the online impact of their massive (or little) offline spend.

I have never known a better strategy then to understand and align yourself with the largest money maker in your company (and no its not online). So there.

Convinced?

Sure you are.

If it is not the enticing prospect of understanding the data better then it has to be craving to be loved by your greater organization. :)

So what's the problem here?

Like in the case of the offline impact analysis, the problem here is also one of the missing primary key (see that post please if you don't know what this is).

Our goal is the figure out how to tie your visit to our website due to a stimulus from a offline campaign. How do I know that it was a tv ad that drove you to the site or a magazine article or a banner by the side of a road etc.

Once you have collected that piece of data you can do any of the other analysis you want for that offline to online traffic stream, be it conversion rates or site abandonment rate or task completion rates or even non-ecommerce outcomes.

Let's get going. . . .

Tips for measuring on-line impact of the offline channel:

Here are the most common offline-to-online movement channels and how to collect data for end-to-end analysis of each:

#1 : Use redirects (vanity url's).

The grand daddy of them all. Plaster your billboard (or magazine article or tv ad) with a easy to remember url and boom (!) you got yourself some tracking.

A magazine ad with the call to action Visit www.usequickbooks.com redirects to www.quickbooks.intuit.com/tracking_code=newsweek_dec_2008 The rest of the analysis is a piece of cake (simply segment out visits with that tracking code).

A box of 12 krispy kreme doughnuts to your IT person will ensure all redirects are coded with the correct tracking code.

If you are super cool like my former employer then you will have web based interface where the Online or Offline Marketer can attached a tracking code to any url, hit save and be in tracking heaven!

Another example:

www.dell.com/tv

(today, dec 22 here in Maldives) redirects to

http://www.dell.com/content/topics/segtopic.aspx/tv?c=us&cs=19&l=en&s=dhs&keycode=6Vc94&DGVCode=TV&dgc=TV&cid=11510&lid=985367

Note the amount of specific tracking that the folks at Dell have attached do that redirect. Bravo folks!

Contrast that to HP which has a ton of tv ads running and has a working redirect that leads to no trackable information:

www.hp.com/tv

to

http://www.shopping.hp.com/webapp/shopping/home.do

Missed opportunity.

In closing it is a crime of the highest proportions if your magazine, catalog, tv, radio, bus, billboards are:

1) not using easy to remember vanity urls and

2) ensuring these vanity urls are permanent redirects and that

3) they are encoded with the correct tracking parameters (version of the ad, name of the magazine, location of the billboard, offer in the radio ad, etc etc).

That was not hard right? You are on your way to measuring online impact of your offline channel!

[I can sense the Smarty Pants amongst you snickering at the prospect of data pollution because people posting your vanity url's online. That is the reason #2 above is important. Permanent redirects (301's) will pass the referrer's data to your website. You can then split out online referrals from the offline referrals (offline will have a blank referrer).

For example the Analysis Ninja at Dell will easily be able to split out visits from my blog from clicking the links above, www.dell.com/tv, and easily exclude. Now stop snickering.]

#2 : Use unique redeemable coupons / offers codes.

Pretty much all multichannel merchants now do something similar to this. . . .

dell e-value code entry page

The offline mediums (a magazine in this case) carry unique promotion (or offer or config) codes that must be typed into a form on the website. This allows those visits to be tracked as being from a external "motivation".

This works well for tv, magazines, radio, catalogs and other such mediums where it is easy for people to remember the codes.

For example the ads for www.1800flowers.com in the NYC Taxi Cab I was riding close to thanksgiving asked me to use the coupon code taxi to get $5 off a $50 order. It was easy enough code for me to remember it and use it.

www.qvc.com and www.hsn.com also tend to use these types of tracking mechanisms as one of them tools in their arsenal to track people who buy on the site while watching the tv show (or later).

Catalogs for pretty much every major or minor company, including yours I am sure, is using this exact strategy as well by providing unique coupon or offer codes.

Bonus Tip: Another great strategy is to use the same coupon code between your channels. For example you are giving $65 off the Apple Ipod Touch. Your ads / catalogs / tv campaigns can say something like "call 1800 Hot Hot Now or visit Hot Hot Now . com and use the code ipod65".

The benefit of this is that you are providing people a choice in terms of channel preference (use the phone or the site if you want) but since the code is the same you can track it delightfully later. This helps you understand channel preference by media type (tv / catalog / radio) and also by product type (electronics / food / meds etc) and by region (what's up with people in Florida 100% using the phone channel to order their Viagra) etc etc.

#3 : Use online surveys / market research.

On this blog we have often talked about using onexit surveys (4Q or others) to understand Primary Purpose, Task Completion Rates & Segments of Discontent (see: Three Greatest Survey Questions Ever).

You can easily adopt that methodology to ask two more questions of your website visitors:

"Which of the following were the source of your visit to our website?" [The answers can be: A tv ad, A radio spot, A google (:) search, You received a catalog etc, wordsmith your options.]

and

"What is the likelihood that as a result of a visit to our website that you'll make a purchase?" (or sell your kidney etc) [The answers will be something like More likely, Less likely etc.]

These two simple questions, drop down single choice, will help illuminate both the drivers of visits to the website (and a real chance here that you'll explain your very high "direct visits" number here) and also the preference in terms of the purchase channel (and of course you can use this for tech support or other non ecommerce websites as well, just twist the second question a smidgen).

In our online to offline article I had talked of the possibility of using primary market research to understand outcomes that happen offline. You can also complement surveys like the one above with primary market research to understand channel use by your customers.

#4 : Correlate traffic patterns with offline ad times / patterns.

My wife was watching HSN the other day (that is how hard it is to find something to watch on tv in the age of 900 channels!) and saw Wolfgang Puck talk about his genius kitchen knives and a "extra special deal just for you right now for only $19.99 plus shipping and handling". (Imagine that with a Austrian accent.)

So Jennie of course went to the website instead of the phone and two minutes later (well a week actually) we were proud owners of Wolfgang Puck limited edition extra special for a limited time only with a chopping board knives (with knive covers!).

Now this is not a unusual customer behavior. Offline media stimulus causes us, lemmings like, to run to the site and do stuff.

Yet it is extremely rare that a Web Analysis Ninja sit down and overlay the company media plan on top of the traffic patterns and deduce the impact online of the offline media spend.

Why not?

Sure you have to beg, plead, and practically sleep with someone to get your hands on the comprehensive media plan for the company. But take one for the team and do what you have go.

Once you have the magazine / tv / catalog / postal mailings / radio / billboard plan for your company then do the correlations with your website traffic and see what the impact is.

My tip would be not to just look at overall traffic (or All Traffic) for the site, you may or may not detect something. Correlate it with your Direct Traffic. You might see a sudden spike in traffic there. Or with Direct and Search (organic or paid) referrals. When you do stuff on tv / radio / retail stores people search (what can I say!).

I had covered exactly this strategy in this post: Excellent Analytics Tip #12: Unsuspected Correlations Are Sweet! You can find out exactly how to execute this analysis in your company from that post.

Here is a picture of the correlations that I had shared there (for one company who ran radio campaigns and the resulting impact, not just directly from visits from the vanity url mentioned in the radio ad but also from Direct Visits, Branded Search etc, very impressive, and surprising, holistic outcome):

Multi-channel tracking for radio ads and online impact

Please see that post for detailed explanation and guidance.

Of course it is always most optimal to identify causality as well (because correlations don't always mean causality). The keywords you drill down when you see the search spike perfectly matching your offline campaigns for example, that has intent. Or you are perhaps running the survey I just mentioned, that will show causality. Or you sold a bunch of Puck knives exactly when, or slightly after, the tv ad ran, that will indicate causality.

Do that.

UPDATE: #5 : Use the power of controlled experiments!

Jim Novo's comment below reminded me that I forgot to add this super awesome way to measuring multi channel impact.

I had covered it as a key strategy in my multi-channel analytics post, tip #6 for measuring impact of online to offline outcomes.

Please see that post for more details. I share stories about using newspaper inserts or conducting geographically isolated experiments in retail stores (or fast food restaurants) etc.

The strategy works for on-line to off-line, and it works even more brilliantly for off-line to on-line.

.

There you go. Four Five simple things that anyone can to to get started on their journey to identify online impact of their offline strategies.

Remember your goal is to identify the complete picture: Offline -> Online -> Offline.

Those of you who have my book, Web Analytics: An Hour A Day, will know this handy dandy reference picture from page 235… it shows how to track your nonline world efforts and capture the key pieces of data to do true multichannel analytics:

multichannel marketing value analysis framework

You can use this multichannel value analysis framework to plan out how you will make sure you are passing the various primary keys and forth and also use all possible techniques at your disposal.

Ok now your turn.

What are the strategies that you have use to measure online impact of your offline campaigns? Have you used any strategies above? What has worked for you?

Please share your experiences, your best practices and tips.

Thanks.

PS:
Hello from the absolutely gorgeous Maldives, more specifically the Conrad Rangali Island. It is lovely here, warm, the staff are wonderful, snorkeling is great (and I am committed enough to all of you to still squeeze in a, hopefully, valuable blog post!).

Here are some pictures. . . .

This is one of the two islands that forms the Conrad:

conrad maldives rangali island

You take a sea plane from Mali to get there, this is one of the island resorts from the air:

manta ray island maldives

A typical sunset:

sunset at conrad rangali island maldives

Lots and lots of snorkeling, absolutely wonderful, and my one high was running into a 4.5 ft shark. It passed by my nose, I was absolutely petrified, it was not [Tweet]. Here's a small one close to the shore:

baby shark maldives

Tiny little hermit crab, hard trying to get this close to it without scaring it:

hermit crab

Other than that lots of personal time (also hard to photograph):

holding hands

Hope your holidays are fun wherever in the world you are.

Ok, don't forget to share your online to offline analytics tips using the form below!

Comments

  1. 1

    Cool pictures. But seriously, good post. Puts it all in one place and starts a discussion.

    Here is a question that goes a little beyond the above. Looking for some thoughts measuring network effect/buzz impact? I can get at positive and negative sentiment scores, traffic pop from certain sources, and maybe tag the links for direct sales but is that the full impact?

  2. 2

    I seriously just saw a local TV ad the other day that pretty much said the following:

    "Save $X by visiting our website right now (with arrows pointing down towards the bottom of the screen where the website's URL was written)". That was great that the local company is trying to use a vanity URL and all, but it seriously looked like this:

    http://www.ourverylongandforgettablecompanyurl.com/freeoffer123456!

    So I would like to add to this very good blog post that vanity URLs / permanent redirects are in fact the "bomb", but only when you can remember them. If they are ridiculously long, then it kinda defeats the whole purpose (and this local company was using a vanity URL – i tracked it down and went there and it redirected me to their main site's URL :)

  3. 3

    Avinash, I've been wondering when you were going to touch on this subject! It's always been amazing to us how few businesses – especially small businesses – attempt to track their offline advertising and marketing. When marketing dollars are limited, tracking is even more important and asking people how they heard about you just isn't accurate enough (although it's better than nothing!). Kudos on a great post and don't forget to make sure those tracking phone numbers carry over to the landing pages via the unique URL's!

  4. 4

    Great article covering an area frequently untouched by many businesses. Also read http://suyoung.wordpress.com/ which outlines some advantages of multi-channel marketing.

  5. 5

    Simple yet effective and I always like the use of examples. Makes things much simpler to understand.

    Strange that I was aware of few of these methods but never got around to using them. Sitting somewhere in the corner of my brain gathering dust. Its only when to read it that you know your knew it – epiphany.

    Thanks for the great post Avinash. It is always a pleasure reading your posts.

  6. 6
    Alice Cooper's Stalker says

    Amazing photographs, Avinash. You are a talented photographer. I wonder what the hermet crab took from it's encounter with you? Hmmm…

    This was a good post.

    Question. Could you elaborate on this point – 'If you are super cool like my former employer then you will have web based interface where the Online or Offline Marketer can attached a tracking code to any url, hit save and be in tracking heaven!'

    Did they build an application to allow non-IT folks to create the redirects themselves without IT involvement? If so, that is cool and beneficial to the IT organization, as well. Chalk up ongoing productivity savings for that effort.

    Quick note…if you are dealing with a campaign where marketing doesn't want to use a vanity URL for a campaign or promotion, you may have to rely on capturing a baseline on some key content and then seeing if it's received any lift once the campaign was in play. I'm in the middle of tracking a campaign where marketing wanted to draw people directly to our brand's core website. They insisted on this. No vanity URL. Yet they wanted to be able to measure the response. Measuring lift of key content was cruicial in this case.

    Happy Holidays, everybody!

  7. 7

    1 word – WOW
    An unbelievably simple an idea that it can so easily be overlooked – Redirects!

    Reading your blog is a constant 'aah!' moment! :)

  8. 8

    Now-a-days people think that newspaper & magazine advertising would not benefit their business. But as per my knowledge it plays a vital role in getting enquiries and sales. All you need to do is, select a proper company which offers cheap advertising, place the ads in popular newspaper which would be accessible to all kinds of advertisers.

  9. 9

    Nice work Avinash!

    If the company has the stamina to pull it off, controlled geographical testing is one of the best ways to get a handle on offline media contribution to web site activity.

    Pick similar markets that are geographically distant and get baseline traffic. Then run offline media in one of these markets and compare to the other, which is "control".

    You can test various combinations of TV, radio, print etc. and look for incremental lift.

    To the extent you can control the distribution of online media geographically, the same kind of test can be done to (for example) measure the contribution of Display ads to Search campaigns.

  10. 10

    Thanks for the post, nice insights. I especially found the re-direct URL tips useful for our purposes.

    One additional aspect that I have started to analyze with more zeal is the brand searches coinciding with offline campaigns when running paid search efforts (notably TV and Radio). I have found that searches on brand name terms or campaign searches correlate rather nicely with offline (though they sometimes suffer slight delays, or sometimes need some lead in time to have an effect). This is especially true if you are coordinated enough to have a branded, slogan-ified (?) campaign running, with which you can marry some nice phrase search terms. There are certainly qualifications for looking at this data, but I have recently found that there is helpful insight in brand terms.

    Keep up the good work.

  11. 11

    Avinash – I see a lot of advertisers using tv tracking mechanisms like the one you mentioned above, where you would go to dell.com/tv36, where 36 would specify some DMA. I would imagine a lot of people don't actually type in the "/tv36", which would dirty the data, wouldn't it?

    The only way that might reduce dirty data is if a) you tied a special offer that could only be reached through that page, that was not accessible form the main site; or b) overlaid your media buy with web traffic to see where the spikes were by city/region.

    What is your experience with this potential loophole?

  12. 12

    #1: I have done experiments, and it seems that with a vanity URL sending people to so-and-so.com/xxxx, a good proportion of them will hit enter before typing what comes after the /. This is seen by a good surge in traffic to index. The longer the URL, the more this phenomenon occurs.

    Make sure you don't forget to leave "scent" of the campaign on the site, or use a subdomain if you want to be sure you're analyzing only the campaign reponse (and event then, it's not clear whether you won't have people still typing the main domain, expecting to find mention of the campaign).

    Finally, I just can't imagine how much shuffling there could have been in the WA blogosphere if that shark had found you appetizing…

  13. 13

    Avinash, true analytic geek to be able to convey a good topic in a place like that. It still surprises me that more online companies do not know or understand the simple use of vanity URL's have a great holiday and be safe.

  14. 14

    Aseem: You are perhaps referring to measuring "viralness" or "buzz" started by seeding say a youtube video or a new promotion to get access to Your Content Site for $1 for 10 years or something like that?

    Or general measurement of the ambient noise about Your Content Site from twitter and blogs etc?

    Both of these are much harder to figure out (and be suspicious of anyone selling you a magic potion that solves this – especially in exchange of a small fortune!).

    One thing I have done in this case is to measure if these "channels" (seeded content or ambient noise) drive Visitors (Unique) with the right kind of "latent visitor behavior", but that I mean they come back to the site again and again (Loyalty) or more frequently (Recency) and even try to measure Days and Visits to Conversion (where conversion is anything, not just a ecommerce submit order).

    All four above are "pan session" metrics that are hoping to tie credit to the original source for behavior that might happen across sessions (over x amount of time – say a month, or you choose).

    That can be a great way to quantify value for the right behavior (with Loyalty and Recency) and right outcome (feed sign up, trial account, lead etc).

    In both cases:
    1) starting with identifying a unique source or set of sources and then
    2) segmenting out the Visitors and finally
    3) monitoring their behavior a few days / weeks / month out.

    Hope this helps spark some ideas of your own.

    -Avinash.

  15. 15

    Beautiful pictures — enjoy the warmth, the nature, and the holidays :-).

    Loved this post, especially since the offline-online-offline synthesis of behaviors is one of the more complex areas for any analyst. And I agree with Jim's comment about controlled geographic testing. We do it all the time to test out products/services in the traditional world (the concept of test market) so why not apply a similar technique for measuring the online/offline impacts.

    I will admit that this is one area where I found it very difficult initially to repress my statistical/analytical side and as you mention try to look for patterns instead (I grew up in an anlaytical world where anything that was not statistically significant was "rejected" :-) ). Thanks for some excellent tips and advice –hope some of the HIPPOs are reading your blogs too :-). It is amazing how folks can spend millions of dollars for a 30second Superbowl spot without thinking of how we can study the impact (if any) of that ad.

  16. 16

    Alice Cooper's Stalker: It was a very simple web based app (web page really, but let's go with the sexy app name :)).

    Three columns.

    First: Vanity url (100's of urls that the company had purchased and were in our database).

    Second: Target (landing page / destination, if there was a landing page already it would be shown here, you had the option to delete the current one or update it to a new url).

    Third: Tracking Parameters (there was a simple framework by which three tracking parameters could be added to the landing page url, these could be pulled up from ones already in the system or set up new ones).

    A developer hacked it in a few hours, it took us a week to get the process nailed down (who can access what and do what kinds of things and how to keep an audit of the changes), but once it was up it got this thankless work out of IT's hair, it gave more accountability to Marketers and Analytics benefited.

    To me this is how you play the game if you want to scale Analytics. Small things that have a huge impact (across a complex org).

    With regards to your second question, some kind of correlation with past data or past behavior would be an option (I would also try to drill down by traffic sources to make sure any lift is not because someone trashed you to twitter rather than from your marketing buddies).

    But honestly without some kind of identifiable source (vanity url or tracking parameters) you are playing a guessing game.

    Jim: Great point.

    I had covered it in my online to offline post (point # 6) but I should have added it to this one since it applies just as much in the case of offline to online.

    I have now updated this post with your suggestion, thank you!

    P D: You are absolutely right. That is exactly what our experience was when we looked, specifically, at radio campaigns:

    https://www.kaushik.net/avinash/2008/02/excellent-analytics-tip-12-unsuspected-correlations-are-sweet.html

    The second graph, or this one directly:

    https://www.kaushik.net/avinash/wp-content/uploads/2008/02/audio-tracking-branded-search-impact.png

    Ryan: You, Mr. Analysis Ninja, would rarely rely just on a simple analysis, i.e. How many came on /tv36. You would do both the things you mention in your comment.

    But.

    You would segment by other stuff, like Geo (based on your company's media plan – if the ads were just shown in FL and all the traffic is from MN something's amiss, but if there is a sudden spike in traffic to the overall site from FL and even if not to /tv36 then some credit might be given to the campaign).

    Stuff like that.

    Jacques: As I mentioned in reply to Ryan's comment that rarely would I, with my full Analysis Ninja leotard on, rely just on one data point, the vanity data.

    I would do multiple correlative analysis, with my primary weapon being using the power of segmentation and an understanding of the company's media plan (offline campaign spend / execution).

    That would get me, to a great extent, around the missing "after /" etc. Not all the way, much of the way.

    Kevin: I think many understand the power, but rarely use them to the fullest extent (like including specific tracking parameters tied to the media plan for example).

    The real problem is not using the portfolio approach when it comes to tracking the true online impact of offline campaigns. No one method is God's answer, I recommend using more than one methodologies (some of which are covered in this post).

    -Avinash.

  17. 17

    Yeah! Great post!

    I'd already see myself short in ideas, trying to do the Offline -> Online -> Offline Analytics way.

    I can tell you that I did something good, but for sure, from now on, after reading this post, I will make something really better.

    Thanks for keep doing these helpful posts.

  18. 18

    […] Here are a couple posts from analytics guru Avinash Kaushik that can help you connect these dots in the ‘non-line’ world (offline – online – offline). First, from a July post, here’s his thoughts on tracking offline conversions. Like a lot of Avinash’s posts, this is long, but it’s worth it — you will find ideas to try, some of which are quite easy (like unique telephone numbers), that will give you a good idea about how much your website is driving offline conversions (online to offline impacts). In his follow-up post near the end of December, you will find good ideas for tracking offline to online impacts. Have fun! […]

  19. 19

    Beautiful pictures — enjoy the warmth, the nature, and the holidays :-).

    and happy new year

  20. 20

    Great post, and it is an area of marketing analysis that is very often undervalued and underdone. In some "story-friendly" businesses, we've seen some impacts of big stories show traffic jumps of up to 20x normal traffic, but this can only be analysed using #4 (correlation in time), which can be extremely difficult to operationalize. An announcement of a trade or an injury in a star player on a sports team, for example, can drive that size of spike.

    How have you seen people move beyond the anecdotal "that spike was from X" into operationalizing the results that are only seen by correlating in time?

  21. 21

    Steven: Overall I would say lots of analysis like this tends to be hard to "operationalize" because of all the possible variables. But. . . . .

    There are a number of standard statistical and financial modeling techniques that you can use. The best strategy to operationalize is to have a resource (person) in your company who has those skills.

    Sometimes having a statistical modeling software is helpful, but I think that in most cases it is not required. Just standard segmentation and business savvy (i.e. Understanding what the company is doing in each case and then doing analysis that might not be repeatable the next time) is key.

    Some parts you can operationalize after some time.

    For example after a couple months we knew exactly where to look for an understanding of how press releases impacted traffic (because after the first few times we had done enough types of analysis to know they caused people to always do x, y and q).

    A peer of mine had done something similar in terms of predicting impact of in store displays (in that case because the first couple times they had done experiments between different Geo's, done analysis of the direct and the search traffic, and understood customer behavior).

    Hope this helps a bit.

    -Avinash.

  22. 22

    Great post Avinash…but the offline campigns not always can be measurement. For example: http://www.filmica.com/audiencias/archivos/008809.html

    See the picture. One spanish daily is in a great place in one road. Only branding! What is better (in your opinión) PPC or Offline not measurement???
    Regards

  23. 23

    Hello Avinash,

    Can you think of one instance when measuring offline "campaign" can be a negative thing? The only one that pop my mind is if to measure the offline impact you need to mess with your brand.

    For example, our 1-800 number have been very branded and is well know by our customers. Using a new number have "mess-up" people in the past.

    Any thought?

    • 24

      Sebastien: I have personally yet to see a negative outcome in terms of brand impact, except of course a company like 1-800-Flowers whose online and offline presence are tied to that number (and hence in their TV and Radio spots they tend to use promo codes to track online conversions).

      Much of my experience has been with large brands with very positive brand recognition and they have successfully used 1800 or 1888 numbers with vanity numbers ("please visit 1-800-HotOrNot for more information" :)) with no impact on their brand. Part of this is over the last few years it has become the norm to have such numbers be used.

      But the most important thing to realize is that you can experiment and measure (so move beyond your and my opinion of what might happen). Use a vanity number or url in TV ads in certain markets or in Newspapers in the month of Jan or in Radio ads airing at a certain time etc etc. Strategic testing will yield the most validating results.

      -Avinash.

  24. 25

    Avinash, you wrote "Contrast that to HP which has a ton of tv ads running and has a working redirect that leads to no trackable information: http://www.hp.com/tv to
    http://www.shopping.hp.com/webapp/shopping/home.do
    Missed opportunity."

    I do not believe that scenario is a missed opportunity because the vanity URL is logged (or tagged) regardless. The tracking information does not need to appear in the redirected URL, as is the case with your Dell example.

    Simple sessionization will ensure that the referral information is retained from one click to the next within each HP vanity URL visitor's session.

    Aside from that comment – great article – great site – and THANK YOU for posting COMPLETE articles in your RSS feed.

  25. 26

    Great Post Avinash!!

    Just one quick question. We will be running a TV campaign that will last a month. Based on your experience when we should start measuring the impact of the campaign and for how long?

    Thanks

  26. 27
    Neha Gulati says

    Omniture offers lots of different performance measures like time spent per visit, pages viewed under a campaign, page depth, entry, exit Visit number, loyalty….and many more

    Could you please guide which of them are critical to be picked & monitor to assure a campaign is doing well.

  27. 28

    Avinash, true analytic geek to be able to convey a good topic in a place like that. It still surprises me that more online companies do not know or understand the simple use of vanity URL’s have a great holiday and be safe.

  28. 29

    This section is very ambiguous:

    In closing it is a crime of the highest proportions if your magazine, catalog, tv, radio, bus, billboards are not using:

    1) easy to remember vanity urls and

    2) these vanity urls are not permanent redirect

    3) they are encoded with the correct tracking parameters (version of the ad, name of the magazine, location of the billboard, offer in the radio ad, etc etc).

    Am I right to interpret this to mean that offline ads should definitely be 301 permanent redirects?

  29. 30

    Ben: You are right, they should be permanent redirects (301) so they can pass the referrer.

    I have fixed the language in the post as well.

    Thanks for the help.

    Avinash.

  30. 31

    I like this thing in tracking visitors – vanity urls but only usable for the big brands I think.

  31. 32
    Minnetonka says

    Online impact of offline campaigns is really great if you do everything right. Good offline advertisment campaign is much more effective that a primitive search engine optimization. But tracking is a problem, and this article gave me some useful tips.

  32. 33
    Ranjan Jena says

    Thanks for this valuable information. As i work for a travel based industry, so using unique phone numbers, coupon codes, redirect urls, tracking code will help analyze my online impact from Offline campaigns.

  33. 34

    What a great post Avinash!

    So far, I always write on my report that the number of direct and search engine visits are related with offline campaign. But, It just based on my prediction, I never think about how to measure it. And after I read your post, Aha! I have to apply it in our project. Because this idea is amazing!

    But, sometimes I have a difficulty in make people believe in what I do. It is hard to explain to them that track every campaign is important. Especially with someone that don't believe in the power of digital campaign.

    Anyway, thank you for this amazing ideas!

    Cheers,
    Mega

  34. 35
    Ojus Naravane says

    Avinash,

    Very nice pics and some funny quotes too! Thank you for the article!

    I have a question regarding giving credit to the original offline referrer when a visitor completes a goal online.

    I am trying to set utm_nooverride=1 in URLs we are using to track leads from radio, posters and other offline sources. Each of these sources will use a custom URL redirected to a URL tagged with a utm_nooverride=1 and other parameters.

    My question is, does Analytics only attribute original visits from Adwords or would these sources also be properly attributed if nooverride is used in the URLs?

    Also how long is the cookie which tracks this is set. Can we customize its lifetime?

    Appreciate your time!

    • 36

      Ojus: If you are using the latest version of Google Analytics then you have access to the Multi-Channel Funnel reports.

      In MCF you can see the entire string of campaigns that were responsible for getting a person to your website in all their visits. See the Top Conversion Paths report.

      That means that you no longer have to play shenanigans to track what you want to track. Just make sure your online and offline campaigns are tagged correctly. Then use the report mentioned above.

      Less pain and more happiness for you!

      Avinash.

      • 37
        Ojus Naravane says

        That definitely makes life easier. Thank you!!

      • 38

        Agreed. However, this issue impact more my Adwords campaigns than my GA reports.

        In Adwords, there is no Multi-Channel Funnel reports, so my ownbrand campaigns, for example, will most probaly get the benefit of the conversion rather than my search or display campaigns.

        So I still need to use the nooverride tag.

        • 39

          Swann: I'm not sure I understand your challenge completely. But if you are using GA then you don't need to use your AdWords reports. You can simply avail yourself of the MCF reports in GA, and AdWords is included in there. You can also apply the advanced attribution modeling on that data.

          More on that here: Multi-Channel Attribution Modeling: The Good, Bad and Ugly Models

          Yes, there are some metrics that are perhaps easier to locate and might only be in AdWords, but they won't impact your strategic analysis.

          Avinash.

  35. 40
    alexandre says

    Awesome post and thank you very much. I have one specific question:

    How do you measure people coming from your website to another using affiliate programs?

    Thanks for your answer. Best

    • 41

      Alexandre: I'm not sure I understand your question but here are two scenarios…

      Coming to your site via an affiliate network: If you are providing links to the affiliate network to spread around then please add campaign tracking parameters to those links. It makes it much easier for you to understand which network people are coming from, based on what offer etc.Learn more about Google Analytics Campaign Tracking.

      Going from your site using your affiliate network links to other websites: You would get most of the reporting using your affiliate network. So please ask them. But you can capture the fact that someone clicked on that link on your website. Just implement outbound link tracking.

      -Avinash.

  36. 42

    Great article Avinash!

    I agree redirects are an effective way to measure traffic… what's been missing is how they convert into phone lead, email leads be them fro the website , newspaper, yellow pages flyers, expos and so on…

    Our preference was to build an app originally for our existing clients to capture exact lead source and volume data and have it attributed to a specific marketing channel / activity. It enables them to run as many different campaigns and channels simultaneously, an d track the lead to the appropriate investment.

    (domoreOptimiser.com)

    We're issuing out unique emails, phone numbers and vanity urls. The results appear in a visual format where the client can compare and optimise their spend according to each campaigns ROI, CPL, Volumes, Sales conversions etc…

    We felt that google analytics alone didn't solve the issue – we needed some old school tracking tools mashed with some new ones to solve the challenge of old school marketing being properly tracked.

    Avinash, it would be great top connect with you further. I'm a fan of your work above and would be a fun chat to continue on with…

    Cheers and thanks again
    Adam Cunningham

  37. 43

    Hi Avinash, I just stumbled upon this post of yours. Firstly, thank you so much for this article, very helpful!

    One question for you though- I have observed offline campaigns tend to have a lot of trickle orders . So, if I want to analyze the efficiency of media spend for that 1 week, how would I be able to do that? Clearly, the week's numbers are being influenced by previous weeks' spend too, correct?

    I am at a situation where media has been on air perennially, so doing an ARIMA would not give me a good idea of baseline. How would you go about analyzing this trickle down effect?

    Thanks,
    Hari

  38. 45
    Nicola J. Strand says

    Another good post. However would this be able to work for B2B? Our offline activity is events and sometimes flyers. Permanent redirects sound like a great thing for a start so I'll definitely look into that. However our sales cycle is a lot longer as our product is a high pricepoint. And even then, new enterprise customers will be more likely to go to a sales rep. I know that we can import that data to GA but I'm not sure that it would be as easy to measure.

    Where would you – or anyone – advise that I start with this as I'd love to be able to show offline attribution?

  39. 46
    Evan Dunn says

    Media mix modeling – why didn't that come up here? Media asset allocation to for modeling the impact of offline channels on sales – isn't this the ultimate proposition for companies looking to quantify offline's impact on online? And once they're built for backwards-looking ROI analysis ("Attribution") they can be built for forecasting/predictions and made to develop recommendations.

    You get at this in your Multi Channel Attribution post (https://www.kaushik.net/avinash/multi-channel-attribution-definitions-models/ ) when you delve briefly into time-decay models.

    We've built these, for multiple layers of conversions. I'd love to discuss with you.

Trackbacks

  1. […] Avinash Kaushik shares strategies for multichannel analytics to track sales driven by offline channels. […]

  2. […]
    Some other articles worth reading this week include:

    Multichannel Analytics: Tracking Online Impact Of Offline Campaigns
    […]

  3. […]
    Avinash Kaushik, a leader in web analytics, wrote a fantastic post about tracking online impact of offline campaigns. With different techniques like redirects, unique coupon/promo codes, and identifying correlation in traffic patterns he describes multiple ways you can effectively track how well your offline efforts are impacting your online efforts.
    […]

  4. […] You should also think about how well you track the influence of offline media as explained in this post by Google Analytics evangelist Avinash Kaushik. […]

  5. […] Best practice in such offline or multichannel tracking has been explained well by Avinash in his post: Multichannel Analytics: Tracking Online Impact Of Offline Campaigns. […]

  6. […] Tracking Online impact of Offline campaigns […]

  7. […]
    Of course, they also want to track the effectiveness of different promotions.

    Best practice in such offline or multichannel tracking has been explained well by Avinash in his post: Multichannel Analytics: Tracking Online Impact Of Offline Campaigns.

    The core technique is to use a 301 redirect which appends a campaign code. He gives the example of http://www.dell.com/tv which redirects and appends a (non Google Analytics) tracking code referencing TV:
    […]

  8. […]
    Omnichannel. It’s a word being thrown around from tradeshow panels to industry publications to the latest feedback management systems. Omnichannel speaks directly to the nonline customer; it’s an approach that connects customers to businesses on the web, mobile, in-store, product and beyond. It’s about hitting your customers where they are, anytime and anywhere, when they are ready to leave feedback. And it’s what our patented opt-in feedback system is built around. Feedback pairs perfectly with the omnichannel world, which is why we are constantly pouring through omnichannel data that improves customer experiences, across a myriad of industry verticals and sectors. And there’s no shortage of data to dig through. Especially when it comes to retail.
    […]

  9. […]
    Avinash Kaushik (@avinash) has written a great post here, describing the problem in a simple way. Essentially, how do you figure out where all of that Direct and, in our case at least, most of the Organic traffic is coming from? I’m not showing figures, but the snapshot below shows the five most important channels for part of Red Gate’s business:
    […]

  10. […]
    What I felt today was different, though. It came when I was looking for some info on a piece of software I use (Google Analytics). I started at the website of a well-known expert in the field, Avinash Kaushik at his The Good, The Bad, and The Ugly piece. That article led me to his Definitions, Models, and a Reality Check piece. That article led me to his Tracking the Online Impact of your Offline Campaigns piece, and that article led me to David Hughes’ website because he coined the term “non-line” that Avinash uses (it means marketing efforts that exist both on- and offline, like the color of a logo).
    […]

  11. […]
    I’ll start by looking at the impact of online activities on offline conversion. I learned an awful lot from a 2008 blog post on tracking offline conversions by data guru Avinash Kaushik. Before I delve into some of Kaushik’s great suggestions, I want to take a step back and think about potential things to measure and why:
    […]

  12. […]
    Avinash Kaushik (@avinash) has written a great post here, describing the problem in a simple way. Essentially, how do you figure out where all of that Direct and, in our case at least, most of the Organic traffic is coming from? I’m not showing figures, but the snapshot below shows the five most important channels for part of Red Gate’s business:
    […]

  13. […]
    With the expansion of multi-channel analytics, we're creeping nearer and nearer to with the ability to present one of these graph with growing confidence. It is serving to us to be a lot smarter entrepreneurs by considering past one-night time stands to how relationships are developed with clients. It is the manufacturers that embrace one of these technique that may have the benefit shifting ahead of their industries.
    […]

  14. […]
    You can set your goals in Google Analytics to track customer engagement and conversions, whether they come from search engines or social media. If you are using online ads, compare how much you are spending versus how much sales you are getting. If you are using an offline channel, make sure that you are directing them to your online channel.
    […]

  15. […]
    Một số lỗi trong hàm hồi quy, có lẽ tôi sẽ giải thích ở một bài viết khác. Các bạn có những cách phân tích và đọc dữ liệu nào nữa, đừng quên chia sẻ cho MediaZ nhé <3 Biên tập bởi: MediaZ Corp Nguồn tham khảo: Kaushik
    […]

  16. […]
    Tất nhiên, đây không phải cách tối ưu nhất để xác định mối tương quan giữa kết quả với các biến (bởi hàm hồi quy không phải lúc nào cũng chính xác, bạn còn phải kiểm tra chỉ số tương quan và các yếu tố khác, để xem biến với kết quả có thực sự tương quan với nhau không) Một số lỗi trong hàm hồi quy, có lẽ tôi sẽ giải thích ở một bài viết khác. Hy vọng với 4 phương thức đo lường các chiến dịch Offline qua các kênh online trên đây hữu ích cho bạn đọc. Tham khảo: Kaushik
    […]

  17. […]
    Multichannel Analytics- Tracking Online Impact Of Offline Campaigns – Omnichannel, offline-to-online conversion, online-to-offline conversion, cross-channel. All these terms are thrown around to bring a 360 (also a term being thrown around) picture of your marketing efforts. This (yet another) Avinash blog will give you the basics (Our addiction to his blog is problematic).
    […]

Add your Perspective

*