The Great Attribution Myth

Can we get real for a minute? I have been thinking this for quite some time now. Let’s be brutally honest – attribution, as we currently know it, is kind of a joke. I know, this is probably not necessarily going to be a popular opinion, and I welcome other points of view in the comments or over on Twitter to this bold statement. But seriously, with today’s technology we simply do not definitively know how a single person behaves across devices and interacts with our advertising definitively or absolutely. It’s not even really close to being definitive and yet we act like it is. I think this does everyone in the process a pretty major disservice.

It’s not anyone’s fault that we don’t have this data. Ad platforms like AdWords and Bing Ads and Facebook are trying their very best to provide it to us. And I do think they are trying – it is in their best interest for us to have more definitive information about exactly how people view and interact with the advertising we purchase on their platforms and the exact roles each piece and each view plays in an eventual purchase/not purchase decision tree. As advertisers we all crave that level of data. But the truth is, there is not a way to actually know all of this.

Attribution, in its current state, is largely based on cookies of some variety or another. And this was an ok method 5 years ago before people became aware of cookies and started to clear them on a regular basis. But even then, it was not great. Today it is even less so. And, cookies can’t follow a person across devices. If I see your ad on my laptop and click through to your site and then look at it again on my phone while I am waiting in line for school pickup, you’d have no idea I’m the same potential customer.

Cookie tracking is not going to get any easier either. Apple is about to make it harder for cookies with their “Intelligent Tracking Prevention” in the next release of Safari (you can read it here from Marketing Land and another piece on Search Engine Land regarding new cookies). Ad blockers stop not only ads, but also cookies before they even start. If your device is set in Incognito mode, no cookies either.

Google included in their regular May announcements their “attribution for all” function and presented it as having essentially solved attribution. It hasn’t. And, I’m going to get even bolder here – I am not sure it ever will. Let’s think about what it will really take from a data standpoint to really, truly be able to show an individual’s journey through a decision process. To really know the process a person follows, data would have to be collected on the following AT THE INDIVIDUALLY TRACKABLE LEVEL:

  • Ad views
  • Ad clicks
  • Landing page views
  • Other searches
  • Views of sites that are not yours to read reviews, look for discounts, etc.
  • Views on more than one device
  • Visits to physical stores
  • Other advertising (TV or Print or Email)
  • Offline purchases

Any or all of these things (plus more I am probably not thinking of at this moment) can contribute to a person’s decision to purchase or not purchase something. We have no way of knowing right now how many of these things happen either at all or for a particular individual, in what order these steps happen and what the final tipping point is to purchase/not purchase. Even companies with pretty sophisticated systems don’t get this right, because they can’t. Platforms don’t allow tracking of individuals like this, nor should they. We as consumers would consider that a tremendous violation of privacy. And it is. Even if you are logged in to a particular platform, which will provide more information about your process than if you were not logged in, there are still plenty of places in this journey that are simply unknowable.

People who work in B2B are already quite familiar with this as much of the process usually happens in technological blind spots, but because the focus is hardly ever on B2B for advertising platforms and therefore attribution, the more trackable and shorter timeframe consumer purchases are what generally make up attribution discussions. But attribution, real and actual complete attribution, is a myth. We are being told that a lot of this stuff is knowable when it isn’t.

So do we just give up on attribution?

No, I don’t think we should give up on attribution. I think we should use all of the information we have available to us to try to mold our advertising in ways that help the most potential customers turn into actual customers. I just think we need to have a better perspective on it and be honest with ourselves and our clients about attribution’s significant limitations. Let’s stop acting like we know more than we do.

In the old days of advertising, before digital, we had very little specific data to go on. We relied on things like an increase in phone calls or people bringing in a printed ad or coupon to a store to tie a purchase back to a specific advertising initiative. Really sophisticated advertisers used tracking phone numbers or phrases like “tell them Ed sent you to save 20%” and those things were tracked. It seems funny to think about still being somewhat in this same type of space with all of today’s technology advances. But in some ways, we still are.

E-commerce sites are probably best positioned to have attribution be most accurate right now. Especially sites that tend to convert/not convert in a single visit. Those types of interactions are easier to track reliably. It’s the ones that have a meandering journey that still are not. I’m not sure how well we will ever be able to track the meandering journeys, at least not until we decide that privacy is no longer a concern and we all consent to having everything we do be tracked and reported or maybe we will all be microchipped some day?

Until then, our job as marketers is to use the tools we have available to pull real data and do our best to tease out other influencing factors for customers. Have a full spectrum view of how your potential customers might progress through a journey from discovery to purchase. Identify places they might go as they go through their decision making process. Put content there. Monitor reviews of your product or service on popular review hubs. Track when your advertising happens (particularly if you’re using social or email) and look for corresponding spikes in traffic to your site, orders, inquires, client’s CMS data, etc. Reasonable attribution requires a big picture perspective and gathering of data from multiple sources for analysis.

What do you think? Am I crazy?  Do you have secrets to connecting the dots? As always, sound off in the comments or hit me up on Twitter (@NeptuneMoon).

Comments

  1. Well, I would like to say this is the best post I’ve ever read about “why attribution is not as good as we think”. Congrats. I loved it and it’s highlighted for reading again in the future.

    I’m a CEO for an attribution company and I’d like to comment few points of your post 🙂

    I agree, probably cookies are not the most fashion and trendy technology, but it still works, even with mobile devices, obviously if users are not accessing through APP.
    Even, users are deleting cookies, adblockers or anonymous navigation… cookies still works, basically because the people that are already blocking cookies are not still enough for deprecating a technology. Should we start thinking to track with other technology/methodology? absolutely yes!, we have started to work and develop our owns algorithms about digital fingerprint. We are currently using with our Ad Fraud Technology.

    Thanks to Digital Fingerprint we can detect the device ID that is clicking in your ads, deleting cookies and changing IPs with each click.

    How can we solve the cross-device tracking? As you said, it’s impossible if you don’t have your own browser with login (well done google!). We’ve tried to minimize this situation working with retrospective stats. I mean, once we have an ID, we check if we have this ID in our Database, and if we find it, we can recalculate completely the cross-device journey of each user. Our clients are working to promote to their users to connect to their webpages from different devices, with the goal of matching cross device data. Currently, it’s working very well with ecommerce, confirm a purchase via email AND WhatsApp.

    As you said:
    “To really know the process that a person follows, data must be collected on the following AT THE INDIVIDUALLY TRACKABLE LEVEL:
    Ad views
    Ad clicks
    Landing page views
    Other searches
    Views of sites that are not yours to read reviews, look for discounts, etc.
    Views on more than one device
    Visits to physical stores
    Other advertising (TV or Print or Email)
    Offline purchases”

    Completely agreed !! 🙂

    Our software tracks each one of previous bullets! And even we track microconversion (download pdf, add to carts,…). While richer is the data, more accurate is the attribution.

    As you said: “Reasonable attribution requires a big picture perspective and gathering of data from multiple sources for analysis.” once again, I agree! And for this reason, we have not done that! 🙂  You’ve said: AT THE INDIVIDUALLY TRACKABLE LEVEL and let me add, “in real time”. Why in real time? Because ALL attribution models are sensitive to any change. I mean, every change produced is going to change the value of the “attribution value”. For example, if you improve the CRO the behavior of your users are going to change, if you improve your SEO rankings, if add new broad keywords on PPC, if your server works slow today, if your company increase/decrease prices,… all these changes are going to change the behavior of your users. Even, if your favorite basketball team wins a match, if it’s raining,…… getting worse: if your competitors increase/decrease prices, server slow…. All these changes are going to change the journey of your user. For this reason, we CAN’T work with a rigid attribution model, we need to work with a “liquid”, flexible model, that is changing and it’s different for each one of your users.

    If you let me add… and this liquid attribution model, that is really tracking the influence (instead attribution) of each click in a single journey, should be connected to Adwords, BingAds,… to increase/decrease bids and budgets based on attribution data, and those channels that we can’t connect via API, access to a “recommendations” screen where the marketer know exactly how much and where should invest the marketing budget.

    Thanks for sharing your thoughts. 🙂

    • Neptune Moon says:

      Rafa:

      Thank you for sharing your very thoughtful comment on this topic.

      Real time is definitely a factor and one that I think will continue to grow in importance as our delivery of advertising and/or offers get more sophisticated. Beacons, or similar technology, could be amazing for delivery of real time ads (and would make redemption/tracking easier!).

      I think there is always going to be a tension between privacy and attribution though. You mentioned the Digital Fingerprint in your comments. If you asked most people if they would be ok with companies tracking them via a digital fingerprint of their devices across the web I would almost guarantee that most people would seriously balk at that idea.

      As someone who works in this field, I am very interested to see how the opposing forces fight it out on this one!

Trackbacks

  1. […] Then she’ll explain why she thinks attribution is kind of a joke. […]

Speak Your Mind

*