What Marketers Need To Know About Conversion Attribution

Data-Driven Thinking“Data-Driven Thinking” is a column written by members of the media community and containing fresh ideas on the digital revolution in media.

Today’s column is written by Alan Pearlstein, CEO at Cross Pixel Media.

I believe that retargeting has flaws and any active participant should be aware of them before engaging in the tactic.  The key flaw that concerns me is conversion attribution; retargeters are consistently taking credit for sales that should be credited to other marketing tactics and advertisers are not getting an accurate understanding of what advertising is working for them and what is not. In my opinion, aggressive retargeting is undermining our ability to track the effectiveness of online ads.

The problem all stems from the idea that the last ad shown to a user(or clicked on by a user) deserves full credit for a sale or conversion.  This concept implies that all ads deserve an equal weighting when measuring effectiveness but I think we all know that this is a ridiculous concept.  And in the case of retargeting, a significant portion of the ads that are being credited for a sale deserve no credit at all (or limited credit) because the consumers were going to come back to the site even if a retargeting ad was never shown to them.

I don’t think you have to be an online media pro to easily grasp this concept – most people don’t buy on the first visit to a site.   Shoppers will research a product before buying it and visit a site multiple times before making the purchase.  Since retargeters start dropping their cookies (view or click-based) the moment after a person visits a site, it is inevitable that a portion of the cookies will be placed on users that were going to come back anyway. When this occurs, the retargeting effort ends up getting credit for a sale that was initiated by another marketing activity, completely screwing up conversion attribution. Does that mean we should not use retargeting?  Absolutely not – there are some people that might not have come back to purchase unless they were exposed to a retargeting ad that lured them back.  But the question is how many?

Under the current model retargeters take credit for every single person with their cookie that that comes back to the site and makes a purchase. And while the problem is exacerbated by view conversion spamming/cookie stuffing, it is still relevant to click-based cookies – just because I click on the ad it doesn’t mean I wasn’t coming back to the site anyway to make my purchase.

Another concern is that a significant portion of the sales that are generated by retargeting come from repeat customers vs. new customers. Obviously repeat business is critical for a company, but many ad campaigns are focused on new customer acquisition.  Is retargeting an acquisition tool or a retention tool, or both?  I would highly recommend that advertisers track these conversions separately to understand the types of sales that are generated by retargeting efforts.

One tactic that marketers are using to understand this issue is running PSA ads (public service announcements) alongside their retargeting campaigns to use as a baseline to determine how many people returned and purchased without being retargeted.

Advertisers are measuring the conversion rates that occur from people who saw PSA’s vs. the conversion rates of people who saw a retargeting ad and the results show that the difference between the conversion rates is nowhere near as high as retargeters would lead us to believe – a lot of the shoppers were coming back to the site regardless of whether or not they saw are targeting ad.  And when smart marketers identify the number of retargeting conversion that are new customers vs. old customers, retargeting campaigns ROI is further diminished.

It is my opinion that retargeting should be a part of almost every online advertising campaign because it is highly effective.   But we need to be realistic about its effectiveness and think we would all benefit if retargeters adjusted their pitch to focus on the incremental customers they deliver and not offer distorted numbers that don’t reflect its true value.

Follow Alan Pearlstein (@AlanPearlstein) and AdExchanger.com (@adexchanger) on Twitter.

Enjoying this content?

Sign up to be an AdExchanger Member today and get unlimited access to articles like this, plus proprietary data and research, conference discounts, on-demand access to event content, and more!

Join Today!


  1. Great post Alan. ‘Last touch models’ – will/must disappear and this is a trending topic. If a business is not building real-time attribution abilities into their offering/mashboard or do not know that advertisers and agencies are doing so right now – they need to research asap.

    Because TagMan are the leaders in real-time attribution reporting for advertisers (it is built into our Tag Management System including Natural Search and Direct To Site) – we have a clear’client X’ report that shows exactly what Alan is talking about and demonstrates where a DSP re-targeting campaign was awarded 5x more than it should on a last touch model.

    However – this works both ways, we can also show where display activity drove incremental business that was then claimed by brand search terms – and the publisher was not correctly awarded credit; so again – if you are not looking at how attribution effects your business on the sell or buy side; you need to, asap.

    Let me know if you want a copy of the Client X report. Happy to walk anyone through it.

  2. Alan — you’re right that advertisers must understand, *for every marketing channel*, the extent to which its contributions are driving new conversions.

    Advertisers should ask themselves, for example: “Weren’t some of those searchers who typed in my brand terms going to get to my site anyway?” Yes – and yet brand term keywords are purchased because marketers feel confident that traffic is incremental.

    Multi-channel attribution is, of course, the holy grail. Last Click Attribution is the problem that everyone acknowledges and everyone says they want to fix… and yet no one’s willing to pay to fix it.

    Why? Because it’s a zero-sum game whose focus is fairness — taking some credit away, giving it to others. And CMOs don’t care about being fair; they care about growing their business.

    Retargeting done well (and with the right business model) can be shown to drive purely incremental transactions. A:B tests with PSA ads as described above are easy to set up and validate — see question #2 here: https://adexchanger.com/platforms/tellapart-ceo-mcfarland-discusses-retargeting-and-its-incremental-benefits/

  3. Tony Chen

    Assuming that we believe that A/B testing works to measure net incremental, I think the real debate is to whether the advertiser is getting a fair bargain. The disturbing part is that the advertiser, many times, doesn’t know what a fair bargain is and here’s why:

    Many retargeting campaigns are sub-optimal not because of technology, but because of the advertisers’ own lack of clarity into the lifetime value of their customers. In a simple example, if I have a website that sells a $2 margin item and a $20 margin item, why would I set the same CPA goal across both items? Of course, fixing this with 2 items is easy, but what about 1000’s of products and customers with varied purchase histories?

    Attribution seems to be the least of our worries if we can’t even figure out what different customers are worth. To Josh’s point, maybe one day someone will pay to fix this, but if your equation looks like x = , looking at attribution only solves one part of the equation.

  4. Great post Alan, you raise some valid concerns.

    “Shoppers will research a product before buying it and visit a site multiple times before making the purchase. Since retargeters start dropping their cookies (view or click-based) the moment after a person visits a site, it is inevitable that a portion of the cookies will be placed on users that were going to come back anyway.”

    Using lists with different durations you can start retargeting at a later point in time. Lets say that a vendor knows that people on average come back 5 days to complete a purchase. In this case you might want to start showing remarketing ads 6 days after the person visited your website for the last time, rather than showing your remarketing ads immediately.

    Second, you ask wether
    ” Is retargeting an acquisition tool or a retention tool, or both? I would highly recommend that advertisers track these conversions separately to understand the types of sales that are generated by retargeting efforts.”

    Again, I fully agree. This also boils down to the question which message I do want to send to consumers who converted vs those who did not. Do not throw them into the same bucket but think about which is the appropriate message for someone who just showed interest in your product offering versus someone who bought your product. In order to transform the latter ones into loyal buyers you probably want to engage with them in a different way. Appropriately tailoring the message in your display ads is one of the key ingredients to retain those customers.

  5. Relevance Police


    It is definitely not true that advertisers don’t understand the LTV of their customers. In fact, the bigger ones clearly understand these metrics down to the different channels. Retail marketing is largely a transactional initiative – it is NOT about LTV of the customers they acquire. Given the diversity of online commerce channels and shopping buys, it actually does not make sense to look at LTV. Most retailers know this very well.

    Hence we are back to understanding attribution based on ROI – which is tied to revenue generated, and media cost. Also, 95% of the conversions happen within one day of the click or view. So, who takes credit for a transaction and the revenue generated from it is hugely important for understanding efficacy of channels.

    Also, a simple combination of average basket size and the price of the product to be presented is a good proxy measure of the value of the impression.

    • Tony Chen

      Agree on your point on LTV. I suppose there’s no accounting for loyalty in online purchasing. My main point is that attribution is only one part of the equation. It’s correct in thinking that everyone is more or less playing a complex guessing game regarding attribution, but I think the value you associate to the attribution (e.g. CPA goal) is not discussed as thoroughly as it should be.

      Also, I would disagree with the 95% of conversions within one day of click or view. I think it varies based on product. Some product categories are “fast twitch.” Others are not. Another way to look at this is the following: If 95% of their conversions happen in the first 24 hours, then those advertisers who allow longer attribution windows must be exceptionally generous.