Stop Paying for Fraudulent View-Throughs

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 Josh McFarland, CEO of TellApart, an ecommerce data and ads platform.

Dear retargeting advertiser --

We all agree online advertising needs a more comprehensive metric than the click.  But the view-through (or post-impression) conversion is not it -- especially for your retargeting campaigns.  In fact, if you're being billed for view-through conversions from your current provider, you are massively overpaying for events which were going to happen anyway.  You've likely witnessed this first hand as you've tried to de-duplicate vendors' conversion reports, typically finding the sales were already attributed to another channel altogether.  Or perhaps you've had the joy of trying to understand why these view-through volumes outweigh click conversions by 10:1 (or worse!).

Why has the view-through become the default measure of campaign efficacy?  If Gian Fulgoni were to be believed, it's because the people who click on display ads are young, stupid, poor, out of work gamblers who spend an inordinate amount of time bidding for junk on eBay;  not only are clicks irrelevant, they're detrimental to the campaign!  Dubious studies aside, I believe there's a more powerful reason the view-through reigns today:  it's easily used to defraud advertisers.

As evidence, I invite you to install Firebug for Firefox and watch its display of third-party ad network traffic throughout your day on the internet.  Visit the homepage of CrateandBarrel.com and - bam! - sixteen (16!) ad network pixels are immediately loaded in your browser.  Then, as you navigate publisher sites which are using ad networks, you'll see how a single ad unit can piggyback tens of additional cookies!  And that doesn't even include Flash cookies or server-side (aka Pixel-Free) cookies owned by the likes of Acerno/Akamai, which are invisible to the user.

What's happening here is not new.  The affiliate industry, which went through a similar and painful learning curve years ago, has a term for it -- cookie stuffing:  the insidious practice of sprinkling affiliate code-enabled cookies on as many users across the Internet as possible... knowing some fraction of them will organically "convert" before the cookie expires, thus giving credit for events which in no way were affected by the affiliate's campaigns.  And retargeting campaign view-throughs are like cookie stuffing on steroids because the targeting focuses on users who are infinitely more likely to return naturally!  Unlike the affiliate world, however, these are not smalltime arbitrageurs;  this is cookie stuffing at scale, performed by big retargeting providers and major ad networks.

Now, it would be naive to assert the view-through has no place as a metric.  Indeed, my own company's rigorous A:B testing of our retargeting campaigns can prove that view-through value does exist.  But this validated view-through lift is complex -- massively influenced by all of your other marketing programs and general business performance, rising and falling in volume from week to week.  It's also hard to measure, requiring large (and costly) control campaigns to be simultaneously run.

And there's the rub -- there is no valid way to enter into a pay-for-view-throughs agreement with a retargeting provider.  Doing so immediately mis-aligns interests and incentivizes the vendor to spam your users with cookies in an attempt to take credit for sales which were going to happen anyway.

This leaves just one conclusion:  retargeting campaigns' value should be tied to the click, specifically clicks that convert.  You should demand high CTRs from your retargeting efforts and from any campaign that uses your own audience data for targeting.  Sorry, Gian -- the click is back -- and I believe you'll see an evolution of display advertising business models that embrace it:  either via click-based CPA or via CPC.

Remember, if you find yourself having to de-dupe your vendors' conversion reports, you've just uncovered the real issue:

You're being duped.

Follow TellApart (@tellapart) and AdExchanger.com (@adexchanger) on Twitter.

31 Comments

  1. Patrick

    So then what is the solution? Because it's hard we should just give up and start using clicks again, despite the very well-documented problems that brings?

    CPC and CPA display advertising are largely the result of lazy and ignorant advertisers. "Hey instead of optimizing my own campaign, since that's my job, I'll just pay you only when you make a sale. But please don't take advantage of that."

    "Cookie Stuffing" or any other gray-area semi-fraudulent activities are the cost of doing lazy business. There is real value in every view, click, and conversion, but advertisers take the easy route and only look at one metric to determine their campaign viability.

    If they are so myopic in their analysis, publishers, networks, affiliates, etc, would be foolish to not take advantage of their laziness. Advertisers are fully capable of doing real performance analysis but don't.

    So they get taken advantage of.

    Caveat Emptor pal.

    Reply
    • When buyers are well educated... caveat venditor.

      The solution - for retargeting campaigns at a minimum - is to return to the click as the single objective function, the measure against which efforts are optimized. And yes -- that optimization responsibility is shared by the advertiser.

      I'm not advocating the dismissal of the view-through as a valid metric, but I don't believe there is a way to charge for them that results in anything but these games being played.

      Retargeting vendors are given access to the purest form of predictive data possible -- if they can't refine that into clicks, then it's time to find a new vendor.

      Reply
      • Patrick

        When buyers are well educated... caveat venditor.

        Knowledge is power. Completely agree.

  2. View-through conversions are a unique beast… no question about it. Strict A/B testing can help validate lift but should be combined with additional analytics. Those include visibility into an engagement map for each conversion to understand the proximity to the sale, quantity of ads shown prior to the conversion combined with data showing user interaction with the ad prior to conversion. Your suggestion that the ‘click is back’ and advertisers should rely on this data point alone is off target. Yes – your retargeting provider should provide a very high click through rate but to suggest advertisers should continue to rely on the old-school thought process of 100% last click attribution is ….well….old-school. Attribution management is a huge issue in our industry and thankfully there are some very innovative companies out there right now working very hard to bring a new-school solution to the table.

    Reply
    • I agree, attribution management needs serious help. I'll note, however, the most innovative company in this space, Convertro, also pays no attention to view-throughs as a metric. (See AdExchanger Q&A on 2/18/10.)

      Advertisers should not rely on the click alone as measurement of a campaign's worth: e-commerce should measure incremental sales, movie studios should count the number of trailer videos watched, etc.

      But in all cases, the view-through is not a metric that retargeting campaigns can be optimized against, nor is it a metric that should be billed for.

      Reply
      • Josh -

        Your logic doesn't make any sense.

        Just because (in your words) 'the most innovative company in this space' pays no attention to view-throughs as a metric doesn't mean that view-throughs should be ignored for optimization.

        If you are unable to create a solution that combines all media actions (views, click, etc) - just say so.

        But statements like: "there is no valid way to enter into a pay-for-view-throughs agreement with a retargeting provider" are simply false.

        The solutions are out there - you just need to come out of your cubby hole.

      • "Optimizing for view-throughs" is a flawed notion because it implies causality.

        With view-throughs, the best you can hope for is to run test and control campaigns and measure the conversion rate differences between. Done correctly and with statistical significance, you can prove that the view-through lift must exist (through correlation), but you cannot prove causality between the ad view and the conversion unless you were to go out and personally interview the users who view-through converted.

        The only way to optimize around the view-through is to become really good at figuring out which users are going to convert anyway... and to then show them an impression before they do so. And that's exactly what I describe as happening above.

      • Josh,

        You can certainly optimize to view-throughs. You can identify (with controls if you like) which audiences (by user and/or site) are more likely to be influenced by the ad.

        There's no fundamental difference between the ability to optimize to view-through conversions as there is to optimize to click-through.

      • You can't measure the influence of an ad through its view. All you are proposing to measure is the composition of the converting audience... and then make sure you get ad impressions in front of those type of people. In that case, you're measuring the targeted reach, not the conversion impact.

        Young-Bean Song of Microsoft's Atlas Institute wrote an article last year which better emphasizes this point. It's called "The Dirty Little Secret of View-Through Conversions."

        http://community.microsoftadvertising.com/Blogs/Advertising/archive/2009/06/23/the-dirty-little-secret-of-view-through-conversions.aspx

  3. Whilst I think view through conversions should be questioned, it is detrimental and nieve to scrap them as a metric. Using your logic, surely a click conversion on a retargeted campaign may well have converted anyway, after all they had already looked at a specific product on an advertisers website? In this case should you then downweight the ROI calculation for a retargeted campaign? When we work with advertisers we use both click and view conversions as metrics but use empirical testing to understand how long those windows should be. We also try not to forget that our advertisers do a lot of advertising in other channels and that no online measurement metric, however much technology sits behind the measurement, is definitive as it only takes online advertising in to account.

    Reply
    • Adam Stalker

      Great response Martin. I completely agree. I'm not aware of any research that has been done evaluating the causation of a click versus the consumer's pre-existing propensity to convert. That would be interesting to see but would logically include consumers that were already going to convert.

      Quite frankly, this is one of the most short-sighted articles I've ever read by someone with any real experience in this industry. It's these kinds of statements from "industry leaders" that set our industry back instead of putting us shoulder-to-shoulder with our traditional advertising peers. Clicks, or their equivalent, didn't exist before the internet and yet advertisers still grew their brands and increased sales through "view-based" advertising mediums. I'm all for critical thinking and facing the harsh realities of our industry but Mr. McFarland, you're ignoring the history of an industry that has shaped consumers' buying habits for generations. Advertising wasn't invented along with the WWW. Please have some perspective.

      Reply
  4. Big Picture

    View based ads are in fact an older metric than click based ads and are very proven and not new at all. Such ads include all found on television, print and outdoor which all existed long before The Internet. Although I don’t entirely disagree with the specific point you make, this situation should have never happened in the first place and could have been avoided by working with a network who is a partner and consultant and not someone looking to turn a quick buck out of you.

    It has long been known that when running CPA that remarketing ads are very profitable to networks. What you may not realize is that the other portion of CPA that runs on RON is usually not profitable or marginally profitable for networks and of high risk. They hedge the success of remarketing with the largely unprofitable RON ads to deliver your CPA campaign that actually can scale. Dismiss the remarketing and you may find your CPA quickly is no longer offered to you. Don’ bite the hand that feeds you.

    When choosing a partner to work with, it sounds like you might have been mislead by a network interested in turning a quick profit. At our network*, we always recommend you don’t run remarketing as a standalone, and when you do we run them on CPM. Perhaps the real lesson of this story is: to chose your networks wisely and work with the largest and best players that have been in the business for years.

    I would also need to full heartedly disagree with your comment on CPA being the result of laziness. We’ve spent over $100 MM on optimization technology. We make real time decisions in milliseconds and can extend audiences via exchanges and determine the bid after we’ve sniffed out cookies. No matter how intelligent and capable you may be, you can’t process 10 billion transactions a day like our technology can.

    * I purposely remain anonymous as to not advertise, which would only cause people to assume I’m trying to sell something and discredit my opinion. I’m here to educate.

    Reply
      • Patrick

        Perhaps the comment is a bit hyperbolic, but written in response to what I interpreted as the regressive spirit of this article. Certainly not all CPA is lazy, but it does put all of the transactional risk on the seller, making it effectively entirely their responsibility to optimize performance. In doing so, advertisers sacrifice margin for security.

  5. I think the key point here is that some (not all) are using this brief moment of vulnerability in the pure click-based attribution model by gaming view-throughs to take unwarranted conversion credit. Josh doesn't seem to be saying that view-throughs are all together bad, but that it's important that advertisers are aware of the pitfalls. If too many marketers get burned, we will have missed out on an opportunity to improve the way we measure the success of display advertising.

    Reply
  6. Fooman Xhu

    Josh,

    Thank god someone has finally said it. The likes of Acerno have been pulling a fast one for too long.

    Reply
  7. Dean Weaving

    Hey Guys

    A point I would like to make is in regards to the re-targeting is who's data is it?

    Is it a case of just a pixel on a clients page and re-targeting previous users or in house data that was collected? Or even brought in?

    Does the data that was used to make the post view conversion not account to how much that user would or would not have converted?

    Reply
  8. CPM is an obvious move... depending on who you are working with you might have to deal with some whining from your account rep because they know the potential of VTC windows on a retargeting campaign. If for some reason you need to lock into a CPA, don't skim over your contract. Even paying 50% on a 7 day VTC window is probably too much in most cases.

    I've been working with one particular retargeting company for the entirety of my retargeting efforts. I've seen major successes because of the metrics they make readily available to you. Like Big Picture I'm not going to list them because I don't want to seem like I'm selling it, but a simple report from them and you will see that VTC isn't worth kicking to the curb. Although they are the backbone measurement of these campaigns, there are other figures which build view-through in, and are much more indicative of a strong retargeting campaign.

    Shop around, you will find that proving your campaign out isn't as hard as a lot of buyers make it out to be.

    Reply
  9. There is no doubt that saying view based conversion are fraudulent is a great way of getting everyone’s attention. Fraudulent is a strong word, implying someone is purposely pulling the wool over a marketer’s eyes. There is no doubt marketers need to use caution when crediting view based conversions. What we have been able to prove to our clients time and time again - through strict test and control studies, brand search lift studies, sophisticated attribution analyzes – is view based conversions have value. It is what I have been saying to clients for years - you shouldn’t credit 100% of view based conversions, but you also shouldn’t credit 0%. Each client will net out differently, but there is no doubt we are able to definitely prove value, much better than any method for proving TV has value…or print, or radio…

    Reply
  10. Advertisers probably should not utilize all major ad-networks at the same time to avoid overlap. A de-duplication algorithm and a powerful full funnel attribution model provider combined with a tiered pay out for click and view conversions will help you to avoid overpaying on view thru conversions. Big Picture is right - if you kill the view you will kill your CPA campaign. CPA campaigns with their inherent risks and high management costs especially if arbitraged will become unprofitable for networks. Live and let live.

    Reply
  11. I'll restate my thesis in partial agreement with the above two commenters:

    View-throughs have value. Their lift varies but can be measured when done correctly.

    But as an advertiser, you shouldn't be paying for them because you will not be able to prevent getting gamed.

    Reply
    • Being gamed or not. There are many ways to price risk into your view CPA payout - not to pay at all means you game your vendor. Most of Advertisers I work with running their campaigns ultra profitable on their back and and are sole DR focused paying for view. Again I can't stress enough to utilize full funnel attribution analytics with their help and a little math you can evaluate how much to pay for view based conversions for each specific vendor on the buy. Yes you might be off better not to pay at all for view thru - but you will not be able to scale your campaign as nobody has interest in running low margin CPA campaigns. Sacrificing a view percentage points margin on your side to keep your top vendors happy - will be rewarded with much bigger scale. Not without reason we see the tremendous move of hardcore DR affiliate advertisers into the display performance space.

      Reply
  12. If we start from the premise that there is inherent value in retargeting display ads to non-converting site visitors; then Josh's belief that view-through conversions have no value is completely incorrect.

    I can think of few mediums whereby an advertiser can have an ongoing conversation with people who are interested but haven't bought their product/service. Retargeting impressions are arguably the most valuable impressions on the web.

    When running CPA or optimizing to CPA, proper attribution is the key. Of course, many of the converters that viewed a retargeting impression would have converted anyway. The key is to determine what that percentage is. If a network thinks it should be 100%; then they are likely scam artists. If the network is willing to work with you and adjust the attribution so that the advertiser can arrive at an effective eCPM for the retargeting impressions, than the value is unquestioned.

    Reply
      • because the eCPM on a click-only CPA may be higher than your eCPM on a CPA that has blend of click and view.

        Make no mistake, the blend should be reasonable. I've worked with advertisers who ran CPA and valued views at 10%. They were able to acheive lower eCPMs than if they valued only clicks.

        The total volume of clicked conversions didn't go down either.

      • Think about this like an online retailer. Holding volume constant, you care about one metric: ad cost per order. It doesn't matter how you blend the numbers to make the eCPM *appear* lower. At the end of the day, you expect to spend X and get 2X (or 10) in revenue.

  13. @Davehonig

    Fair Points. You are not figuring all the " clicks " that were driven from Google resulting in a conversion, a conversion that was clearly helped by our friends mr 300x250, 728x90....... This is why ad servers today are weighting the conversion and taking away credit from Google based on rules. Now, I am not saying the VTC is perfect. But I challenge you Josh to practice what you preach and charge your customers based on a click to conversion. Talk to me in 6 months and tell me how that worked out for you.
    Perhaps its the advertisers fault and they should limit to only a couple of networks to run their re-targeting initiatives. Perhaps its frustrating to you because you are not getting the credit you feel you deserve. One thing is clear, low margin CPA campaigns are NOT scalable.

    The click is so far gone, you are speaking a language that many of us wrote 7 years ago, and amended 4 years ago. Times have changed, innovation is at blazing speeds, understand and use technology, it will allow you to think big picture but take small steps to get there, use data, amplify audiences and more importantly bring value to your clients. I can assure you, there is not one real brand out there that will see value to your strategy of click to conversions on re-targeting campaigns if its not scalable. Prove me wrong, come to NYC and dinner is on me. Have a great weekend!

    Reply
    • "But I challenge you Josh to practice what you preach and charge your customers based on a click to conversion. Talk to me in 6 months and tell me how that worked out for you."

      We accepted that challenge when we started the company nine months ago, and the model has performed fantastically and at scale -- feel free to call any of our clients for reference. 🙂

      http://www.TellApart.com

      Reply
  14. Josh - great post and obviously generating a good deal of comment from various sides. Attribution is a huge issue for us as an industry and VTC is not the answer as a standalone. View-through conversion are there to be gamed and anyone who has been in the space for a while knows how at least one very large ad network routinely bought millions of low-cost impressions in certain high-volume spots to get a cookie on users' machines to get credit for conversions happening anyway. IN a world of incomplete answers and overlapping intent/attribution, the click is much more defensible since there's only one last click. A related point - companies selling retargeting on a CPC or CPA basis as a standalone are (often) disingenuously taking advantage of clients who do not understand they are essentially leasing the company their data for free. It's easy for folks who work in a zero-transparency world to cross-subsidize crappy campaigns with retargeting on a CPA... when clients would be better off splitting them out, and knowing where their stuff is running, and also realize that one of the reasons advertisers want to run your inventory on a CPA is because it is long-tail bs sites and typically performs poorly. Better inventory can and should be inventory that can be sold on a CPM and command competitive pricing because an ad network isn't the only one who wants to buy it.

    Reply
  15. Hi Josh,

    Very interesting post, indeed. However, I have some insights about all this :

    1. When launching a retargeting campaign, you should ask for a PCC price and a PVC price. For example, 50€ for Post clic Conversion and 5€ for Post View Conversion. That's forst step.
    2. You should prepare a full Views/Clics reporting system. So that you can check for each user which page he has viewed and which ad has been clicked before the sale.

    Reply

Add a comment

XHTML: You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>