The Economics of a click vs. impression

Today no one can run from the economic downturn we are all facing. As the credit crunch continues to tighten its grip on the economy we are seeing advertisers demand more from every dollar they invest. In recent years we’ve seen the rise and rise of digital marketing as an effective way for brands to invest their dollars due to two key factors; the channel is the fastest growing on the planet and perceived level of accountability it offers. What is beyond me though is the lack of transparency and understanding most clients have around their digital investments and the blind faith they place in the reports they get from their media agencies. Without wanting to be alarmist most acquisition and engagement focused clients could save upwards of 30% of their digital display budgets with little or no impact what so ever on their campaigns performance. Even worse they can in most cases reinvest this wastage into performance based media and Search for a 50% – 200% increase in campaign performance.

The problem…

The primary problem is most (not all) campaign performance reports delivered by media agencies use blended metrics of post impression vs. post click success events (CPA etc). In some cases I’ve seen this blended metric as high as 95% post impression and 5% post click. With cookie windows of up to 60 days this introduces a high level of bias into reports. But today the problem goes deeper, some digital media plans are unethically designed to introduce a cookie spraying methodology into the media plan. Cookie spraying is a deliberate tactic to buy as much low performing, low placement, cheap remnant digital inventory  as possible. This type of media strategy can ensure upwards of 70% of an available Internet population ( e.g. Australian internet users) ALWAYS have a cookie on their machine.  In some instances I’ve seen a single client drop 4 billion post impression cookies in a 12 month period, with only 13 million active  Internet users in Australia you do the math. So a user may never ever click, engage or even see a brands ad, navigate to the brands site either by direct entry of another medium ( e.g. Search) and the value is attributed on a post impression basis to an advertising / media plan that had little or no effect on consumer behavior. Think how many times a consumer may visit a social networking site (www.facebook.com and www.myspace.com), use an IM program ( Microsoft Messenger, AOL Chat etc), use a free email service ( sponsored ad email) (Yahoo mail, Hotmail etc), or check the news or stocks from their favorite site. Every time they do this a cookie is dropped on a users machine.

The economic impact:

  1. Clients are relying on false or skewed data to make digital media investment decisions
  2. Clients are double paying for performance in some instances ( i.e. paying an affiliate commission, the cost of an impression or a Search click) – lack of ability to de-dupe
  3. Clients are missing their true value creation opportunity as their campaigns are being optimized and planned on fictitious data sets and not investing in high yielding digital media
  4. Some media agencies are falsely claiming media rebates and commissions they are not morally entitled to
  5. Some media agencies are doing this by deliberate design to meet their contract commitments to networks and publishers to gain OMI (other media income) and discounted wholesale rates that they mark up due to a group buy – all in the face of not operating in a clients best interest and not disclosing this as part of their strategy ( why would they?)

The impact on performance numbers

The impact on this practice on a clients performance numbers is significant. In some cases I’ve seen digital success metrics overstated by by to a factor of up to 20 times.

Example

Industry                              Blended Metric                                Post Click Metric(actual CPA)

Telco                                    $86 CPA                                             $679 CPA

Credit Card                         $54  CPA                                            $1894 CPA

Homeloan                          $205 CPA                                            $29,452 CPA

The Solution

  1. Base all value attribution ONLY to last cookie (last cookie should win in all circumstances)
  2. Understand the difference both post click & post impression metrics. A post click engagement is worth SIGNIFICANTLY more than an impression.
  3. Have a common cookie window policy across all digital mediums (display, affiliate networks, search etc)
  4. Understand conversion latency (time from click to conversion – they differ greatly between digital mediums)
  5. De dupe conversions / cookie pools etc
  6. Understand the consumer journey and what exposure a consumer has had to all digital advertising (what has lead to what)
  7. In the case of post impression metrics attribute value only within a very finite window ( i.e. 6, 12, 24 hrs from LAST impression)
  8. Apply a universal set of data business rules to digital value attribution
  9. Break down your media buy in transparent terms. Understand what components of the digital media buy is CPM, ROS (run of site) CPC, Remnant inventory etc.  This will allow you to understand cookie spraying if it is going on. Also get all performance reports broken down on the same basis, the results may shock you.
  10. Optimize a campaign investment on your value creation terms, not the agencies hidden objectives of OMI, media commissions, groups deals etc.


Examples of Cookie Spraying – you placement / low yield inventory

Woolworths Everyday Money - Ad at bottom of page below the fold, little or no post click value

Woolworths Everyday Money - Ad at bottom of page below the fold, little or no post click value

Low yielding HSBC ad - significantly below the fold, little or chance of engagement or a click

Low yielding HSBC ad - significantly below the fold, little or chance of engagement or a click

Europcar - Display ad at bottom of page significantly below the fold

Europcar - Display ad at bottom of page significantly below the fold

About these ads

12 responses to “The Economics of a click vs. impression

  1. So one one hand you say…
    “Base all value attribution ONLY to last cookie (last cookie should win in all circumstances)”
    yet on the other…
    “Understand the consumer journey and what exposure a consumer has had to all digital advertising (what has lead to what)”

    So which is it? Clearly there are other impacts on a customer journey other than the last cookie (or click) so your first point would really be a retrograde step. There is value in most steps of the consumer journey, so the trick is identifying which ones, not just the last capture channel.

  2. Oh, and obviously you working for a search (ie click based) agency, you’d want as many sales to come under the last click as possible, really, as search is often the channel (being a pull channel) that gets the last click attributed to it.

    Now I’m not justifying cookie spraying – I think it’s abhorrent and I know of an agency currently spamming 14 day PI cookies around the net for a large campaign – but I think we have to acknowledge that just because a branded search gets the last click, doesn’t mean all the value should be attributed to it.

  3. Media Dude – I don’t disagree with you at all. I believe all media can have an impact on a consumer actions and conversions. Lets just look at what actually leads to a conversion within a defined time window with some logical business rules. If a consumer has had exposure to display and a Search has occured within 12/24 hrs then understand the consumer journey / exposure and reward the other medium. Anything more than that is self serving.

    Yes I work for a Search agency (currently, prior to this all creative agency side), but my motives aren’t mixed. If display can claim the conversion, then claim it based on a common set of business rules.

    Lets just stop the absolute BS on post view CPA’s and the extreme wastage that is going on.

    Also all conversions from a last click do not come from branded term searches, some do but not as many as you may think.

    Thanks for your comments & thoughts, appreciate the debate.

  4. Hi Justin,

    I believe we’ve come across this discussion many times before particularly as we’ve worked together and i both agree and disagree.
    I agree that paying out on a post-view CPA thinking that it equals a sale is incorrect BUT (and it’s a big but) if you change your perception of CPA buys to actually look instead at what your eCPM is (effective cost per thousand impressions) then you may find the results are really quite interesting and you are getting low cost display ad coverage with high reach frequency which could actually be costing you less on a CPA rate than if you bought on CPM….

    i wrote a piece on my blog a few weeks back about exactly that: http://quintessentiallydigital.blogspot.com/2009/01/performance-performance-performance.html

  5. Hi Zoe,

    Don’t get me wrong, I believe there is a place & role for everything. No one medium is the silver bullet in digital marketing. There is a role for display in building a brand & that has an impact on all other digital mediums that yield a return for an advertiser.

    I just believe in simple terms that there should be a level of transparency for what is and isn’t creating economic value for a clients investment. I don’t believe that for CPA focused clients, that some current practices (i.e. cookie spraying, post view CPA’s 95% post impression / 5% post click) give them a true picture of what value they are really getting. If you ask most clients, they don’t understand what is actually being reported and if they understood the data they’d make different decisions based on rational insights.

    Some current practices & reporting techniques are biased in the agency’s / publishers favor and not in the best interest of the client. All I advocate is transparency & a common set of business rules applied to all mediums.

    Cheers

  6. You have made some pretty sweeping accusations about media agency practices and their transparency, when actually I would say we are a pretty transparent bunch. Particularly when search and display agencies experience a harmonious relationship and share tracking tags. Yes, that does actually happen, and SEM usually benefits from this being the last click to win and all. Giving specific advertiser examples for clients you don’t work on is interesting when you also don’t have access to their results. With slightly more insight into the campaigns you mention, I would suggest your comments are unnecessarily alarmist to say the very least

  7. Media Agency (Anna) – appreciate your views.

    Firstly, not really sweeping accusations – more observations from working on a range of clients. Have actively seen a bunch of misleading reports with a use of blended metrics. Again as I said above, some not all. Most clients also aren’t aware as to what constitutes the CPA’s that are being reported.

    Secondly, my comments are based on bringing a new level of analytical thinking for CPA & ROI focused clients & challenging norms and conventions to find incremental marginal value from every dollar spent on digital media. With some (not all) media plans having a heavy bias towards cookie spraying and post impression metrics I truly believe this issue needs discussion.

    I also don’t have the conflict of trying to support one medium vs. another from a legacy business model and revenue perspective. It is the freedom coming from working for an independent.

    I’ve never stated anything other than the application of common business rules about post click metrics and conversion latency applied to all CPA focused digital media. I believe in my current role at an SEM agency, I also believed it when I worked in creative agencies as a Head of Digital / Direct & Data. The role of any good direct medium as experienced DR people will agree is to reduce the wastage across any medium.

    From that perspective (again for a CPA based client) let the data tell the true story in full transparency. The majority of large clients (again not all) I work on don’t get full and open transparency in their reporting. I’ve never seen a business rule sheet with blended CPA metric business rules appended to a weekly / monthly report. If you do I think that is brilliant & congratulations.

    Thirdly, never disputed role of display. Only the way it is bought & reported to a client. Harmony in cross digital media usage and impact – not disputed, just believe a common set of business rules should apply to metrics across all mediums. Cookie spraying should NEVER be supported and then claim false CPA metrics on a majority of post view metrics with large cookie windows (14 days plus is a joke).

    Specific examples – sorry they are campaigns our teams do work on and the metrics are real that our clients have asked us to comment on after they give us full campaign data, it is the way we work with the majority of our clients to make informed rational investment decisions. Mostly we get asked to help develop value attribution business rules across digital media that serve the clients business first and upmost.

    I could also give you a bunch of other metrics we’ve tested with a range of clients who stopped questionable “performance display” who actually had no drop in any volume acquisition numbers, only significant drops in CPA as the wastage was reduced.

    Also as this is my personal blog, I don’t deem it appropriate to divulge confidential campaign stats related to any specific client. Surely you can understand that.

    If you work for a media agency that reports with 100% clarity then there is no issue now is there? You’d be happy to discuss this openly – yes? Because you’d be reporting across all post click & post impression metrics separately, applying confidence intervals statistically to conversion latency and measuring the recency & frequency of an impression with insights and business rules about how impressions lead to conversions across any digital medium right? You’d also get active buy-in from your clients on what the level of blending should be, incorporating conversion latency, cookie windows etc. Also if you use a common tracking tag for your clients then none of this would be an issue at all…right? If you are doing all this then why the agro?

    Again – as I stated, this applies to some, not all agencies. If your agency is doing all this I applaud you.

  8. advertising analytics

    Justin,

    couple of observations:

    1. cookie windows with a statistically insignificant time period (i.e. 60 days) are as much a problem post-click as they are post- impression. Don’t kid yourself. They are pleny of methods to manipulate post-click as there are with post impression.

    2. cookie windows need to be based on the purchase cycle of a particular product and calibrated towards the number of ad providers and channels you use. Not much use to give at static one-fits-all number; dangers are that you either price yourself out of the market or overpay.

    3. Below-the-fold placements are a cheap and easy target but you could be wrong in more than just one instance: exit placements under articles have been proven to be very effective not just for display: some of the highest yielding ad sense placements for publishers are there.

    4. No doubt cookie spraying and stuffing is a significant problem – but so is the widely spread belief among misinformed clients that there is no difference between a branded display ad and a distribution text link.

    • Thanks for comments, always appreciate feedback.

      In order:
      1 & 2 . Couldn’t agree more, regardless of cookie window length the most important metric in my view is conversion latency periods. From my experience the vast majority of conversions usually occur within a very small time period. We have some clients that have upwards of 75% of conversions occur within 1 hr of a click, others up to 2 days. Anything more than 7 days for the majority of product classes is kidding yourself. My issue is with mindless spraying of cookies that result in up to 80% plus of the internet audience having a persistent cookie on their machines.
      3. Below the fold – maybe? Maybe it is also because they are more contextually relevant to the page they are being served on? Lets not kid ourselves though, the vast majority of below the fold placements are cheap, remnant inventory that bias the case of the publisher & media agency, who in the majority of cases have the same motivation.
      4. Agree – I think there is a place for everything & branded display is very important. If it is branded display then it should be treated that way & not presented in a blended way 7 passed off as CPA based acquisition. Maybe you’ve seen more transparent reports but the majority I’ve seen try to masquerade as CPA on post-click blended metrics.

      Thanks for your comments & views.

      Cheers

      JH

  9. I have recently presented evidence of a 3-week delay of search based on display activity for an automotive campaign that ran over a 6 month period.

    Think about consumer journeys… you are on a bus to work monday morning, you see a sign saying “20% off…” are you:
    A. getting off bus NOW?
    B. going back in lunch-time?
    C. going back at weekend?

    So in real world, consumer behaviour is so drawn out, give a little slack to the reality of people’s lives and stop keep looking for instant-recipes to a complex issue…

    the end of the day, before companies like Eyeblaster were able to group display and search activity together, we were all looking at skewed data….

  10. Pingback: Interview with Justin Hind, Chief Operating Officer, Downstream « Maloney on Marketing

  11. Pingback: Economics of New Media Formats – Cut the BS please publishers « Economics of Advertising

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s