Removing the Measurement Cap !

Measurement hasn't perhaps got the attention across the wider marketing ecosystem that it deserves. While the challenges of digital measurement - Person vs Cookie, Third Party vs Walled Garden, Cross-device/platform measurement and attribution, etc.- are out there in the conversation , it tends to be mostly among specialists.

The cost from the lack of measurability should be assessed more rigorously and appreciated for what it is - a cost. And this is where a ground-level micro perspective may play a part.

For example, in the simple - though not simplistic! - case of skippable online video (OLV) advertising when average viewing frequency per cookie is known , the duplication rate makes a significant difference to the actual 'person'-reach and therefore frequency and cost.

Imagine a hypothetical OLV campaign of 500,000 paid views with a specific creative copy with an average VTR of 20% targeted at a demographic base of 1 million individuals with an average cookie view frequency of 1.5. As the chart at the bottom shows, depending on what the duplication is in reality, reach can be anything from 33% all the way to 5% and below. Cost per reach obviously increases accordingly.

But it goes beyond just  cost per reach.

Assume that the actual 'person'-reach is close to unique cookie viewers, i.e. an actual 'person'-frequency of around 1.5-2.0 ish (meaning the user wouldn't watch the same spot more than a couple of times or three - a reasonable enough assumption on anecdotal evidence in the absence of anything else ), what happens when another 500K views are bought , say, in the following month ?

Now, at 20% VTR, around 2.5 million impressions would have been already served first time around to generate those 500K views. These impressions would have covered most if not all of the addressable 1.0 M TA base already. So in terms of incremental audience in Month 2 , what are the chances that users opting to skip or drop out the previous month would choose to view the same spot now ? Or that those who viewed it last month would view it again this month ? Neither case is impossible or even improbable but ,well,it doesn't sound very probable either ! The math just does not stack up. Now this becomes not only a question of X% additional cost per reach but also the very tenability of the campaign, i.e. the possibility of a 100% additional cost

This example is obviously illustrative - and,yes, extreme ! In reality, a buyer would take audience size and related info into account before deciding on the buying volume. Equally important, copy would be refreshed regularly. And this is only a case of purely demographic-targeted buys which in reality is a relatively small number of buys.

(Measuring outcomes differently - say, through engagements, etc - doesn't affect this argument. Firstly, they are not mutually exclusive - one doesn't preclude the other. Second, this goes for those measures too, e.g., substitute 'click' for 'view' and the same Person vs Cookie discrepancy holds. Third, 'engagements',for example, tend to be low and stable in this format and , moreover, still a function of scale)

The point here though was to illustrate costs that may fall in the cracks of micro campaign management away from headlights and headlines.Should they - and numerous other more complex cases across channels and formats - be thought about, quantified and aggregated, it could provide the urgency and push which would expedite the move towards better measureability sooner.

Almost the entire illustration here is conjecture built on assumptions. Only the facts and figures could really tell. And for that to happen requires an understanding from the ground-up and cooperation among both marketers / buyers and platforms / sellers. Most questions are not easily answerable and require advanced measurement , including (especially ?) Third Party but some 'clues' could also be provided by platforms - for example, viewing distribution even if at cookie level. The buyer needs to think about that and ask , the seller needs to appreciate the market growing potential of such moves and provide ! The onus is on both because the benefits go to both.

As a post script : talking about frequency and such leads me to a sign off on Frequency Capping. It's a no brainer that Frequency Capping is a huge boon in today's digital era.

But how meaningful is it in the context of served impressions for skippable videoads with completed (or at least paid) views as KPI ? In the above hypothetical example of 20% VTR and an average View frequency of 1.5-2.0, how meaningful is a Frequency Cap of , say, 5 (or 4 or 6 or 10) here ?

When it is highly unlikely that a person will watch your ad three or four times , the cap becomes redundant at best. And at worst, you are limiting the chances of future exposures by not serving it again.

Also,as an aside, when a viewer has actively opted to watch an ad multiple times, is that still a waste ? One to ponder

Two Elephants in the Ad Measurement Room

The first elephant is the difference between digital consumption and digital advertisingconsumption. All the usual issues of viewability, bot fraud and, well, plain unnoticeabilty - and I'm looking squarely at you here, little banner ads ! - is why the percentage of ad exposures will remain relatively stable even as online consumption grows exponentially.(Though,obviously, the absolutes of ad exposures will grow with it)

The good thing with online video advertising where the difference is clearly quantified as the difference between ad impressions and ad views is that the elephant is easily sighted, understood and can be responded to.

The second elephant in the room is the less obvious one : TV ad avoidance.

There is simply no way of knowing whether people really watched your ads on TV. All that commercial ad break ratings tell us for certain - and this is keeping aside markets such as ours in MENA where these are not even available in the first place ! - is that people didn't switch off the TV and didn't zap channels.

Whether they walked out of the room or switched their mind off or buried their noses in their phones or , indeed, watched the TV commercials with love and adoration during those three or five or ten long minutes we do not know. And have no way of knowing.

So in effect we are penalizing online video for being transparent while not holding up TV to the same level of scrutiny and accountability. So TV ad exposures* all over the world are likely to be overstated simply because of a quantification gap (see figure)

* This post is only from the limited point of view of ad exposure. Engagement, impact, sales outcome et al are a different- though surely correlated !- matter.

While realistically speaking this gap can't be eliminated, can it be reduced ? Perhaps through syndicated sample survey-based research or,say, through more pervasive individual advertiser-level A/B experiments ? Hard to tell - but as viewer consumption boundaries blur and the market battles intensify, more attention will probably need to be paid to this TV elephant to size it up to some reasonable approximation.

(click to enlarge picture)

Data and the problem of plenty

"Won't be nothing, nothing you can measure anymore" - Leonard Cohen : The Future

As I listen in remembrance to one of the greatest pop poet-musicians of all time in the wake of his passing, I am also thinking of the problem of plenty in marketing data today. Namely : the more data there is the less data there is.

That paradox sounds (almost) Cohenesque but it's certainly a lot less than poetic for us practitioners out there !

Marketing data today has immense depth but very little breadth. There is a treasure trove of KPI metrics for each digital 'walled garden'* in your marketing mix but little or nothing by way of the same metrics aggregated for the entire digital mix (let alone the entire overall mix as a combination of online and offline)

(* I borrow the term in this context from an excellent piece by senior and very respected industry leader Gowthaman Ragothaman which you can read here)

Later if not sooner this is likely to place at least some constraint to business growth for these advertising dependent gardens - and which is why I am sure a solution will arrive sooner rather than later ! The recent announcement from Facebook (read here) about the launch of a 'Measurement Council' along with Third Party verification measures is a step in that direction. More will no doubt follow from more players. After all, it is some very important value delivered to their customers (viz. advertisers and agencies) - and as we all know to superior customer value goes the spoils !

Even leaving the whole third party thing to one side , improvement in existing reporting within each single platform is where it can and needs to begin. An obvious if simple example is unduplicated individual person-level reach. Sure, it is complex but something that can be solved given the right commitment and resources. Far more complex problems have been solved after all ! The question is when it will be deemed a necessity rather than an also-good-to-have. I suspect it will be sooner rather than later. One step at a time and the rest follows. This diagram illustrates very roughly how that can unfold. We await that first step !