The universal unit of advertising measurement is the issue that will define advertising research and analytics over the next five years.

Since the time of Don Draper, GRPs (Gross Rating Points) have been the predominant unit of measurement for Print, TV and Radio advertising.  Originally born at a time when advertising was not measured at all, GRPs are the product of % reach against a target audience multiplied by average frequency against that target (e.g. 20% reach x 3 frequency = 60 GRPs).  Traditionally this unit of measurement is reliant on syndicated advertising research surveys to estimate reach %, where a "random" subset of the population is asked to fill out a survey about media usage (magazine readership, etc..). These massive surveys that ask respondents to recall whether or not they read hundreds of magazines or TV shows (some are 100s of pages long) are then indexed and logged in databases where media planners can access the information and craft media plans designed to hit specific reach and frequency goals.

Up until about 4 years ago everyone was perfectly happy using GRPs as the predominant unit of advertising measurement.  On the off occasion when advertisers actually wanted to know the effect their advertising was having on the people who saw it, they could fund expensive custom research studies that flogged customers to death with survey requests and relied on shaky statistics to validate findings. However, even after conducting custom research, there was, and continues to be, no definitive way to track the effectiveness of advertising.

GRPs are not capable of representing how consumer perceptions change in response to advertising. Rather, advertisers are left to their own imaginations to project how effective their advertising is at changing consumer purchase behavior.  This limitation was accepted by advertisers and agencies who took comfort knowing that GRP measurement was "the best we can do." And at the time, it was the best we could do.

Enter digital media. Banner ads, paid search, email marketing, custom microsites, homepage takeovers, roadblocks, rich media expandable ads, online video and mobile ads.

The innovation of digital media allowed advertisers to somewhat effectively track advertising response using two rudimentary tools, internet browser cookies and HTML tags (that can look for browser cookies).  By planting browser cookies on every user’s computer that sees your ad, you can look back at the time of conversion (granted the conversion takes place online) at which ads that computer saw prior to purchasing a product or service.

All of the sudden the problem of tracking advertising response was turned on its head. Digital media is good at tracking advertising response, however, it’s lousy at tracking who actually saw the ad (the exact reverse problem that traditional media had with the GRP). 

Television ads have different lengths and content, but by and large they’re all pretty much the same user experience (all TV ads are videos that occupy 100% of the screen for a limited period of time).  Because all TV ads provide a relatively equivalent experience, it makes sense to assume all exposures have roughly the same impact on viewers.  Contrast that with digital media.  How do you compare a 300×250 banner ad to an email advertisement? And how do you compare those exposures to a paid search text ad?  Inside the realm of digital media there is such a diverse range of advertising experiences, you can’t possibly bucket them all together. You need to differentiate them with a measure of impact.

Looking back over our exploration of advertising measurement history, we can see three main categories of advertising measurement. 1) Reach and Frequency 2) Impact on the consumer 3) Response.

Most practitioners of advertising research and analytics would agree with me up until this point. However, from here, most thought leaders have tried to combine these three measurement categories into a single measure.  Everyone is so used to dealing with GRPs, the natural progression is to look for the “online GRP” so that we can compare all media in the same currency.  Rather than mash together apples and oranges, why not split advertising measurement into three separate units?

One unit will be a measure of reach and frequency – similar to the GRP, but split out into reach and frequency units (combining them was never a good idea because an appreciation for the proportion of reach to frequency is lost).

The second unit will be a measure of impact.  A standard of measurement needs to be innovated here, but for the sake of simplicity, let’s say that each advertising opportunity will be rated in "impact points" and it takes 100 impact points to empower a person with the knowledge needed to decide if a product is right for them. Note – this does not mean that all of the people who are reached with 100 impact points will purchase your product, as not every product is appropriate for every person (this is a separate variable). Impact would be best rated by an independent rating agency (the advertising equivalent to Moody’s or Standard and Poor’s) who could audit each opportunity (for a fee) and quote advertisers their unbiased rating of the opportunity.

The third unit of measurement will be response. It’s important here to note that response is not the same thing as impact. Response is when a consumer sees an ad and that ad guides them directly to make a purchase. Impact, on the other hand, is the cumulative effect of advertising to which each advertisement contributes, eventually empowering a consumer to make a purchase decision (moment of empowerment). Confusing response with impact is (in my opinion) what has caused the debate around advertising attribution and to some degree what has contributed to making paid search the most profitable advertising channel in the history of the world.

So where does that leave us?  My proposition for the future is that each advertising opportunity, whether digital or traditional (or something that has yet to be invented), be rated separately on all three attributes: 1) Reach/Frequency, 2) Impact Rating 3) Historical Response. These three units will help inform the next generation of media planners and overall create a more efficient advertising ecosystem.

The Future of the GRP
  • A few notes:

    I agree with you that it was silly to combine reach and frequency at all. It’s probably because reach and frequency are pretty much the same, from a COST perspective. It’s the number of times the ad is shown (on TV); if the same group of people are watching it over and over, or its a new group every time – well, the cost structute is identical. But it’s irrelevant for the ad buyer.

    Impact:

    1) What do you mean by “advertising opportunity”? Would that be an advertising campaign, or each “impression” of a campaign? I’m sort of confused – you seem to imply its by impression in identifying impact points and needing 100 to get “decisive action” for or against – but that seems both difficult to measure and disingenous. As you know, the “impact” for each additional impression, in each channel, is going to be different; I’d suspect the cumuluative effect would be similar to an S-Curve. But *order effects* (TV->are likely to matter as well, at least for some permutations – and teasing out that would be a bitch of a statistical job. Well, if you’re working with more than 3 channels.

    2) I mean, I think a real question to address is: What level of penetration (GRP?) is necessary in a population segment to drive action?

    You’d expect advertising to, at some point, generate word-of-mouth marketing if people begin to purchase it. Advertising is partly awareness; if you can dramatically increase WoM effectiveness in a population by “priming the pump” then broad-scale advertising become really, really hard to measure. How do you know when you’re ABOUT to get to that point, or if it won’t work entirely?

    When do you cut your losses? Dial down the spend? Or, do you pour more money into a budget you now realize wasn’t quite sufficient to hit that threshold?

    3) Since you mention rating agencies, would you measure the impact of each single campaign? Or throughout the campaign?

    Impact would have to be marginal, probably averaged over the number of prior views. As you well know, impact for the 5th time you’ve seen the ad is different from the first time. There’s likely to be a sweet spot, perhaps (cumulatively) some sort of S-Curve.

    The idea of having some unitless value, from 0-100, it interesting – because it’s analagous to the percentage impact before a consumer buys something – but I think it’s ultimately misleading. You’d have to have different impact measurements for different groups of people… a total average wouldn’t give you an accurate view of the impact across groups.

    4) As you mention, not all viewers/people impacted would be inclined to purchase, even if the advertising is effective. How do you measure that? Would you rely on survey data for that?

    In other words: how do you know the point at which people decide not to buy something?

    5) I’m also not sure of the value of “rating agencies.” That really didn’t work out so well in the financial industry, and I can’t see it working out better in advertising (due to incentive structure – first of all, who’s paying these agencies?).

    Response:

    I’d think response is the easiest thing to measure, here (which doesn’t make ti easy). That is, how much did sales increase based on how much advertising was purchased? Dealing with time-series data is always tricky, particularly since you can’t isolate advertising-driven sales.

    Unless by “response” you mean the “response rate” i.e. the rate at which actions (view/click) correspond to purchases? So, search ads have a very high response RATE, because people who have been exposed to advertising in the past and know what they want are inclined to click on the text ad. But the text ad isn’t responsible for the bulk of the impact the customer has experienced. Is that it?

    The “attribution problem” is really, really hard to grapple with.