Commentary

GRP Is A Lazy Metric, Exposure Isn't Performance

The discussion around online GRPs is the wrong conversation.

There's been lots of discussion about creating a Web version of the gross rating point, which is a number used to measure the size of an audience reached by a media vehicle or plan. It's the product of the percentage of the target audience reached by an ad times the frequency they see it in a given campaign.

This is an exposure-based metric created when it was very difficult to measure the performance of an ad campaign. All we could measure was exposure, so we did. Now there are many channels that allow us to closely measure performance, especially the digital channel.

The source of this exposure-based info is typically from Arbitron for radio, Nielsen for TV and MRI for print. Arbitron, Nielsen and MRI measure the size of an audience by asking people in a national projected representative surveys whether they watched, viewed or listened to a particular media vehicle.

While many have always been uncomfortable with the survey methodology to accurately measure all audiences, statisticians have verified it. These companies have done a very good job at devising the best possible solutions given the technology; after all you can't survey everyone in real time.

While I understand trying to apply a GRP metric to online media, I think it's the wrong discussion. Instead, I'd like to try and force other media to be more accountable. If other media can't be more accountable, they should lose budgets to those media that are accountable, trackable and measureable.

I'm not suggesting that all budgets should move from other channels to online, at least not yet. There are many marketing requirements that digital still can't deliver. I'm also aware that when we run multichannel campaigns, the overall performance of a campaign improves, e.g. TV drives faster online conversions.

But what I love about online media is that it's trackable and as a result, easily optimized. I wish post-view/listen/read for TV, radio and print were as trackable and optimizeable. In fact, I don't care how many people I reach online. What I care about is what they do. Because I can measure what they do, I don't care how many I reach.

The only reason why we care about how many people we expose to a message in those other media channels is that it's so hard to measure what they do after they see a commercial.

A popular argument for exposure and reach, particularly around brand budgets, is that we need to inform lots of people about our brand proposition. We're told that brand awareness is very important because it's an indication of importance and vitality. I'm not so sure if that's true any more.

I don't care if people know about a brand if they're not going to take some action, either by buying it or promoting. Online, I can measure both. I can easily create ways for consumers to buy a product or take an action that indicates with a high probability they will buy the product, e.g. request for more specific information). Or, I can persuade them to share or comment on a message in a way that allows them to promote the brand.

And now, I can even assign attribution to each digital exposure. If it takes 30 digital exposures for a consumer to take an action, I can assign a value to each and every exposure.

In the past, we also used the argument that it was important to create brand awareness so when consumers were ready to buy, they would know a particular brand existed. Now that we can tell when people are in-market, predict when they will be in market, and begin sending them relevant messages based on behavior, creating awareness months ahead is a very expensive marketing tactic few brands can afford. We no longer have to spray and pray, nor should we.

It's no longer enough for advertising to just create awareness. Awareness without a meaningful measurable action is a wasted budget. A marketer that doesn't ask a consumer to take a measurable action is just lazy.

GRPs are lazy metrics because they're just measuring exposure. I don't care how many people are exposed to my message; I care what they do as a result of seeing the message. Our communication should be constructed so we can capture behaviors and we should invest in media that allows us to do that.

We shouldn't sell plans based on how many people have an opportunity to be exposed to our message. Our clients deserve far more certainty. It would be far better for our industry to create metrics tied to performance than exposure.

8 comments about "GRP Is A Lazy Metric, Exposure Isn't Performance".
Check to receive email when comments are posted.
  1. Paul Benjou from The Center for Media Management Strategies, August 15, 2011 at 9:18 a.m.

    You can't teach an old TV dog new tricks. Try as you must, I've been at it for 35 years and the TV marketers listen but refuse to budge from the "GRP gold standard".

    So ... What's your plan?

    Paul Benjou
    www.MyOpenKimono.com

  2. Jessica Lampron from Mediabids.com, August 15, 2011 at 11:03 a.m.

    On the print side, we've been developing a performance-based program that allows advertisers to pay for print ads in certain newspapers & magazines on a per-response basis (call/lead/sale).
    I think your point is exactly right; with definitive measurement tools available for online campaigns and a way to pay for online ads based solely on response, the bar has been raised for other mediums to provide similar offerings. - Jessica Lampron, http://www.mediabids.com

  3. Paula Lynn from Who Else Unlimited, August 15, 2011 at 5:46 p.m.

    You need to talk to Dave Morgan.

  4. John Grono from GAP Research, August 15, 2011 at 8:42 p.m.

    Where to start?

    First GRPs is a communications planning tool that is not TV specifc. TV has just done a darned good job at utilising them.

    The 'beauty' of GRPs is that they are pretty much media agnostic - they are a theoretical construct that measures the 'weight' behind each medium in a campaign. The problem is that the way GRPs are calculated and utilised by each medium - not all GRPs are created equal.

    While I share your enthusiasm for online tracking, I have also seen serious flaws in such an approach. The first (and most obvious) is that the attribution of the action (a click, a purchase, a video view etc) is generally credited to the last click - ignoring all previous actions. The second is that as you only have identified online data all attributions are credited only within the online sphere.

    Smart marketers are aware that not all GRPs are equal, that not all sales are online, and that attribution within the online sphere is myopic.

    So, what to do? The use of sophisticated econometric models that link to sales (generally) provides great insight. The key reason is that the model doesn't really care that you are pumping in GRPs for TV, ads x readership for magazines, and clicks (or whatever the latest cool metric is) for online.

    What the models do (in simple speak) is say ... well look at that ... every time I put 'x' GRPs on TV with 'y' magazine ads while getting 'z' impressions online then my sales spike .... so let's have a bit more of that !

    Clearly 'x' GRPs and 'z' impressions is purely about the weight so within each medium the media vehicles chosen and their placement is also optimised.

    As I said, the smart marketers know this and look at the BIG picture. Brands like Coca-Cola have become behemoths by focussing on mass marketing to drive mass sales. They simply do not need to know who/how/why/when each person bought a can of Coke - they just need to know that when certain marketing plans are executed that they sell truck-loads of fizzy drink! I'm yet to see the marketer that has their annual goals and KPIs based on individual sales.

  5. Damon Ragusa from ThinkVine, August 16, 2011 at 9:15 a.m.

    I think I understand where Barry's POV is coming from. In a world where it is perceived that you get immediate measure of "real" consumer behavior it is hard to look at anything else. But everyone should be a bit concerned about the unadulterated acceptance of the "last click" phenomena advocated here. You may not care about those things that cannot be directly measured but that doesn't mean they don't exist and don't have an impact. The notion that if you can count something implies you can optimize is nonsense. You cannot possibly optimize digital media if you are, at the same time, utilizing non-digital media vehicles, without considering the impact of one on the other.

    But here is the rub, even the most sophisticated econometric models are still challenged at differentiating simultaneous and synergistic impacts across all marketing touchpoints in even the most rudimentary marketing plans. And in increasingly complex digital marketing plans even the tag/track methods can become intractable quickly. I think this issue has created some of the divide between those who count (digital analytics) and those who model (see MMMers). But there is a place in the middle that can utilize the tagging and tracking of digital media and audience/reach measures of traditional media. The common ground is in building from an understanding of how people behave. Utilizing dynamic, bottom-up models of how people use TV, radio, web, e-mail, text messaging, YouTube, etc. is the key to building an understanding of how all media interacts to drive the outcome marketer’s desire: sales.

  6. John Grono from GAP Research, August 16, 2011 at 7:52 p.m.

    Fair points Damon. In the modelling I have been involved with the simultaneous nature of cross-media advertising stimulus was a problem in some situations.

    We had an understanding client who, wanting to improve the predictive capability of the model, agreed to split-tests. In half the nation we ran simultaneous and in the other half we didn't - then we flipped it. The model had the data it needed.

    There was/is however a more fundamental underlying problem with these models (and indeed all advertising measurement in general). That is, the inherent assumption that advertising sells ... now. There is a lag between the advertising stimulus and the consumer action. This lag differs by category and brand of course, but also by advertising execution (though the latter is generally very small and can be largely ignored). However, the lag by medium can vary a lot. Clearly the online lag is much shorter than traditional media. (For example UK studies using the SuperPanel have shown the average to be a 16-day lag for FMCG products advertised on traditional media.)

    Thankfully, we were able to use 'raw' activity/audience counts and then 'smooth' the 'weight' of that activity over the upcoming weeks to reflect (i) that just because you advertise washing powder on TV on Tuesday doesn't mean that everyone rushes to the supermarket on Wednesday to buy (ii) memory ad-stocking effects.

  7. Neal Burns from University of Texas at Austin, August 16, 2011 at 8:10 p.m.

    Good discussion. To me the metrics of engagement and sentiment that we enjoy with on line media make it clear that GRP is what it always was . . . an approximation of content audience and an overstatement of advertisement audience delivered. And, the presumed delivery of a million viewers on TV is quite different than the engagement of a million who go , for example, to you tube to see a video.

  8. Brian Rock from Network Ten, August 17, 2011 at 2:11 a.m.

    I don't understand why this discussion, like so many others, polarises into GRPs vs response metrics, instead of looking the benefits of GRPs and response metrics.

    In a very narrow sense saying "I don't care how many people I reach online. What I care about is what they do" is correct. Reaching 98% of people who ignore your message is pointless. Better to reach 5% who actually buy something. So far so good.

    However in a broader sense saying "I don't care how many people I reach online" is effectively saying "I don't care if my campaign could be even more efficient and more effective".

    Just to pick one example, if Campaign A gets results X with a 40% reach, it doesn't seem a radically crazy idea to suggest that reaching more prospects might improve sales. And if we have the appropriate response metrics we can find out if it does, and if so by how much.

Next story loading loading..