Forget CPM - It's Time For Time To Be The New Metric

There's something in the air and if you want to know what it is, take a look at Biz Stone and Evan Williams. The Twitter co-founders' Medium site has signed up its first content sponsor, BMW, and the big news is that the car brand is not paying by impressions or click. Instead, the metric is time.

It's been a prediction of mine for quite some time that while cost per click (CPC) will probably have a future in direct-response campaigns, particularly around ecommerce, cost per thousand impressions (CPM) as a metric will likely be replaced a time or audience share metric, or possibly a combination of the two.

Ad men and women cannot complain. As the industry has been mechanised, it has made CPM virtually meaningless. Why count how many times an advert has been served when more than half are unviewable -- although there is a new IAB standard here -- and an estimated half of those which are clicked on are selected not by a keen consumer but rather an ad "bot." The figures aren't exact and they vary from one source to another but the phrase that you waste half your money in advertising has never been truer than online display.

If you can't rely on clicks alone nor impressions, it makes sense that advertisers will seek to look for additional metrics.

This search comes at exactly the same time as video is the hot topic. It's massive because people genuinely like to be entertained in sound and vision, but there is no agreed viewability metric for the medium yet in place. It's being worked on but it's far harder to say when a video is viewable or has been viewed than it is to say at least half a display MPU appeared above the fold for longer than a second.

Video, however, has a metric that broadcasters have come to know and trust. It's not all that precise, but then the problem with mechanised advertising is it's too precise and so open to massive manipulation by "bots." 

Broadcasters measure share of a defined audience, and there is a common belief that this will carry on into the online world and could progress from video into other forms of advertising.

Native advertising would be the perfect medium for this, particularly around branding rather than direct response -- let's face it, if you want to sell widgets and widgets alone, you'll probably just grin and bear the faults inherent in CPC because at least they measure clicks rather than sentiment or favourability.

If you're branding, however, time spent by a certain share of a certain audience is probably as good a metric as any. At least it goes beyond showing that the material was available to be viewed and doesn't get too bogged down by whether a human or a "bot" instantly clicked on a link to earn a commission.

Mark these words. CPC will probably be kept alive by direct-response ecommerce operators, but as the Web moves to embrace branding as well as direct response, and as video becomes a key component in that, we will see more moves like BMW's sponsorship of design-led articles, where time and share of audience are the metric.

They are terms that will become a common part of adland's vocabulary, just as they have been in broadcast for quite some time.




Recommend (20) Print RSS
13 comments about "Forget CPM - It's Time For Time To Be The New Metric ".
  1. Andre Parreira from WebSpectator , July 31, 2014 at 10:52 a.m.
    I couldn't agree more with Sean. WebSpectator is helping the industry to shift from impressions to time. The best is that we can now measure and buy time. The Cherry on top is that this new currency is MRC Accredited. The new accredited metric is the GTS - Guarantee Time Slot Impression and can be used with any kind of Multimedia element (Eg. video, display ads, text,...). For more info, please visit this link: http://www.webspectator.com
  2. Ed Papazian from Media Dynamics Inc , July 31, 2014 at 11:26 a.m.
    I agree, in principle, too. The problem is that so long as money is changing hands, in that the advertiser is buying access to the publisher's audience, CPM calculations will be with us. What happens next is that in many cases the buyer never lets go of the CPM finding and becomes so fixated on it that all other considerations are subordinated to it. This is exactly the same problem that plagues many of the traditional media, where a certain kind of buyer takes CPMs, which should be the starting point for evaluating a medium's value, and makes it the end point as well. Fortunately many advertisers get this and support media which offer engaging environments, less ad clutter, superior targeting capabilities, etc. even if their CPMs are higher.
  3. dorothy higgins from umww , July 31, 2014 at 3:08 p.m.
    Quite likely the time metric becomes an additional factor in developing CPMs that weight various components such as comp, coverage, time spent. There is no one metric.
  4. LLoyd Berry from Moving In Media , August 1, 2014 at 2:41 p.m.
    with a 1 second viewability standard - are we going charge by 1/1000 of second - so it still backs into a CPM? How are you going to RTB that?
  5. Craig Mcdaniel from Sweepstakes Today LLC , August 1, 2014 at 5:28 p.m.
    I agree in principle with the time element. However I have run thousands of text link ads and used the month and week method. Basically the old billboard model of pricing. Why this works in online sweepstakes promotions is with over 300,000 members I would sell out the sponsor's inventory. No joke. If the sponsor is giving away a car or a large amount of cash it would not be uncommon to see 25,000 to 50,000 sweepstake click through/entries. The big Fortune advertisers love my rate system. My problem is big advertisers are very slow to understand this. Maybe you can explain why an big agency couldn't make this work when I can?
  6. John Grono from GAP Research , August 6, 2014 at 9:59 a.m.
    Ed, just a thought. I assume the article is primarily referring to television CPMs. As you and I know (but a lot of the market has forgotten), TV ratings are an 'average minute audience estimate'. Given that the ratings data is available at the minute level (well it is here in Australia - we post-analyse on minute-by-minute data), CPM can be thought of "Cost Per Thousand People Minutes". Given that most ads fall within that time frame (and the variation within a minute is not that large), then TV basically already is what Sean is proposing, is it not? Basically, I think he is saying he wants duration based audience data ... or have I misinterpreted? Cheers.
  7. Ed Papazian from Media Dynamics Inc , August 6, 2014 at 10:24 a.m.
    John, I assumed that Sean was referring primarily to online metrics, not TV. As for TV, the basic problem , as I have noted many times, is the likelihood that so-called average minute ratings, even when they are isolated only for commercial minutes, do not tell you whether the "viewer" is watching or, if so, for how long this occurred. What you are getting, basically, is a set usage measurement, coupled with the assumption that a person who logged in as "watching" when the set was turned on or a channel was changed is "viewing every second of content that follows----unless a cessation of viewing or absence from the room is reported. Since this almost never happens, even though in real life we do leave the room and/or cease paying attention---especially during commercial breaks---- it's asking too much of the data to apply it in the manner you propose. Just my opinion, of course, but we have to take all average minute ---or second----"viewing" estimates with a large dose of salt.
  8. John Grono from GAP Research , August 6, 2014 at 10:44 a.m.
    Thanks Ed. I was in the privileged position around 10 years ago to be able to analyse a week's prime-time TV viewing (on an All People) basis at a minute-by-minute level stacked up against independently verified ad logs. I'm shooting from memory ... but the average ad-break (of around 3-4 minutes) was down around 4-5% from its surrounding minutes. First and last in break suffered the least degradation (around 2-3%) while middle of the break was just under 10% (again I stress that is from memory). The thing that amazed me is that I could see that in the panel of over one thousand people, MANY were logging out when they left the room during the ad-break. By far the most common activity was channel surfing (and I stress that was in the pre-DVR days in a five-channel FTA environment) which produced the oddity that for the 'recipient channel' of the surfing often say its audience increase! Weird eh!
  9. Ed Papazian from Media Dynamics Inc , August 6, 2014 at 6:40 p.m.
    Interesting comments, John, as it confirms what a number of agencies have found when they spent a fair amount of money to explore people-meter data on a "granular" basis, as regards commercial audiences. The idea was to garner insights about how different commercials performed and, of course, issues about ad placement within breaks, ad clutter, etc. They, too, found that the largest 'audience losses" came at the middle of the breaks while the first position and the last spot performed above par, the latter mainly due to incremental or new tune-in. In all cases, however, the absolute levels were tiny----on the order of 2-5 percentage points, typically. Otherwise, the data seemed to show that average second "viewers"----or average commercial segment "viewers" were not particularly discriminating regarding subject matter, creative execution, etc. The reason, in my opinion, was that the data did not reflect either the overt or, the more subtle changes in attentiveness that take place throughout the commercial exposure experience during a break, which, typically may feature 5-8 ads of various lengths, plus promotional messages. With 90-95% of the "audience" assumed by the system to be "watching" this is hardly surprising. The core assumption underlying the system has self-defined program viewers "watching" just about everything that appears on their TV screens, including just about all of the commercials.
  10. John Grono from GAP Research , August 6, 2014 at 7:11 p.m.
    Agreed again Ed. But we also have to consider why we have TV ratings ... to sell ads. The broadcaster spends mountains of money to create a programme that attracts (say) 20 million viewers. Given that they are pretty much glued to that content we can pretty safely say that they are 'watching' and 'engaged' (whatever that word has come to mean of late). Yet in 30 seconds we can knock a 10% or a couple of million people off that quantum because of the quality of the ad content. Let's be clear - this is not the broadcasters fault! I was also lucky enough to be involved in early stage neuro-science work on ad-breaks here Downunder. We found some initially surprising things. In a sporting event such as one-day cricket (there are 30-second breaks between overs so a natural for ad-breaks ... or between wickets falling), we found a diminution of positive activity for the ad despite cricket being one of our favourite sports and TV shows. The reason was that the brain was in sensory overdrive due to the contest, that the ad simply didn't measure up. Putting that same ad into a low-rating pretty dreary daytime soap saw the brain light up! It's easy to see why post-event ... but common logic was the inverse.
  11. Ed Papazian from Media Dynamics Inc , August 6, 2014 at 8:40 p.m.
    John, I would disagree that people, once they are defined as program "viewers" are, in effect, engaged. The evidence on this subject, including, dial switching rates, "holding power" studies, viewer attentiveness measures, program content recall studies, etc. shows wide variations in response and involvement based on program content, and demographics. Over and above that, studies of a similar nature, plus a number of observational efforts ( cameras and "spies" ) reveal even greater attrition in audience interest, attentiveness, etc. when commercial breaks appear. Of course, this varies slightly by commercial positioning and there are marked reductions when breaks extend beyond three or four messages. But, the largest variations are due to the subject matter of the commercial, the way the message is executed and how often it has been seen before. The best evidence of this is offered by thousands upon thousands of commercial recall studies. These use all sorts of methodologies and a good deal of cuing to stimulate the viewer's memory. In general, you almost never get recall higher than 60% no matter what the ad is and many spots fall into the 10-20% category. This can't be, if the actual "audience loss"during commercials is only 10%, or thereabouts. I agree that, except for excessively ad cluttered breaks, the fact that many viewers don't pay attention to a given commercial airing is mostly on the advertiser, not the network or station. But that's a fact of life and hardly surprising. Most people watch TV for program content, and tolerate the commercials. Advertisers accept this and wage their promotional campaigns over extended periods, where the cumulative buildup of awareness and buying interest develops over the course of numerous exposures to their ad messages. If it were really true that 90% of the audience watches each time a commercial appears, normative ad recall levels would be far higher than they are; worse----from the medium's ad sales perspective----advertisers could safely reduce their media budgets, since their messages are able to get their points across so easily whenever each spot runs.
  12. John Grono from GAP Research , August 6, 2014 at 8:51 p.m.
    ... or if only the ads were as good as the content.
  13. Jan Jilek from ad-net , August 12, 2014 at 7:50 p.m.
    Great discussion. Ed, John, all of research online or offline have shortcomings. E.g. TV - did he realy watch that ad, Online - did we realy reach that number of people, or we have reached that number of cookies? And there is many more examples like this. Also advertising is very complex and to many factors influence end result of campaign that is even difficult to discuss about it. But there is one crucial part. Has ad been seen and if yes did he got enough time to transfer his message to user. So time is crucial for message delivery, so it have to be a part of online ad equation. Like TV has a standard of 30'' so will display or any other ad in a future. One of my conviction is that time should be common denominator.