Ad tech is awash in acronyms – PII, DMP, CPM, CPV – but that doesn’t mean there isn’t room for more. “MTO” stands for mean time to optimization and it’s becoming pretty important. As a metric it describes the time between when an online campaign begins and when it has started to perform at its most successful level. The greater the MTO, the greater the cost in terms of wasted impressions, wasted ad dollars and wasted engagements.
Big data and programmatic systems show the potential to improve MTO but there is still much to be done.
Before going into how MTO has evolved over time, it is worthwhile to look at the three elements that contribute to campaign optimization:
- Audience – Audience design and targeting has become one of the central themes of digital advertising. Audience targeting and programmatic methodologies are adding incremental transparency and substance to the audience definition process. Designing the optimal audience today is both an art and a science, but the pallet of tools available today is rarely able to deliver the masterpiece your client is counting on. With ad blocking, duplicate cookies, and a lack of mobile device support, cookies often only represent 20 to 30 percent of a potential target audience. Optimization is critical. But when you only have a quarter of an audience to start with, optimization can be more for show than for dough.
- Media – Optimization of the media mix has been around even longer than audience optimization. It’s logical to conclude that targeting the right person means being able to reach them on the right site(s). A good ad planner can certainly be trusted to determine the top 5 to 10 sites but after that, let’s face it, it’d all pretty subjective and all we’re doing is guessing. Perfecting the media mix in a programmatic universe of 100,000+ sites per category has to become an empirical and automated process.
- The Analytic Toolbox – They say you can do it right, on time or cheap but you can only choose two. Every DSP today has bidder optimization. They often work very well and are quite good at meeting campaign impression goals at something less than the maximum price. How much less is difficult to know. Optimization schemes today only start when the first ad is served and while they learn and can improve pricing and/or results over time, there is a “mean time to optimization” that comes with a cost.
In the distant past, advertising was a two dimensional process. If you were advertising in a newspaper the audience was the audience and there wasn’t much you could do about it. Sure, you could choose a section but there wasn’t a whole lot of fine-tuning that could be done. With radio and TV things got a little better but in reality the options for reaching specific demographics and metrics were based on channel viewership and location the size of a multi-state DMA.
The arrival of digital brought us the promise of third dimension – at least in theory. Marketers are supposed to be able to do a much better job of designing their audiences based on past behavior and some inferred demographic attributes. The tools are still pretty crude, however. Cookies, still the dominant method for building an audience, are part of a loose onboarding process that often trades off accuracy to create reach. Mobile, private browsing, do not track and the fact that a single individual might be represented by as many as a dozen cookies leads to inconsistent accuracy and a lack of unique user reach. In the end this makes it tough to achieve the promise of audience optimization and is frustrating many agencies and brands.
When it comes to media optimization the picture isn’t much better. RTB exchanges have literally made most site inventory available all of the time in 10 ms or less. Site buys are becoming more rare and expensive. To be able the handle the volume and nuance involved in selecting from hundreds of thousands site pages each second means that subjective media optimization is no longer an option.
Today there’s plenty of talk about “real-time” performance optimization. This sounds very attractive and very fast. But it also requires that a learning process – and that learning process means “real-time” is not fast enough. Optimization that starts once the first impression has appeared suggests a campaign that is far from its optimal level and is learning “on the job.” This equals cost.
As an industry, we’ve made great strides in bringing down mean time to optimization but the next step is the one that counts. Given the volume of data and improvements in analytic automation, MTO can be reduced dramatically. The laws of physics tell us that MTO can never be eliminated but, an automated analytics process that is able to measure and report against the three-dimensional interactions between audience and media, BEFORE a campaign ever begins, will reduce MTO and improve results significantly. That is a standard we should strive for.
Wired article: http://insights.wired.com/profiles/blogs/understanding-mto-reducing-campaign-mean-time-to-optimization-and?xg_source=msg_appr_blogpost#axzz3JM2OoOsC
Ray Kingman : October 14, 2014