New empirical research revealed Tuesday by a mix of respected industry experts made the case that marketers need to shift from a decades-old practice of “reach optimization” to new methods of “results optimization,” especially when factoring the most precious media in their mix. The research, which was presented during a special webinar from the Advertising Research Foundation and at a live event at programmatic TV developer Simulmedia’s New York headquarters, showed that when brands measure the return on their TV advertising investments based on the incremental sales they generate, they get better results.
One of the examples presented, an explicit case showing actual sales results for an ecommerce marketer participating in a campaign managed by Simulmedia, resulted in an ROI of $45.17 for every TV dollar spent. Significantly, the result was not measured by modeling, fusion, single-source panels, or theoretical research, but an actual dollar-in/dollar-out analysis using the advertiser’s TV budget and proprietary sales data.
“Linear TV advertising is now as accountable as digital,” proclaimed Simulmedia Founder and CEO Dave Morgan. As one of the pioneers of digital ad accountability -- first at 24/7 Real Media and then behavioral targeter Tacoda and AOL -- Morgan should know. And he’s built Simulmedia on that the same framework -- that if you could reorganize the way people use television to target consumers and measure the results empirically, you will make it more effective and more accountable.
The secret, Morgan implied, was not in math, but in the fact that TV advertising has powerful effects and that if used more scientifically, can work even better. The case study, as well as examples presented by industry guru Bill Harvey, founder of single-source measurement platform TRA, and CBS research chief Dave Poltrack, made the same point -- all with empirical proof, not fusion, modeling or theory. Just dollar-in/dollar out analyses.
The cases drew another important line for the ad industry, especially for television, effectively moving it from a so-called “long-term brand effects” medium to a performance medium. That’s something the TV industry has historically resisted, because performance media, like direct-response and early-stage digital platforms, historically commanded lower media costs than media that are perceived to have more powerful branding effects. The presenters at the ARF webinar seemed to be making the case that a new way of thinking about performance media is now required, one that downplays long-term branding -- for now -- but keeps it implicit.
Several attendees reacted immediately to that point -- including the 4As’ Mike Donahue and industry consultant Bob DeSena -- raising questions about the implications of that shift for branding.
“It seems to me the major trend is a direct marketing mentality coming to brands,” direct marketing vet DeSena noted, adding: “What direct marketers have always known is the two most important things -- after product -- is the target and the offer. How does that translate when you say don’t take it out of TV?”
DeSena’s question addressed the fact that all three presenters made the case that reductions in TV spending -- especially more expensive, high-rated mixes of TV inventory -- always result in lower returns. Those cases were a direct challenge to the current trend among big marketers and agencies to boost digital spending by shifting it out of TV. The cases argued it is not just antithetical, but counter-productive.
CBS’ Poltrack responded by noting that it was always very easy for direct marketers to prove the return on their investment, because they were able to correlate dollars spent vs. dollars returned via a closed-loop system. Now, because of research that companies like CBS, Nielsen, Simulmedia, TRA and others are doing, it is possible to correlate it with campaigns that historically have been viewed solely on a long-term branding basis.
Simulmedia’s Morgan added that brand advertisers will have their cake and eat it too, because the new methods will enable them to target TV more effectively in terms of results, but they will still be able to buy it on a CPM basis.Morgan asserted it will happen with TV, because its already what digital platforms have already done. “You get long-term sales effect for free,” he concluded.
While I certainly agree that ad campaign results should be measured on a "performance"basis---if possible----the question remains how many TV ad campaigns can this standard be applied to? Again, the obvious distinction between direct response and "branding" campaigns rears its head. Where a direct marketer makes his case for consumers to buy his product in a single commercial---without reference to competitive brands-----the branding advertiser is often engaged in a long term effort to position his brand relative to rivals. This requires a constantly recurring pattern of exposures---in effect, reminders to those who were convinced and a lure for still more new prospects, defectors from rival brands, etc. The results are not always reflected in instant sales responses in the marketplace. Sometimes you are doing just fine if you maintain your normal share of market. Also, many TV branding campaigns are not of the packaged goods variety. Many are "image" campaigns or national overlays for promotional spending or local advertising by dealers, franchisees, store chains, etc. In such cases, tracking results and tying them to particular TV " ad exposures"is not going to be an easy task and it may even be inappropriate. So, "yes", but only where relevant----not always.
Ed: I'm not an expert, and I'm just reporting on what experts are saying (including you), but I think the idea is to do it where you can measure it and use those empirical measures to benchmark the ones you cannot measure. Obviously, every brand has a different KPI, and this is easier to do when you have a direct causal result, like sales or a call-to-action. That's something direct response advertisers have been doing for TV since people began using television. I think the idea is to up the science, apply some of the rules learned from digital, and start proving TV's overall effectiveness -- short and long -- based on the most empirical measures you can gather, or benchmark them to. That's my read. Would love to know what you or others think. Personally, I think people often look at media through their own blinders. I think this is an attempt to show some upside for the science of TV as a performance medium beyond direct response. Use some of the tricks of the digital trade. And get away from outmoded GRP measures and temper or fine-tune reach optimization planning.
Ed, very good points. The results we showed at the ARF event we're intended to suggest that the entire ocean of all brand ad spend and sales effects can be boiled down to perfect knowledge. However, it is now indisputable that you can account for the short term and mid-term sales effects of TV ads for virtually any direct to consumer brand advertiser - retail, entertainment, restaurants, travel, financial services, telecom, e-commerce - and many manufacturer categories as well. You may not know all of the commercial communication events that contributed to the sale, but you will be very confident that "but for" the TV ad the sale would not have occurred. That is an enormous step for TV media, which never wanted or could be accounted for that way.
This is not new news. And this is an echo of the CAB research that also demonstrated increased sales to ecommerce marketers when they used TV. For media professionals this is good news and helps us to move the dialog forward. For example, if we merge this empirical approach with the move to "buying audiences" we first identify different audiences and their brand-related media journeys. We assign to each a desired response or brand-related KPI and consider the channels, content, devices, locations which are most likely to contribute to positive results. The key step is the definition of media-accountable KPIs and where each media opportunity fits strategically to deliver against them. Perhaps the success metric is increased awareness....or greater favorability among those aware....or recommendation....or sales. T/V is a marvelously effective information provider and our job as media pros is to help align its attributes to client business needs and KPIs to which we in the agency business can be held accountable. And..., we still need to define and measure the GRPs and reach translation to deliver scale and ensure the optimal audience contacts on a weekly basis.
A few more points that advertisers need to consider are stage in campaign and campaign momentum. I have had occasion to review a number of ROI-type studies and, inevitably it matters quite a bit if you are at the outset of a branding campaign ----when consumers are being exposed to a new positioning strategy and/or a new product for the first time and the message is not overly redundant--- or well into the campaign when wear out is at play. This, of course, ties into the "momentum" that the campaign has been building in terms of gaining and holding awareness, garnering word-of-mouth endorsements, getting "new" buyers to try the product, etc. The studies I have seen have had a difficult time dealing with these particular aspects. Any thoughts, guys?
Agree entirely. We do not plan TV as a one size fits all messaging opportunity. What is the role for TV and the actual message for the various audiences we are trying to reach. Are we introducing the brand to newcomers to the market (a 365 day a year proposition for most brands)? Are we trying to get the stuff off their pantry shelf so they will use it and buy more? Do we want the gatekeeper to recommend a product to the end-user? Every campaign has its own goal metrics and we need success metrics and media alignment scheduling to address each of them individually.
Dear MediaPost: The inability of your commenting system to produce paragraphs has prevented me from responding here. Please join the 21st Century (perhaps Discus? or even Facebook?).
Ed-at PD, we could do it on virtually any type of KPI. You can even set up simple perfectly matched sub-dma mini-markets to do test and control where you only change one variable at a time but what we found was clients did that for about two weeks and said let's just roll it national because the results can be staggering. You can measure just total sales, you can measure website visits if you want, for autos I have seen people look at visits to showrooms and numbers of people using the car configurer on the model's website, there are tons of things you can choose as your measure instead of the mere proxies of "brand awareness" and gobbledy gook like that.
The results are truly staggering. Our biggest new business problem is the gains didn't seem credible. The LOWEST gain in top line sales we saw for a client was 9%. The highest was literally 72%. Those were numbers counted by the client in actual sales increases. The big vs small ratings is simply a function of targetability. Truly broader based consumer goods could use bigger rated programming. But when you are targeting on over 400 variables it is unlikely that you would get truly broad. Also interesting is the anti-targeting you can choose to do or avoid. Plenty of cosmetic advertisers would buy one show in particular but in truth it was only right on age (they wanted 18-34 and the show was 31) but virtually every other variable spoke to the age is right but the audience was absolutely non-cosmetic buyers. Part of the efficiency gain is avoiding the non-buyer.
Very interesting, Jon, but not surprising about advertiser attitudes. In my experience, especially in the packaged goods category, I have seen a number of cases where ambitious ROI projects were undertaken, trying to get at issues like media weight, media mix, scheduling variations, etc. Invariably the market researchers, who are woefully underfunded at many of these companies, were under great pressure to produce usable findings, yet they didn't have the resources to gather all of the relevant data. I'm talking about ad awareness, GRP delivery by demos, reach attainment, commercial impact measures, promotional activities, brand image data--where relevant----, distribution factors, etc. and, for all of these metrics, how the competition was performing--- on a month by month basis, if possible. What they used, instead was a simple correlation , usually of quarterly sales with their own GRP weight and sometimes not even that, with dollars used as a substitute for GRPs. Not surprisingly, their ROI machinations were highly insensitive, indicating only "trace" responses to GRP weight and not much consistency in the findings. This is probably why many of your clients are in a rush to come up with solutions and why they are so surprised at sales or share of market "lifts" of 15%-25%, or more. Their own investigations rarely show anything like this.
Using non-aggregated STB data is a great thing for TV measurement. However, how do you account for the other media touch points (display, social, etc)? It seems like this approach to TV attribution is similar to view-through measurement approaches of display campaigns whereby display ad vendors are taking credit for a sale just because they could link the banner view to a sale - when in reality it was a variety of media touchpoints that led to the sale. Don't get me wrong, it's great to finally be able to dish it back to digital by using empirical data to assign credit to TV, but how do we counter critics or clients that pose this type of question to us?
Ed, you touched on the key....sometimes researchers try to look at too many variables. The key is keep it simple. One variable at a time. Strict, clean test/control. Just like if you are trying to find a cure for a disease. Then after the single variables are handled you can start looking at 2, then 3, etc together.
Steve-attribution is usually bull shit. You can due more with stb than aggregated. You can pii friendly match homes viewing (and people better than most think) people viewing to purchases, webpages, searches, etc. Plus remember the concept of strict scientific methodology...hard science like physics not soft sciences like marketing has been previously due to technical limitations. The key is to not make the mistake of confusing correlation and coincidence with causality.
Similarly, EMM has found, although somewhat rare in 15% of thousands of commercials evaluated, TV advertising that can produce double-digit sales growth for branded CPG products. In these cases, spending payout is achieved much faster than advertisers typically expect. More importantly, EMM has diagnostics at both the commercial execution and the overall campaign level to reliably fix and optimize under-performing TV advertising to produce double-digit sales growth for even mature brands.
Not surprising, David. Scanner panel research has been showing sales or share-of-market lifts for packaged goods TV commercials of anywhere between 5%-25% immediately after exposure for many decades, with a few ads delivering even higher results---usually for small brands---while a small percentage of ads ( 1-2% ) actually have the opposite effect and generate negative reactions. The initial lift, quickly dissipates, however, and must be reinforced with added exposures over time to sustain a campaign's momentum. Does your research lead to any systematic findings as to what types of creative executions and/or positioning strategies are most effective? Can you share any of these general insights with us?
When we find effective commercials, the weekly incremental penetration lift ranges between +15% to +200% with an average of +83%. We find an ad effect only when on-air when consumers are in the marketplace and no ad effect whatsoever when off-air. In other words, there is no carryover effect on sales. Indeed, Erwin Ephron referenced our work as evidence to support his Recency/Continuity/Propinquity theory in his book: "Media Planning - From Recency to Engagement."
We have found that how the message is delivered greatly determines whether a commercial successfully captures viewer attention at the time exposure. This is about how the message is delivered, not what the message should be. The commercial format and their modifiers determine viewer attention, while how the brand is introduced determines brand linkage - or who paid for the commercial. Altogether there are over 50 executional characteristics and their combinations.This body of diagnostic knowledge enables evaluation of TV copy at the level of storyboards, rough edits and finished commercials.
For example, we can assess how a Testimonial commercial will do vs. a Voice-Over. If it is a Testimonial, should it be a Celebrity vs. a Non-Celebrity. And if it is a Celebrity, should the celebrity be identified or not identified.