OPINION30 October 2013
Marketers are relying increasingly on automated systems, and research is ripe for automation too, says George Terhanian.
As an example, she pointed to last season’s Super Bowl, when, just after a massive power outage halted play, the Oreo brand tweeted, “Power Out? No Problem. You can still dunk in the dark.” Re-tweeted more than 10,000 times in the next hour, those three sentences became the most influential marketing event of the entire game, Steel said. Given the more than $220m advertisers spent on Super Bowl advertising that day, that’s a remarkable achievement for a free tweet.
Timing can be everything. Steel explained that marketers rely increasingly on automated systems, rather than, say, traditional RFPs (requests for proposals), formal negotiations, and insertion orders, to “show the right advertising message to the right person at the right time – no matter whether a person is surfing social media, watching television, checking their mobile phone or walking into a store”. She also noted that technology companies focussing on advertising “are racing to build the next generation of automated advertising systems”.
These systems – which replace roles filled traditionally by people – make it easy to purchase advertising at a moment’s notice, with creative content and placement optimised, in theory, to ensure that the message delivers on its promise of reaching the right person at the right time. As Steel observed: “The new technology places a huge emphasis on building more robust profiles about consumers.”
How widespread is automated, or programmatic, buying today? Within the US market, consultancy eMarketer estimates that advertisers will spend more than $3bn this year, a still-small slice of the total online display advertising pie of $17bn. But eMarketer expects the figure to nearly triple by 2017, experiencing several years of double-digit growth along the way. If you add television advertising to the mix, the number may skyrocket.
Such automation won’t be limited to ad buying. Today, most survey research agencies continue to employ a factory model, with one group of workers responsible for sourcing respondents, a second for sample selection, a third for survey programming, a fourth for analysis, and so on. Research buyers, meanwhile, spend about half of their money on projects involving tracking, satisfaction measurement, and advertisement pre-testing. They buy these services from agencies that have typically developed and marketed base versions – some with catchy names – that utilise a standard set of questions, analysis methods, and reporting templates. The agencies offer them off-the-shelf or customised. All told, this translates to an annual online survey research spend of approximately $2.5bn worldwide.
Survey-research assembly line
Much of this work is ripe for automation. It is easy to imagine a future in which tech-savvy agencies figure out a way to automate the survey-research assembly line, driving down cost and increasing speed without necessarily sacrificing quality. It is just as easy to picture a highly-skilled marketing scientist or brand expert coming up with a way to infuse these automated systems with question, analysis, and reporting modules that replicate work once done by hand.
“Automated systems conducting survey research can be, in some sense, dumb. They rarely offer advice or guidance, nor do they usually include the checks necessary to prevent common mistakes. But it would be naïve to assume that they will forever remain dumb”
In fact, it has already begun to happen. The people involved in this process would probably even tell you that they are “racing to build the next generation of automated survey, reporting, and analysis systems”. Already such systems enable users to choose the exact people they want to interview, ask the questions they want to ask, pay via credit card, and view the results in real-time. Some systems are even more sophisticated. For instance, they include high-end features such as push-button, state-of-the-art sampling and weighting methodologies, thereby reducing the likelihood that – in the case of a tracking study – a brand will mistake changes in sample composition for changes in key measures of interest.
Brainy entrepreneurs are also developing self-serve questionnaire and analysis modules – positioning, pricing, ad pre-testing, concept testing, package testing, brand equity measurement – that combine great technology with the best thinking of survey methodologists, marketing scientists and marketers. Although the Oreo brand did not pre-test its Super Bowl tweet, it will be able to do so in the future by relying on automated research systems. In a matter of minutes, Oreo will be able to select its target population, type in or re-use previously stored questions, upload copy, click a button, view results, and make an informed decision.
Automation will take hold in the research industry, as it already has in the marketing industry – it is only a matter of time. But these are early days. And there are still downsides to relying on automated systems to conduct survey research, largely because some systems can be, in some sense, dumb. They rarely offer advice or guidance on, say, optimal survey design; nor do they usually include the checks necessary to prevent common mistakes. But it would be naïve to assume that they will forever remain dumb. In every industry, including the research industry, talented people will continue to develop technology-enabled systems that improve performance, reduce human error and lower prices. That vision, ultimately, will drive change.
George Terhanian is chief strategy and products officer at Toluna