After applying a new bias-correcting technique to 93 published advertising studies, Joseph Korkames, PhD, has some advice for marketers: ‘Be very pessimistic of studies, be pessimistic of journals and be pessimistic of the publication process.’
Korkames used recently developed statistical methods to eliminate publication and aggregation biases in advertising meta-analysis papers — specifically, those that used econometric models, like MMM, to calculate the results. His findings, which were published in the International Journal of Research in Marketing, showed that when these data were put through this new analysis, advertising elasticity estimates fell substantially.
While previous techniques put the average effects of advertising somewhere between 0.09 and 0.47 (meaning that, if a brand spent 1% more on advertising, it could typically expect sales, consumer choice or market share to increase by between 0.09% and 0.47%), the updated range provided by Korkames’ method is between 0.003 and 0.1.
These more humble figures bring the advertising elasticities produced by econometric models closer to the lower estimates that are frequently seen when researchers try to discern the effects of advertising from controlled lab experiments.
In some respects, Korkames makes for an unusual academic because he’s not a full-time researcher or teacher. His job is to help run the family business, Famous Chili, which sells bricks of concentrated beef chili in retailers and food services across 10 or so states in the US.
Korkames’ foray into academia was not an altogether happy one, either. He tells MediaCat that he became ‘disgruntled’ with the institution, believing that some of his ideas were copied by fellow academics, and that the journals repeatedly argued against his results because they reflected badly on the authors of the studies he’d analysed, some of whom were involved in reviewing his research.
But Korkames’ scepticism of marketing research was born out of the data — namely, the funnel plots showing the distribution of results from previous meta-analyses — not his frustration with academic machinations.
‘Generally speaking,’ he says, ‘funnels should be symmetric around the true effect. So, if there’s a whole bunch of estimates that are not very precise and are very positive, but there’s no [imprecise and negative results] on the other side, then the literature is cheating.
‘It could be intentional or unintentional. It could be that [authors] want [the effects of advertising] to be high and feel like they’re more likely to get published if they’re high. It could also be that everybody’s sending in all the data, but the journals only publish the ones that are high. But for whatever reason, that bias is very large in advertising. And I’m willing to bet that it’s the case in most fields in marketing.’
Korkames’ disillusionment with marketing research is not a refutation of advertising altogether, however. While his father didn’t really advertise Famous Chili, he does, spending most of his budget on digital channels with ads that inform people about the product or remind them that it exists — Korkames has found that nostalgia tends to bring a lot of people back to the 100-year-old brand.
Rather, Korkames’ scepticism is limited to how academic papers say advertising works, and how those results are frequently misinterpreted or extrapolated beyond the relevant range. The effects he calculated in his own research, for example, cannot tell you what would happen if a brand were to double its ad spend because the data doesn’t contain that information, but all too often other papers will make that kind of leap.
A big determinant of the extent to which advertising will work for a brand, according to Korkames, is ‘the way the product or service is given or experienced.’ Movie theatre advertising, he notes, produced noticeably larger effects than the other kinds included in his data, for example. Korkames hypothesises that this is because movie trailers often act as a kind of sample of the product, rather than an ad, and because a cinema screening takes place over a fixed period of time in a localised area, which increases the informative value of the ads.
But even if advertising produced no effect whatsoever on business outcomes, there would still be reasons for brands to do it, says Korkames. ‘One is your rating on Walmart and Google. If you’re getting a bunch of bad reviews from people saying your product is too thick and dry when it’s a concentrate that you’re supposed to add water to, the informative purpose of advertising, to ensure people are using your product correctly, is sometimes more important than how it’s going to directly increase your profits.
‘And now, certain retailers expect you to advertise on their platforms and pay money to them to sell your product on their platform, and if you don’t, they might give your competitor an edge. So, a lot of time, it’s advertise or die.’