Samuel Scott wrote an article for The Drum titled Digital attribution is dead! Les Binet tells us why marketers need econometrics in 2023 (23 February 2023).
« What is better – information that is cheap and wrong or information that is expensive and accurate? Soon, marketers may have to decide. »
« there was a 96% correlation between the number of cold sales emails that our business development team would send out and the number of Google searches for our brand name and resulting website clicks. Put simply, we inferred that people would receive an email, wonder who Faddom is, search Google for the name and visit our website.
Now here’s the issue. Google Analytics logged those visits as organic search engine traffic because they did indeed come from the search engine’s unpaid results. But in such specific cases, the cold emails – not SEO – should get the credit. The decline in organic traffic had almost surely come from the team temporarily sending fewer emails. »
« An entire new generation of ‘digital’ marketers now thinks only about what is stupidly called ‘performance advertising’ because it is deceptively simple. You put an ad on Twitter. You see how many people click to visit your website and buy. You attribute those purchases to Twitter. Easy. »
« The Fallacy of Immediacy. Marketers often assume that an effective ad will convince someone to buy or become a lead immediately. But many ad-driven purchases occur long after the advertisement appeared – and especially long after the ability to track the sale with digital attribution has disappeared. Just remember one of Binet’s classic charts from his famed work with Peter Field on The Long and Short of It. »
« The Fallacy of Last-Touch Attribution… The Fallacy of First-Touch Attribution… “The idea that one channel can be given ‘credit’ for a given lead or sale is nearly always nonsense,” Binet said. “Each sale is usually the combined result of multiple channels working together, often over a period of months or years. Instead, the question marketers need to ask is: What is the incremental effect of each channel? If I dial spend on this activity up or down, how much will my sales rise or fall?” »
“Attribution modeling overestimates the ROI from direct response communications and underestimates the ROI from brand communications,” Binet said. “If you just follow the attribution data, you end up just doing short-term stuff. You never build a brand, you don’t grow the customer base, you don’t grow the base level of demand. And it’s a recipe for disaster in business.”
Everyone today is obsessed with being so-called ’data-driven.’ But a lot of data is simply bad information. Good research, common sense and gut instinct are often more accurate.
Eric Stockton, senior vice-president of demand generation at cloud desktop provider Evolve IP, perfectly summarized the situation late last year on LinkedIn: “In an ironic twist of marketing fate, the channels that aren’t the easiest to measure are often the best contributors to pipeline and revenue … correlation of buyer behavior [is better than] direct attribution by channel.”
« Econometrics – also known as marketing mix modeling (MMM) – might be a solution. After all, the attribution-based online marketing world that many have known for the past two decades is rapidly disappearing. »
« Essentially, econometrics models take every single thing that might affect sales – from advertising campaigns to the weather to pricing to overall economic conditions – and narrows down the factors from hundreds to the dozen that are the most important.
Then, the model creates an equation that describes the relationship between those factors and sales. Marketers can then run the equation to see that if they do X, sales will change to Y. Governments use econometrics to measure things such as the effects of tax rate changes on GDP. Marketers can use it to determine the best marketing mix – because marketing, of course, covers a lot more than advertising and communications. »
« “Attribution – quickly and cheaply – will give you an answer that is precise and wrong. Econometrics – slowly, laboriously and expensively – will give you an answer that is right,” Binet said. “Attribution modeling will tell you that things that are unprofitable actually made money. It can be incredibly misleading. And it will tell you that things that really are profitable aren’t.” »
« The argument against econometrics. For econometrics, one prominent skeptic is Byron Sharp, director of the Ehrenberg-Bass Institute for Marketing Science in Australia. “The people proposing econometrics as a solution [to the problems of attribution modeling] all seem to be people who sell econometrics… and marketers are not known for their mathematical knowledge, so [they] are easily fooled.” »
« Sharp gave me a 2018 paper that he co-wrote for the International Journal of Market Research with John Dawes, Rachel Kennedy and Kesten Green. I have made it available for download in PDF format here. »
« most econometrics models need at least a couple of years of historical data on hundreds of variables. Large, established brands certainly have that. But small businesses and new high-tech startups, for example, do not. »