“What impact will the vanishing cookie have on econometrics?”
In some ways, this is an article that doesn’t need writing, for the short answer to the question, “What impact will the vanishing cookie have on econometrics?” is “nil.” So, as much as we might personally lament the passing of those little text files that clutter up our hard-drives, the impact on econometrics should be negligible. Feel free to stop there.
However, if you want to know more, carry on reading, for there is a bit more to this topic that’s worth considering. It’s helpful to look at the Great Cookie Massacre through two lenses – the implications for measurement and the implications for marketing effectiveness.
Implications for Measurement (including Econometrics)
First and foremost, self-evidently any form of measurement that relies on analysing cookie-based data is going to struggle in the absence of the aforementioned cookie. It has the potential to seriously banjax any form of direct tracking which relies on recording people’s browsing habits on the web and associating that information with the individuals concerned. Good for privacy, perhaps, but bad for tracking. This stretches to digital attribution modelling, or Multi-touch Attribution (MTA) as it’s also known.
The good news for all current and would-be consumers of econometrics is that econometrics is left largely unscathed by the scythe being wielded against cookies, for the simple reason that econometrics doesn’t rely, and has never remotely, relied on cookie-level data. The whole premise of econometrics is that it is a top-down approach based on analysing correlations between variables such as marketing spend and sales and drawing inferences about the relationships between them, characterised chiefly by the ROI in the marketing-sales example.
In building an econometric model, therefore, what we start with are weekly time series of these variables – weekly sales, weekly deployment of TV (measured in TVRs or GRPs), weekly deployment of Display advertising (measured in impressions or clicks) and so on. At no point do we use customer-level data, such as that recorded in cookie files. Wherein lies both the strength and weakness of econometrics – it can analyse potential sales drivers for which customer-level tracking does not exist (in practice, broadcast media), as well as those for which it does (e.g. Facebook advertising). The downside is that, while it can tell you the average uplift in sales from doing £1 of TV (the ROI) it can’t tell you that that uplift was due to Mrs Brown in Sheffield seeing a TV ad. But then again, I know of no other technique that can.
So, in a nutshell, if you are doing, or considering, an econometric study you’ll be as fine post the cookie-cull as you were before it, whereas if you’re using cookie-dependent approaches to marketing attribution then you are royally stuffed.
Implications for Marketing Effectiveness
But there’s a secondary potential knock-on effect we should consider, which is the impact of cookie loss on marketing effectiveness. Remember, we don’t do marketing mix modelling – econometrics – as an end itself. Instead, we do it (or should be doing it) to improve our marketing deployment to maximise the benefit to our organisations. From an economist’s standpoint it’s all about making the best use we can of scarce resources, aka the marketing budget, all the more so in these times of economic headwinds, when marketing budgets can become very scarce indeed. So, as much as we all find it inherently interesting to measure anything we can, the real purpose of doing so in this context is so that we might work out what works more or less well, and adjust our marketing mix accordingly. In other words, marketing effectiveness is the name of the game.
And that’s where the absence of cookies may well have an effect that goes beyond hampering our ability (but not econometrics’ ability!) to determine what is going on in our mix. Advertisers don’t serve cookies for the fun of it, but instead to be able to target people according to their preferences as revealed by their online behaviours. Knowing that someone frequently visits the AutoTrader website gives us a vital clue as to what they are interested in and, moreover, what they might conceivably be in the market for. Knowing this, we can serve them a targeted advert for a product we think they might buy, be it a second-hand car or underglow neon lights, which might go some way to explain ads for tasteful accoutrements appearing onscreen when you’re looking for a used Citroen Saxo.
It follows, then, that the loss of targeting ability implied by the loss of cookies has the potential to impinge on media performance, by making it more difficult for us to identify individuals we think are most likely to purchase what we sell. On the flip side, we have an ongoing discussion about the best trade-off between targeting and reach. Whilst targeting seems like a good thing, the question is whether we end up overdoing the comms with our target group, whilst ignoring potential customers outside of it.
It’s also worth delving into the detail here. What Google is proposing to eliminate is third-party cookies, not first-party cookies (placed by a website itself), nor second-party cookies (placed by partner organisations). Needless to say, this will not affect advertising in any logged-in environments, such as Facebook or Amazon.
Also, as much as they undoubtedly care about our privacy, Google/Alpha is not in the habit of making it harder for advertisers to advertise, which accounts for the majority of their income. You won’t be surprised, therefore, to learn that they are in the process of launching new tools (‘Google Sandbox’) that allow advertisers to do what they did before in new privacy-compliant (well…) ways. Indeed, some commentators have expressed concern that the ultimate effect of the ban will be to concentrate even more power, and revenue, in the hands of Google. Fancy that.
So, what should I do about it?
If you’re a heavy user of third-party cookies, then you are going to have a problem come 2024, or whenever Google does decide to kill the cookie. As a starting point, why not ask your marketing managers, agency partners or media suppliers how much of your marketing budget is currently tied up in at-risk marketing activities. We’d suggest you don’t wait until 2024 to start this process.
But if you’re considering commissioning an econometrics study from a leading, independent econometric-modelling specialist, then nothing really has changed in this regard and you should carry on regardless.
On which note, if anything we’ve said here has piqued your interest and you’d love to know about econometrics and these particular leading, independent econometric-modelling specialists, then why not download our ebooks on the subject or drop us a line.
Philip Gaudoin
