Just when you thought the Christmas marketing emails have passed their sell-by date, along with last year’s turkeys, allow us to remind you once more of the festive season.

As you’ll know, if you’ve seen any of our posts on Econometrics (or read our fabulous ebook with your post-turkey port) econometricians do it backwards. That is to say, we’re always looking over our shoulders at what has just been, to help us work out what just might possibly be.

So, that’s our excuse for talking about Christmas in January.

Christmas – not such a wonderful time of the year for econometricians

Christmas always poses particular problems for us. After all, people don’t act normally in the run-up to, and over, Christmas.

As it approaches they go mad buying all sorts of stuff they don’t buy the rest of the year, and in staggering quantities. Come Christmas day itself, nobody buys anything. Until 3 pm, when they get on their phones under the pretext of messaging Auntie Nel, but really to carry on shopping.

Recent years, of course, have brought the much-hyped, completely artificial sales occasions ‘Black Friday’ and ‘Cyber Monday’, although this year Covid rather put a dampener on the punch-ups in the queues for cheap tellies.

And it’s the other C-word, also (alas) not entirely in the past which brings me to the point of this blog. As if all of the usual weirdness at Christmas wasn’t enough to give us a headache, we have the added complexity of the well-known coronavirus to contend with. Our clients are asking the highly relevant questions of:

  • Just what DID our first Covid Christmas do to sales?


  • Did it play havoc with our big Christmas ad campaign?

If you’re asking these questions, there are a few things you can/should/must do to get to grips with the recently departed yule.

1. Get some econometrics

Yes, yes, I know we always say this and I know we can hardly be expected to be neutral on the subject, but you need a model. Seriously.

How else can you possibly hope to tease apart everything that happened and work out what worked and what didn’t?

I mean, if you had a blinding one, I suppose you could just let your media take the credit, but –what happens when you run the same thing again in 3 months’ time and don’t get the same results?

Similarly, if your industry was suffering over Christmas, how are you going to know just how much worse it WOULD have been without your media if all you’re doing is looking at total sales data?

Get some econometrics. We’ll keep saying it.

2. Watch your base

Christmas usually adds a chunk to (or takes a chunk out of) base sales. That is, the underlying sales you would have achieved in the absence of media and marketing activity. Christmas is one of the reasons that econometricians like 2-3 years’ worth of historic data.

After the model sees a few Christmases and gets a feel for what an average one look likes, it’s better able to attribute the right number of sales to it. And, all the more so this year, our first Covid Christmas. As we build our crop of January models, we’ll be looking for evidence of a change to the shape of Christmas thanks to Covid-19.

This will help our clients answer these questions:

  • Did it boost your Christmas peak, or dampen it down?
  • Even if your Christmas overall was about the same as it always was, was the profile different this year?
  • Did customers buy earlier or later?

Incidentally, there’s a common misconception about econometric models that they must be up and running during the period you are modelling. Not true. So, if you’ve read our advice under Point 1 above and are lining up to do your first econometric project, fear not. You can start your models in January, looking back at Christmas 2020, Christmas 2019 and, in fact, as many Christmases past as you are minded to.

3. Mind your multiplier

Quick recap: Media tends to work multiplicatively.

That is, instead of thinking of an ad in absolute terms as driving 50,000 extra sales, it’s usually more accurate to think of it in relative terms as driving an extra 10% sales. That’s what we mean by ‘multiplicative.’

In fact, this is what’s behind the general truism that the biggest driver of ROI is… wait for it… brand size. Clearly, a 10% uplift is worth far more for a brand that sells a lot in a year than for one that sells a little. And that is how media tends to work (there are exceptions).

It follows that anything, Covid included, that boosted your Christmas sales, may also have boosted your media ROIs. And vice-versa.

So, when modelling, remember to look out for evidence of a change to your media ROIs that might just have been caused by Covid. Isolate everything else first – your media buy, your creative, a typical Christmas – and then look for what’s left. Of course, it will call for some interpretation – it won’t come to you gift wrapped in ribbon marked ‘Covid’ – but it’s worth some careful investigation to see what sort of a handle you can get on it.

Bah. Humbug. But at least it’s not boring

No-one goes into marketing econometrics to have a boring career and Covid-19 has certainly made that true. So, if you’re facing these uncertainties around the impact of your first Covid Christmas and media spend, rest assured – there is a way to accurately find out the truth.

Philip Gaudoin

Sign Up