Tom Vladek from Recast shows how their media mix modeling platform accounts for holidays and promotions when measuring marketing performance. The platform lets you model different types of sales spikes separately, so you can see which marketing spend actually drove sales versus which came from a holiday bump.
Hi, I'm Tom Vladek, co founder of Recast, a media mix modeling platform where the world's most sophisticated brands measure and optimize their marketing spend. Today I want to talk to you about one of the most important and I think coolest features of Recast, which is how we model holidays and promotions. We call them spikes because they look like spikes. They're points in time where your sales can increase dramatically. And for many of our customers who have lots of promotions, it's incredibly important that we model this correctly. So I want to show you how we do that. So you're on our dashboard. And you can see on our Insights tab we have a spike summary. Now, this is demo data, and this simulated data only has three kinds of spikes. The start and end to Black Friday and a promotion. Now, you can see that these spikes are arrayed over time, and since they're not many, I'm going to zoom in on just one, which is Black Friday. This is a perfect illustration of the flexibility that Recast brings in modeling these types of variables. Now, everybody knows that on the day of a promotion or on the day of a holiday, your sales are likely to increase dramatically. But what's really important is to capture what happens on either side of that promotion correctly. A lot of our customers have some vague idea that their customers may be delaying their purchases to take advantage of a promotion, or maybe accelerating their purchases to take advantage of a promotion. And capturing the sort of pull forward or pushback effect on, on the shoulders of a holiday or sale is really key. Now you can see that in action here. So this is the Black Friday promotion and we're capturing the effect that in the run up to Black Friday, customers are actually delaying their purchases. They're piling them on to Black Friday itself. And then some of those customers have actually pulled forward their purchases. And so in the days following Black Friday, sales are depressed as well. This isn't the end to the kind of flexibility we can capture. Holidays can be wholly positive or wholly negative, or the negative effect could only be on one side, meaning there could be only a pull forward effect, but not a pushback effect or vice versa. Now, to see another example, we could zoom in on the promo. If I get rid of that. And this promotion apparently had no positive effect. It was just a discount targeted at customers that would have purchased anyway. And this shows up in revenue. So for any volume of sales, customers are just spending less than they otherwise would have. And so this effect is completely negative, although sometimes the effect can be completely positive. The flexibility in the shapes that we can model gives us this power. In the spikes that we model, we show them in two ways, grouped and individual. Grouped means aggregated or sort of the average effect. So what this line means, this Black Friday end line means that for the average end to Black Friday, Here is what we're looking at. And when we forecast to the future, this is the average effect that we will forecast for the next Black Friday. Now, we don't only do it at a group level, we also do it at an individual level, which means for this specific end to Black Friday and for this specific start to Black Friday and for this specific promotion. The data that we require for this is very simple. We just need the dates of your promotions and of the holidays that matter to your business. And with that, it will automatically get modeled. It will be shown visually like this and there will be tons of data that you can download and use as you like. Now, if promotions are important to your business and you want a media mix modeling partner that can handle them correctly, just go to getrecast.com and hit book a demo. Thank you very much.
