Not All Sales are Made Equal. Some are Incremental. ??

Exactly 2 years ago, I wrote a #WednesdayWhizz "In Defense of Marketing ROI":

It started off with an ROI formula. And this started off with "?????????????????????? financial benefit gained as a result of marketing investment":

These days, incrementality is the talk of the town.

Old wine in new bottles

There is nothing new about the desire to find out which % of sales is In marketing/media induced vs. which % would have happened anyway. Actually, the benchmarks on HOW LITTLE sales is actually incremental is sobering. No matter if you are a fan of #MMM derived findings, the averages look very similar wherever you look:

The challenge

In my view, incrementality is such a hot topic these days because many marketers got a bit tipsy (some got drunk) on the promise of digital marketing which goes a bit like: "with better targeting comes perfect incrementality measurement".

That might be true for certain situations: If you are an ecommerce company advertising on a limited set of purely digital channels and you run a perfectly controlled experiment on one of these channels in which you know (from cookies or UTM) where each sale comes from -- incremental sales might be easy to calculate. In many other situations, that is just not the case.

A specific danger in this context is to equal Attribution with Incrementality, that is simply WRONG:

The demand for granularity

On a tactical/strategic altitude I am convinced that modern MMMs are doing a very solid job in measuring incremental marketing and especially incremental media impact. Please see more on this here:

However, the issue is that

Marketers (and not just the tipsy ones) and marketing controllers tend to require proof of incrementality on a level of granularity that is way more detailed than MMMs are suited for.

In fact, with the rise of MMM popularity I also see a rise of inflated expectations towards MMMs to perform on levels of granularity they are not made for: Folks would like to know the incremental effect of single creatives within a single campaign within a single week. Beside the fact that this is not an MMM domain it poses a lot of challenges with any methodology.

My esteemed colleague Prof. dr. Koen Pauwels has summarized these issues around incrementality and required granularity very well in one of his recent blogposts below - and as much as I like his piece, I recommend the comment section even more that shows a few of the debates around this vividly:

The upshot

The whole debate about incrementality is also an important discussion piece on

• Every method has its sweet spot and
• How to integrate experiments with modeling,

where I believe many companies are still finding their own optimum on.