Typically, math tools have been used to measure advertising effectiveness (often post-mortem).  

Math can take us beyond that…by causing effectiveness to happen on each and every campaign.

Improving advertising effectiveness comes from doing better along three dimensions:

1. tactics/channels/context
2. targeting
3. creative

Math is key to systematically doing better on all three.

The first point is well known…use Marketing Mix Models, MTA, and test vs. control comparisons to know which tactics and ad placement options are working best for a given campaign. Still many marketers measure effectiveness episodically rather than systematically.  That could be improved.

The second point, targeting, has needed improvement for a long time. The main culprit is a reach-based mentality that prevents marketers from seeing the math of consumer response.  Correcting this has been an area of focus for me over the past 3 years.

The third leg used to be the domain of Mad Men but it is being transformed with math…AI and dynamic content creation. (Neural nets are the basis of generative AI and Linear algebra and calculus are important underlying tools.)

Let’s dive deeper…

Tactics/channels/context

Marketers…you should commit to ruthlessly measure the effectiveness of everything you do using best available approaches.  For walled gardens, that might be RCT testing.  For programmatic that might be MTA.  For linear TV that might be MMM.  Then, use math to put these diverse signals into one holistic equation of what works (this can be done with linear algebra and partial derivatives and/or by using Bayesian methods).

From measuring to causing…in your advertising allocation planning, favor what is measurable and proven to deliver above average results.  Alan Wurtzel, the former head of research at NBC used to say, “If I can’t measure it, I can’t sell it.” Similarly, I’m suggesting to marketers, “If you can’t measure it, maybe you shouldn’t buy it!”

I also suggest to marketers and their media agencies that you drop any notion of reach as a surrogate for effectiveness. Media planning has traditionally been based on finding lowest CPM plan that delivers the most reach against broad targets…suppose we just point the math to what we really want…outcomes.

Value-add: +30% (vs. spotty and less effective measurement)

Targeting

Working with the MMA, its partners (e.g. Transunion), and marketing members, we have developed and confirmed a science for consumer response.  Using math, we identified that a segment called “the Movable Middle” (those with a 20-80% probability of choosing your brand) should be hyper-responsive to your advertising, delivering 5 TIMES the ROAS.  In fact, case studies confirmed this expectation and shown that the multiplier can be as high as 23X.  On a dozen or so cases, the Movable Middle has ALWAYS outperformed non-Movable Middle segments. In a profession historically known to fail on 50% of ad campaigns and 80% of new product launches, I know of nothing else in marketing that delivers such assurances of success.

Value-add: By shifting 20% of your ad budget from low loyal/unresponsive consumers to Movable Middles, 30-100% improvement in ROAS is possible.

Creative

Now onto the third leg…math has now enhanced the domain previously ruled by Mad Men like Don Draper. Generative AI is being used to create the ad assets themselves.  No, we are not taking humans out of the loop but we are giving them tremendous tools to do their jobs even better. Furthermore, creative can be constructed on the fly for each programmatic ad serving instance, via matching the ad to the context and content of the placement. 

Norm de Greve, CMO of CVS says the company is already seeing good results with machine-generated content.

“We can train some of these tools on our images and our voice, and it is writing about our own brand better than our ecosystem is doing today because it’s so dispersed. I can see that going globally, and it’s going to be a serious tool.”

Value add: I have seen reports that matching creative to content can increase returns by 30-150%.

Where do we go from here?

Meta analysis of advertising effectiveness from 2011 has documented an average advertising elasticity for existing brands of about 0.12 (i.e. doubling spending drives up sales by 12%). 

Using math to enhance each of the 3 dimensions that drive advertising effectiveness, you should be able to double that. For a $100MM ad budget, that means you might be able to add $50-200MM to your sales base with math. If you aren’t planning for this, you should be.

Personally, I am fascinated by what ChatGPT, Bard (bard.google.com), and Bing can do. I use it to create a dialogue like it’s a co-worker which helps me think through a problem.  I have used it to solve differential equations and to write code in R and Python for me. I have asked it to help simplify my writing. For example, I asked ChatGPT for a 6 word story that explained the problem with broad reach media strategies and it came back with: Many reached, few responded. So inefficient.   I thought that was really good!

There is a new science of consumer response to advertising…driven by math-based principles, machine learning, and AI.  It promises to consistently deliver exceptional results.

At the MMA’s Possible mega event in Miami, LL Cool J was asked about AI and how to control it. He said, “That’s like standing under Niagara Falls and saying, “Wait, Let me explain…”

Notes: Title and graphic from AI (ChatGPT, Bing/DALL-E)