Analytic Partner’s Paul Sinkinson on reality returning to marketing
Paul Sinkinson’s decade-long tenure at research company Analytic Partners has been underpinned by a clear insight: analysis for its own sake is worthless.
Sinkinson, Analytic Partners Australia and Asia managing director, says that unless the work changes decisions, and those decisions go on to help a business, it is unjustifiable.
“The prettiest dashboard in the world doesn’t help anybody grow, does it?” he told Mumbrella.
Paul Sinkinson
This utilitarian idea lies behind the company’s lack of a sales department.
“AP is a little bit different. We don’t have sales teams, client service teams, modeling teams or anything like that. Everybody’s on the tools.
“So it’s literally dealing with clients, getting data into models, running models, supporting clients with simulations, all that sort of stuff. It was the reason why the company was founded.”
Nancy Smith began Analytic Partners in New York in 2000. Smith had held senior marketing analytics roles at Clairol and Ipsos ASI.
“She figured, better if the person who’s going to build the model, who’s going to figure out what data to put into the model, is aware of the context and the strategy, and who knows what the client’s doing, then that model’s gonna be built better.
“And when somebody’s asking why am I seeing this result? If you’re asking the person responsible for that, you get a much better answer.
“We have people with a variety of skills in the business. But everybody’s a problem solver. Everybody’s deep into econometrics.
“When you are the person who’s gonna have to deliver the work, you are a little bit less salesy and a little bit more realistic with the promises that you make of what you can do. Because you’re putting your own head in a noose.”
Sinkinson says the team models all aspects of a business, not just those related to marketing.
“If you model just marketing – the decisions that marketers make – the reality is you’re modelling somewhere between 5% and 20% of the volume drivers of a business,” he said.
Instead, Analytic Partners expands the scope of traditional marketing mix models to what it calls “commercial mix modelling”: using different signals like economic conditions, product changes, legislation, pricing, and distribution.
“You can also avoid that classic problem that happens with marketing measurement, where marketing come back and go, ‘Hey, we are responsible for that big sales uplift’. The sales director goes, ‘Bollocks, it was us. We did that with our sales promo’. The product team go, ‘No, it was our new formulation of product’.”
The different approach can change how teams operate.
“If you’ve got the full gamut [of signals], you can get the whole business behind it … you can also see the synergy between things.”
Sinkinson said he had seen the remit of CMOs shrinking, but that there was reason for optimism.
“It’s clearly a shame. It’s suboptimal for businesses. You see it in some places, but there are strong CMOs in other places taking the mantle back.”
The data-driven approach is what underpins some of AP’s more counterintuitive findings — including the idea that most ad campaigns are pulled too early. “We looked at 50-odd thousand pieces of creative … we found 14 of them had worn out, the other 51,218 hadn’t.”
“Worn out” was defined in this case as the new campaign at least matching the old creative and covering the cost of its own creation.
Sinkinson says the implication is simple: fewer campaigns, better creative.
“If you think that your campaign has worn out, the likelihood is it has not.”
When it comes to AI, Sinkinson is clear-eyed about the risks and limitations.
“The big problem with AI is is not a technical challenge. It’s a legal challenge,” he said. “Anybody who’s using ChatGPT, anybody who’s using off the shelf AI, you are putting that client’s data to the AI engine and then getting it sent back to you, which is a data breach.”
AP avoided this problem by building AI in a “containerised” way.
The other problem with AI was the quality of the initial data.
“The reality is our data’s terrible in the marketing industry. It’s been input by somebody at an agency who’s 20 years old, who got into marketing because they’re creative and they wanna be creative and make things, and somebody’s put them in front of a spreadsheet and given them no training and no support.”
“If you dump that data into an AI engine and think it’s all gonna go well, good luck.”
For all the technology, he insists the core hasn’t changed.
“There’s a bit of reality coming back into marketing … you’re seeing the upskill of marketing leaders, the breadth of marketing leaders … it makes me feel good about the future.”
Have your say