Political parties, campaigns and brands can’t get an accurate and cost-effective understanding of opinion in small geographic areas, like the constituencies of lawmakers. This is a big problem in political campaigning. And all political campaigning now has a huge online element, as we know. We also know political turbulence is one of the defining themes of our age.
But one thing is clear: All the players want faster, cheaper, more accurate and a more granular understanding of consumers and voters. In the age of AI, survey predictions are influenced as much as so many other machine-learning technology products.
Focaldata is a U.K. startup that thinks it has some of the answers to these quandaries. Their integrated consumer analytics and survey workflow application claims to give customers a more accurate and granular picture of consumers than traditional polling using machine learning. At the same time, they say their workflow software cuts down on the cost and time that market research takes.
The idea is that they employ a new machine learning-based technique (MRP) to generate survey “results.” This new methodology can use more information (such as old survey data or public statistics) than conventional methods, which lets them get accurate predictions in small geographic areas from the same sample sizes.
Founder Justin Ibbett had done MRP manually on his laptop a few times for some existing market research firms and realized how fiddly it was. “I felt a dedicated software application would reduce the complexity whilst making the results more accessible and useful — our early incarnations just delivered a spreadsheet!” he told me.
Much of Focaldata’s business has been in politics. They have worked with the pro-Remain group Best for Britain and the anti-Racism charity Hope not Hate on combating Far Right sentiment. However, most demand is now from large brand owners, such as ABInBev, a recent client.
They now have more than 10 paying clients, including big brands like M&C Saatchi.
Competitors include YouGov, Survation, Dalia Research (a Balderton-backed company) and standard market research agencies like Kantar and Ipsos Mori.
But against traditional agencies, Ibbett says their ML-based data processing engine sets them apart, allowing them to go very granular and get more accurate over time.
The market research market is £5 billion in the U.K. alone (PwC report, 2016) and global market research is a $40 billion market.
The startup has raised a £1.1 million seed round from notable U.K. angels, including Alex Chesterman, founder of Zoopla and Martin Bolland, founder of Alchemy Partners. Previously they raised a small pre-seed round from three other angels, including Xen Lategan (backer of Magic Pony and ex-Google, former CTO of News International).
CTO and co-founder Calvin Dudek was at Google for five years as a product manager, and ran Data Science Innovation at the DWP. Chief Data Scientist Takao Noguchi is a cognitive scientist.