By Dr Grace Kite

Algorithms are powerful new agents of double-jeopardy

Share

The marketing playing field has never been level. Effective marketing is easier for big companies, it’s a well-established fact. The theory on why – double jeopardy – comes from respected academics including Ehrenberg, Sharp and Romaniuk, and the evidence is indisputable too.

But what isn’t well known is how the tilt in the pitch has been changing as technology transformed the way consumers and marketers go about their business.

The pendulum has had two swings: Starting in the 1990s, the internet made it easier for start-up businesses to make their products available, market them, and grow. But since then, the direction of travel is increasingly in the opposite direction. With ever growing advantage for the big players.

The reason is that algorithms are mediating ever more purchases and they prioritise things that are already popular. That means businesses that are already big. Smaller businesses are served up less often, pay more for their online ads, and have to devote more cash to defending their existing sales.

The two swings net out at a situation where it’s easier for small businesses to launch than ever before, but also where there’s a drag on how fast and far their growth can go. Where challengers can only challenge so much, and big business remains at least partly insulated from competition.

It’s better to be big

The chart below plots return on investment for advertising in 343 UK econometrics studies between 2015 and 2019 by size of the advertiser’s business. It’s clear, ROI is higher if you’re a bigger company.

The explanation is that the products that bigger businesses sell are more readily available, to more customers, than the products that smaller businesses sell. If Tesco puts an ad on TV, it’s much easier for someone that lives in Birmingham to respond than if the ad was for a corner shop in London.

The chart above is big picture – across lots of different categories. But more detailed data shows it’s better to be big too. The chart below shows the double jeopardy pattern – something that’s been observed in a huge number of categories worldwide.

Bigger businesses – shown here by bigger bubbles – are nearly always observed to have both more buyers (higher penetration of households), and buyers that are more loyal (higher purchase frequency).

The explanation is again about availability. Customers split their purchases more or less evenly across options that they know of and can get hold of. And the products that bigger companies sell are more mentally and physically available.

Algorithms as the new arbiters of availability

When e-commerce first became a realistic possibility in the 1990s, being available suddenly didn’t have to mean an expensive physical shop. Small businesses could put up a website and customers would be able to find them, right alongside bigger competitors.

Economists quite rightly said this would be good for consumers, who’d have much more choice, and good for the economy via healthier competition too.

Then, as time went on, consumers increasingly turned to algorithms to help them navigate all the new options. Calculations in the Google, Amazon, and Facebook machines decided more and more often what will be made available, especially to the 70-90% of people who never look at search page two.

And all the important algorithms prioritise things that are already popular. That means businesses that are already big.

Google’s organic search algorithm is the first important case. It moves businesses up the page if they have a low “time to long click”. That is, if they are clicked on often and users dwell when they land. The first part of this criteria – clicked on often – favours bigger businesses, so that it’s much more often big business at the top of the page.

And the bigger-business advantage widened when Google, Facebook, and others monetised their search pages. Because they used popularity-first algorithms to mediate how businesses would pay for positions on the page.

Paid search auctions offer ads at a lower cost per click if the algorithm expects that an ad will be clicked on a lot. Bigger businesses have that higher click rate, and, as a result they have the option to be at the top of the page for a lower price. In Google’s auctions, for example, this plays out via the “quality score”.

Many big businesses use this advantage to attack the increased availability for smaller competitors that the internet delivers. They do it by bidding on smaller competitors’ brand names as search terms.

Here again, there is an inequality. A big business bidding on a smaller competitor’s brand name will get a better “quality score” than a small one bidding on a big competitor’s name. This means that smaller brands are disproportionately under attack, and they disproportionately have to defend their availability by paying for ads against their own name.

Merchant media is the next big frontier because this same set up is increasingly being played out in retailers’ results algorithms. Amazon is a particularly important case because during the pandemic it became the biggest single engine for product searches with 53% in 2020.

The top of search on Amazon is increasingly available to purchase, and advertisers have turned out to be very willing to pay. The chart above shows, in orange, Amazon’s revenues from paid search advertising. Last year the figure was bigger than YouTube’s entire advertising revenue.

Meritocratic algorithms would’ve been better

It’s unlikely that the Google, Amazon, and Facebook engineers intended to bake double jeopardy into the way algorithms work. No doubt they were much more focussed on the challenge of finding a way to deliver millions of good recommendations with minimum computing power and wait time.

Crowd-sourcing solved that problem and produced a neat way to prioritise that made sense. If lots of people are buying it already, it must be good, right?

But it wasn’t the only possible solution, and in fact, Amazon’s organic algorithm is a good example. Absent any paid influence, it is meritocratic as well as populist. It looks at match between text entered and product, price, reviews, and in- vs out-of-stock.

More meritocratic algorithms throughout the e-commerce set up would’ve been better for small businesses with a good product. And it would’ve been better for the world economy.

Unfettered competition between small and big firms keeps an economy efficient, ensures there is progress, and enables consumers to get the best possible deal.

And that would’ve been better for all of us.

© 2024 Website &AGENCY