PR and Marketing Pros: A case study in algorithm fail, critical thinking win

Bill Gates said in his ‘’ blog, “The age of AI has begun”. Far be it for me to correct the Microsoft founder turned billionaire philanthropist, but I think he’s got his words in the wrong order. AI has been around for a long time, from robots to big data and predictive maintenance.

What he really meant to say was AI has come of age. And with the hands of Microsoft and OpenAI steering the power of the AI rudder of the Bing search engine resurgence, we need to keep a critical eye on the role of those algorithms.

Here’s the question – where would AI be without The Algorithm?

Why AI has views and opinions

AI algorithms behind Chatbots seem to get facts wrong but even more troubling, have views and opinions about people and events. 

We’ve already seen cases with Apple and Amazon where hidden algorithmic bias in AI and ML tools led to gender discrimination in setting credit limits and recruitment. Now we have the woke algorithm where one chatbot refused to tell a joke about women, another that couldn’t find anything nice to say about Donald Trump, only said nice things about Joe Biden and one that couldn’t find anything bad to say about Jeremy Corbyn.

I’m entering controversial waters here so I will take a step sideways and look at the use of generative AI and Large Language Models in the world of PR and Marketing Pros.

I have already looked in a couple of blogs recently at what they can do well –  they are excellent at marketing speak! – and what they will never be able to replicate. Now I take a deeper dive into their marketing shortcomings – with an example of algorithm fail and a win for human ingenuity.

How two booksellers thrive while Amazon’s algorithm fails

Barnes & Noble Booksellers is a Fortune 500 American bookseller. In 2018 the company was in dire straits. Competition from Amazon forced it and rivals into retreat and led Barnes to shut more than 400 bookstores.

Fast forward just four years and the same company opened 10 new stores in 2022 and is planning to open 30 more stores in 2023 as part of huge real estate expansion. It has launched a $40 annual membership program offering discounts, free shipping and more.

Amazon on the other hand opened some real-world stores but is retrenching. It has now closed a string of shops – including two in Boston where Barnes & Noble is now expanding. The reason? The choice of Amazon titles in its bookshops was curated by a robotic computer algorithm – based on what was selling well online. Bookshops it seems have a different clientele – and what’s popular on the page doesn’t translate well for machines.

The Barnes & Noble magic formula? A ‘personalized approach’ with a curated choice of books that customers actually want. The choice varies from bookshop to bookshop, location to location (central Boston from Alabama for example).

You see readers are not robots, and algorithms are not human.

The revival is due to British bookshop pioneer James Daunt who started his first bookshop in 1990 in Marylebone High Street, central London. He masterminded the strategy that saved Waterstones from oblivion after he was recruited in 2018 by new owners Elliott Management to turnround the ailing UK bookshop chain.

Elliott Management then bought the struggling Barnes & Noble chain in 2019 – and parachuted Daunt in to import the strategy that had saved the British bookstore.

It’s not called book browsing for nothing

As at Waterstones, customers who now walk into each branch of Barnes & Noble are greeted by a display of titles chosen by that shop’s staff.

Readers are intelligent human beings who like to hear ‘You’ll really like this, or if you like that, you’ll also like this’. They do not like ‘Here’s our best-selling choice, which is based on statistics’.

Daunt took advantage of the Covid pandemic in early 2020 to give bookshop staff the opportunity to redesign their stores, literally book by book. No longer did they look like small libraries, they became places to explore and browse, bookshelves were used to create “rooms” for different subjects such as history, cookery, manga, romance and so on.

The key was giving booksellers the power to decide what did and didn’t work in their own branch rather than to comply with a major marketing drive dictated by an AI-based general algorithm at head office, as was the case at Amazon. Both used algorithms but the data was different – ‘hands-on’ insight versus universal trends. ‘Human’ versus ‘artificial’ intelligence.

Witness Harry’s book

Then there’s Mary Sheldon, a kindred spirit to Daunt and owner of the Tecolote Book Shop at Montecito, California, home to Prince Harry and Meghan Markle.

On the day that Prince Harry’s controversial headline grabbing memoir Spare officially became the fastest-selling non-fiction book in history and when it was difficult to find a bookstore not stacked with copies of the promotion, the bookstore in his adoptive hometown of Montecito in California did not move a single copy.

Owner Mary Sheldon is reported as saying a few weeks after the launch that she’d sold about 30 copies, with a few more reserved for customers who’d promised to fetch them in person. “I think most people up here think of it as a soap opera,” she remarked. There’s one lady who knew the reading tastes of her clientele better than any computer-generated algorithm.

In fact, according to Nicholas Schou writing in the Guardian newspaper, “finding a local who owns a copy of Spare is challenging – much less one who has finished it.” He quotes TC Boyle, the short-story writer and novelist who is also Montecito’s most distinguished literary luminary: “Didn’t read it, will not read it”.

My own straw poll among office colleagues, friends, and family has similarly drawn a blank. But stop the press! Yesterday, I did speak to one person who was reading a copy – but I had to go all the way to Canada to find them!

Message for marketers

Critical thinking is fundamental to understand cause from correlation. It is necessary to introduce rules to govern and amend AI algorithms based on the data they are processing. Well before machine intelligence, the Greeks had a word for it. They called it ‘nous’ – ‘common sense’.

Something that doesn’t come automatically with AI-based solutions, and doesn’t come easily to some well-honed marketing departments it would seem.

Judith Ingleton-Beer is CEO at IBA International

Leave a comment