Why marketers must win algorithms over before they woo consumers

I had sent armies of market researchers to stalk potential customers in malls and kirana stores, organized focus groups with freebies for participation and wired consumers to eye-tracking machines. I called this measuring sentiment.

Fast-forward to today, the AI era, and the marketing cosmos looks like it has been hijacked by a mischievous sorcerer. Consumers are still human, yes. They still binge-watch serials, pretend to read nutrition labels and quarrel over Android versus the iPhone. But now, looming between the brand and the buyer, sits a new gatekeeper: the large language model (LLM).

Suddenly, it isn’t enough for humans to like your brand. You need AI algorithms to like it first. If ChatGPT, Claude or Perplexity decide your brand has the charm of a damp sponge, your sales funnel collapses before a single human even gets a whiff of your product page.

AI has become the invisible maître d’ of consumer choice, whispering recommendations, warnings and sometimes outright nonsense about your brand to millions of curious searchers. Welcome to the new era, where marketers must woo customers via LLMs and manage their sentiment.

Before the rise of AI, we had our toolbox of sentiment detectors: Surveys with questions designed to look casual but worded to avoid bias; focus groups where a few strangers debated brand colours under the gaze of a one-way mirror; social media listening tools that scraped posts for emojis, hashtags and the occasional meltdown; brand trackers that measured awareness, recall and preference in painstaking quarterly charts.

It wasn’t perfect: humans lie, forget, or say one thing while meaning another. But the object of analysis was clear: the human psyche. Marketers were playing psychoanalysts of the masses, trying to decipher intent from a storm of contradictory signals.

If someone said your burger was “fine," was that good? Bad? A prelude to defection? One learnt to read between the lines, to decode the shrug emoji and obsess over Net Promoter Scores as if they were stock prices.

Today, the game is stranger. LLMs don’t buy toothpaste, but they tell humans which one to buy. They don’t need smartphones, but they can crown one as the undisputed king while relegating another to footnote status.

This shift means marketers must ask not only ‘What do consumers think?’ but also ‘What does the algorithm think consumers should think?’ The marketer’s playbook now includes a bizarre new chapter titled ‘AI Sentiment Optimisation,’ and it reads like something out of a Black Mirror episode. Anyhow, here’s what must be done:

Track sentiment across all major AI platforms: If ChatGPT gushes about your eco-friendly sneakers but Claude mutters about “overpriced gimmicks," you’ve got a split personality problem to solve; so set up automated sentiment alerts. Because nothing ruins a Friday faster than finding out an AI agent just called your product “outdated."

Benchmark against competitors: It’s not enough to know your score. You must know if Bard thinks your rival’s battery life is “stellar" while describing yours as “adequate." Adequate is death.

Analyse context: AI doesn’t merely say “good" or “bad." It offers essays. A paragraph of praise for packaging might be followed by a damning line about customer service. One must understand the why behind that sentiment.

Take corrective action: This could mean gymnastics over search engine optimization (SEO), content partnerships or simply fixing the actual problem.

The irony is that all this effort is still in service of influencing the same old human shopper.

The woman standing in the toothpaste aisle has no idea that before she arrived, an invisible algorithmic jury had pre-filtered her options. She trusts the AI’s guidance without knowing how much lobbying, nudging and algorithm-courtship went into making sure Brand X appeared on her radar instead of Brand Y. We marketers used to say “perception is reality." Now we must add, “And AI’s perception precedes reality."

The implications are as dizzying as they are comical: Brands are now fighting to become the teacher’s pet of algorithms. Imagine a toothpaste brand sending flowers to ChatGPT on Valentine’s Day. Public relations crises can now start when an LLM “misremembers" your history. Competitive intelligence includes asking how Claude describes us versus them, as if one must eavesdrop on a cocktail party. SEO is no longer just about Google’s spiders, but about training the very personality of AI platforms to recall one’s brand fondly.

It’s as if the brand’s audience aren’t people anymore, but omniscient, occasionally cranky librarians who control the flow of all knowledge. Convince them, and you convince the world.

I live in a great marketing paradox: people still want shinier, cheaper things, but I now woo the machines that woo them. One day I’m conjuring TikTok magic for Gen-Z, the next I’m working to fix why ChatGPT called my product “clunky." We have little choice but to let algorithms chuckle at our efforts and brands. It’s to AI that we now first pitch our wares. Best to cozy up with them. Welcome to algorithm schmoozing.

The author is co-founder of the non-profit Medici Institute for Innovation. X: @MuneerMuh