AI is breaking your brand measurement model
We need new models for content that serves humans and machines
The tidy carriages of the user journey – running along steady tracks from awareness to consideration and conversion – exploded into a thousand social and search pieces years ago. But with AI search, measuring your marketing effectiveness is like nailing hot jelly to a wall.
As more digital knowledge is neatly summarised and packaged by AI chatbots, how your brand shows up in AI queries needs a radical rethink.
The great content heist
Many content marketers are losing their mittens as organic search traffic falls off a cliff. Google's AI overviews and Perplexity, an AI-powered search tool, blur the lines between search and AI; many now use chatbots and search interchangeably.
Gartner predicts organic traffic will halve by 2028. The sands are shifting: AI chatbot traffic surged 81% year-on-year to March.
Let's not fear the reaper just yet: the baseline remains low. Chatbots account for just 3% of search engine visits, and people are 34 times more likely to use traditional search.
Although the baseline is low, it's growing rapidly. AI traffic referrals to US retailers jumped 1,300% during the 2024 holiday season. The demographic split is telling: consumers using AI to shop are typically younger, wealthier and more educated.
For this new "AI-first" generation, AI may be preferable. Broder and McAfee's research shows a cognitive load for search users who swoop like magpies to madly switch between search results pages. AI-first audiences will find this behaviour quaint but archaic.
It's like visiting the butcher, baker and candlestick maker on the local high street. Or sodding that and getting a supermarket delivery direct to your house.
Winning at AI search will be a twisted and bumpy ride. As models develop longer memories, they'll craft responses based on individuals’ interests and tastes, spinning up billions of different answers for the same query.
Ben Steele waxes cynical about Google's recent I/O Conference, warning content producers should consider ditching the Google game entirely and focus on gated communities (like this one).
"Thanks for all the content. We've stolen it and we're serving it to users dynamically. We don't need you anymore. Oh, but you should keep creating."
Ben Steele, Content Marketing Institute
The existential threat to marketers is real. Around half of search queries worldwide show Google AI overviews before top-ranked organic pages, leading to Google Zero. It does what it says on the tin.
As search stalwart Neil Patel says, zero clicks doesn't mean zero value, though it complicates matters. Showing up in Google AI overviews builds trust and sends less but more engaged traffic whose appetite is already whetted.
AI is flattening the creative curve
A big problem for marketers – and anyone with eyes – GenAI is flattening the creative curve. Its outputs are "50 shades of beige" by default – stripped of tone, memorable visuals and attributes that help your brand speak in ways your audience may find appealing. (Or cute, or sassy, or reassuring or relevant).
We've been here before with the web. In my noughties "webmistress" days, I standardised sites made by creative designers in Adobe PageMaker. A history and travel site used tribal mask icons to get to the next layer of pages, and in another section, a different colour scheme featuring linked icons of boats.
These quirky and fun non-standard sites were replaced with boxy, standard navigation, thanks to WordPress, et al. Today's standardised sites better serve users for accessibility and information, and help build user trust in websites.
AI chatbots are like a carton of orange juice: you can forage for fresh fruit for your juicer if you want the full experience. Such as when you want to deeply understand a topic. But if you just want to drink the juice, AI bots are ready to serve.🍊
The squeezed middle
Andrew Holland, SEO Director at JBH, believes AI has rendered the typical marketing funnel useless. Users forget brands by the time they're in-market (which is only 4.5% of the time for B2B brands, according to Prof John Dawes).
Holland proposes a new SEO funnel: serve top-of-funnel digital PR to strengthen brand 'mental availability' while focusing information delivery on bottom-funnel sales for 'physical availability'.
But the 'consideration' middle gets squeezed as chatbots will make more choices for us. Low-funnel content becomes the new normal for search marketing.
Mike King from iPull Rank pours scorn on SEO folks who think their tried-and-tested techniques will work with AI. Google and ChatGPT want to serve info juice and keep users addicted engaged in-platform for longer, just as social networks do. Serving users a page of 10 ranked outbound links is no longer in their interests.
James Cadwallader of Profound declared SEO "an antiquated function" at a recent SEO Week keynote. The new AI bots will be "answer engines." Google penalised spammy practices where site owners created content for the Google robots. The future of web content could be the opposite: well-structured content packages designed for ingestion by AI bots.
Share of Model: From clicks to citations
Changing times, mainly bleak. But some good news for web managers: it's not about producing huge volumes of articles to prove depth of expertise. Content teams need fewer smart free-form articles (like this one) and more practical pages with structured 'passages' answering specific issues in different formats like FAQ pages, tables and charts. It's not just text: videos and infographics optimised for readers and voice search get more visibility in AI search.
Digital PR must be more ambitious than own-channel thought leadership, a sad PDF languishing behind a gated webpage. Build citations by getting others to talk about your programme or product. Original research is a great way to get quoted. Certain sites carry greater weight with chatbots: Wikipedia, Reddit, Quora, industry publications and high-authority blogs.
Forget the old SEO techniques of building backlinks from authority sites; citations are the new clicks.
Now we need to measure it. There's a new acronym in town: brace yourself.
"Share of Model" (SOM) measures AI recommendation patterns. It's becoming as important as Share of Voice (public conversations and media coverage) and Share of Search (intent through personal searches).
Digital agency Jellyfish defines Share of Model as how often, how prominently and how favourably your brand appears in AI responses compared to your competitors.
The key difference with AI compared to search is that it's binary: your brand either features in answers or it doesn't. There's no below-the-fold or page 2 for AI chat responses. Be there, or become invisible.
Solutions trump storytelling
Jellyfish's fascinating analysis of US automotive brands shows how the game is changing. The researchers categorised brands into four quadrants: "Cyborgs" like BMW and Tesla (current consumer rage aside) succeed in traditional brand metrics and AI recommendations. "AI Pioneers" like Rivian punch above their weight in AI visibility despite lower brand awareness by delivering solution-based information.
The stragglers are "High-Street Heroes" like Jaguar – established brands, but with poor AI representation. Jaguar's recent controversial ad campaign focused on aspirational messaging around fashion and 'vibes', but this doesn't translate into information AI bots can parse. "Emergent" brands like Polestar (who? exactly) lag in both brand and AI visibility. The methodology is brilliant, segment names are terrible. Not all laggard brands are heroes. Even just for one day.
The research shows AI tools prioritise utility over aspiration. AI chatbots want to connect problems with specific solutions. Brands need to define "semantic niches" –category positions AI can understand and recommend, shifting from emotional positioning to problem-solving.
Instead of saying "we sell deluxe coffee machines for your chic lifestyle," opt for "our dual-boiler system maintains consistent temperature for espresso extraction."
This could make brand web content more dull, but ultimately more useful.
A less Ordinary success story
Science-backed skincare brand The Ordinary is an *actual* hero of Share of Model in practice. Rather than traditional lifestyle beauty marketing, it built its brand around transparency and education. Product pages read like science manuals, explaining active ingredients and concentrations, connecting each product to specific skin concerns.
Even with advertising-resistant consumers like myself, they've gained trust and cut-through (rather like their stinging PHA facial peel). It works: The Ordinary often ranks in AI queries above beauty brands with significantly more clout.
AI is morphing product discovery
The shift to AI information builds on Google's E-E-A-T evolution, where search rankings reward Experience, Expertise, Authoritativeness and Trustworthiness over keyword stuffing. But this time it's different. Where Google sought indicators of authority, AI evaluates information for utility, making content quality and relevance more important.
The transition from search to AI conversations will shape how brands think about discovery and recommendation. Winners will adapt their content strategy to serve both humans and machines, as online content sifts from persuasion to precision.
Brands optimising for Share of Model will be well-positioned for success as more audience opt for AI-first search. It won't be a quick or easy win (what content marketing ever is). AI models are becoming multimodal; Google advises enriching your content strategy with different formats. That's easy for the world's most valuable tech brand to say, harder for a two-person boutique team in Bognor Regis to do.
Better-resourced brands *should* be able to move faster. If you're starting out or revamping your web content, David can leap over Goliath. Plan for structured content that the bots will love (and people may tolerate), matched with weightier expert-led content to build customer trust (and maybe even love).💓
How are you planning and measuring AI discovery in marketing? This post is free to access. Share with someone who may find it useful and add your thoughts. 💭
🔁 Until next time, keep it curious
Susi O'Neill
EVA trust in tech www.evadigitaltrust.com
People-first tech communications 🤝 Tech content strategy 🎙️ Tech talks and inspiration This season’s keynotes 🤙Need this? Get in touch