Five years ago, I noticed something that most marketers had already stopped looking at.
Every consumer category I worked on — travel, banking, cameras, opticians — had a thick layer of dashboards sitting on top of it. Share of voice. Keyword rankings. Sentiment trends. Competitor benchmarks. Teams stared at them, updated them, reported on them in weekly meetings.
Not one of those dashboards ever told anyone what to do.
That's not a dig at dashboards. They measure things. Measurement is useful. But measurement and structure are different problems, and the tools built to solve the first were being asked to solve the second. Predictably, they failed.
The question that kept nagging at me wasn't how do I measure this market better. It was: is there a shape to this market I'm missing?
The observation nobody pays attention to
Here's the idea that eventually became Theia. It is deceptively boring.
Search engines have already segmented every market.
Google, Amazon, YouTube, TikTok — they all spend billions of pounds a year doing one thing: matching queries to results. A person types best retinol for beginners into Google. Google, at that instant, retrieves, ranks, and returns the 10 URLs it believes most likely to answer the query. Multiply that by eight billion daily searches. Multiply that by two decades.
That matching is market segmentation. Running continuously. At planetary scale. Paid for by the biggest companies on earth.
Every keyword cluster — retinol serum, vitamin c serum, niacinamide the ordinary, best drugstore retinol — is a pocket of demand with a specific intent. Every domain or ASIN that shows up for those queries is a consumer voting, in real time, for a solution. The structure of the market is already there. It's readable. It's refreshed every hour.
We just hadn't built the tools to read it properly.
Existing keyword research tools, the ones that most marketers still pay for, have three problems. They're built around a single query at a time (volume for one keyword, rank for another). They're built around a single domain (your rank, your competitor's rank, a crude share-of-voice number). And they struggle with the messiness of language — "retinoids" and "retinol serum" and "best retinol for 30s" get treated as separate things when they are obviously the same market. The tools ask you to supply the structure; they don't give you one.
So they stay tactical. And strategy, which is what CMOs actually need, stays out of reach.
What the alternative looks like
Imagine instead that you could take a consumer category — let's say the US skincare actives market — and produce, in a week, the following artefact:
- Every relevant product across Amazon, DTC brand sites, and Sephora/Ulta — clustered into six to ten market segments that emerge from the data, not from a retailer's category tree. Retinol serums in one cluster. Vitamin C in another. Budget exfoliants separated from clinical peels.
- For each segment, the five to ten keywords that define it — measured not by volume alone, but by how concentrated their traffic is in one segment versus spread across all of them. (We call this distinctiveness, and we measure it with the Herfindahl-Hirschman Index, which economists invented to measure industry concentration. It works just as well on keyword clusters.)
- For each brand, its share of generic traffic versus share of brand traffic, and which converts into sales more efficiently. (This is where the interesting stuff hides. Budget brands like The Ordinary tend to convert generic traffic several times better than premium brands — because their winning products get validated individually on YouTube, rather than resting on brand halo.)
- For each product, the top twenty properties customers actually talk about — not the features the brand puts on packaging, but the things reviewers, Reddit commenters, and YouTube creators spontaneously bring up. Sentiment score for each property. How that sentiment has moved over the last quarter.
- For each important query, the AI Overview answer in ChatGPT, Gemini, Perplexity — which brands get cited, which get ignored, which are miscited.
Not a dashboard. A structured graph of the market, queryable from any angle, refreshed continuously. And — this is the part that turns structure into action — a set of agents that reads the graph and produces briefs: a category strategy, a product positioning analysis, a content calendar, an Amazon listing rewrite.
That is what Theia does. It's what we have been quietly building for five years, first as a keyword clustering tool I shipped in 2021, now as a full-stack intelligence platform running on four data pillars across twelve countries in our data library.
Five years of the method in public
I don't write this piece because I had the idea yesterday. I write it because we have five years of evidence that the approach works, published across four markets and three continents of clients.
In 2021, the first public case was the UK opticians market. Five clusters emerged: high-street opticians, sunglasses, online glasses, medical eye conditions, designer frames. Specsavers dominated overall visibility but underperformed in the highest-value category, designer. That piece still ranks on Google today. The method worked on a market of roughly a hundred brands.
In 2022, Econsultancy published the method applied to UK financial services — a market of hundreds of players, heavily regulated, with famously complicated brand architectures. Six clusters emerged: retail banking, challenger banks, insurance comparison, mortgages, savings products, investment platforms. The cross-vertical proof.
In 2024, the UK gift market — seven clusters from personalised gifts to experiences to flowers to hampers. The method handled a seasonal, fragmented, long-tail market.
In 2025, Best at Travel — mapping the UK luxury Caribbean travel market. The client named three competitors they worried about. The graph showed them the actual competitive set, which included three sites they had never heard of.
In 2026, Bose Germany. 38,127 customer voices from web articles, Reddit, and YouTube. 3,796 competing products. Eleven market segments. One headline finding: Bose converts generic search traffic into sales ten times more efficiently than Soundcore, but captures thirteen times less of it. A €1.8M growth opportunity, invisible to dashboards, uncovered in ten days.
And Canon — fourteen thousand ASINs, four countries, eleven imaging segments, refreshed monthly. The enterprise scale proof.
One method. Six verticals. Consumer electronics, travel, financial services, retail, healthcare, gifts — and now, beauty. A study of the US skincare actives market is in build as I write this. Coming to theia-digital.co in two weeks.
The four pillars (and why most platforms cover one)
If the graph is the structure, the pillars are the data that flows into it. Theia is built around four of them, which correspond to the stages of the customer journey every consumer brand cares about.
Demand — what the market wants. Search volume per query, trend curves, intent clusters, the queries AI Overview answers. This is the "upstream" signal. Every strategy should start here.
Visibility — where you show up. Rankings on Google and Amazon, which domains AI assistants cite, each brand's share of traffic in each keyword cluster. Rankings are not share of voice; they are share of attention, which is not the same thing.
Sales — how attention converts. Unit sales, revenue, market share, price. First-party if the brand shares it; third-party estimates (JungleScout, Stackline, Keepa) when they don't. Without this pillar, perception work is philosophy. With it, perception becomes predictive.
Perception — what the market thinks. Product reviews, Reddit discussions, YouTube creator coverage, editorial reviews, AI Overview summaries. Harmonised across sources so that "autofocus" in a German review and "autofocus" in an English YouTube video are recognised as the same concept, and "battery life" and "autonomie de la batterie" collapse into one canonical property.
Most platforms cover one pillar. Brandwatch does perception. SparkToro does audience-led demand. Similarweb does visibility. Evertune does a narrow slice of perception — AI visibility. Each does its slice well. None of them builds the graph that connects them.
That connection is where strategy lives. A brand that rises in AI citations but loses sales is misreading the market. A brand with declining search volume that's winning sales per click is in a valuable niche, not a dying one. You can't see either of those without the full graph.
The honest part
We are not a dashboard. We don't want to be a dashboard. There are good dashboards and you probably already have one. Keep it.
We are not a research agency either. Agencies deliver a very expensive quarterly PDF and disappear. We deliver a graph that refreshes continuously and agents that generate the briefing whenever you need it.
We are not a replacement for strategic thinking. Theia gives you the structure and the evidence. Your team still has to decide what to do with it. We try to make that decision obvious, but obvious isn't automatic.
And we're not for everyone. If your category has fewer than fifty products and no search volume signal, Theia isn't the right tool. Go to a research agency. If you need a single-brand social listening stream, use Brandwatch. If you care only about AI brand visibility, Evertune is narrower and cheaper.
Theia is for consumer brands operating in crowded, complex categories with commercial pressure, where the question what is actually going on in my market keeps recurring and never quite gets answered.
What's next
Over the next sixteen weeks, I'm going to publish — on this Substack, on theia-digital.co, and on LinkedIn — four flagship pieces. One new build, three refreshes:
- The US skincare actives market (new, coming in two weeks) — the beauty study
- The UK gift market, five years on (refresh of 2024) — how the gift economy moved
- The UK financial services landscape in 2026 (refresh of my 2022 Econsultancy piece) — challengers, open banking, AI advisors
- The UK opticians market, revisited (refresh of 2021) — the first Theia case, five years later
Alongside these, a glossary of the terms we think the industry needs a shared vocabulary for — intent distinctiveness, generic-to-brand conversion ratio, perception trajectory, AI answer share. Forty terms. Properly defined. Open for critique.
And a demo chatbot at theia-digital.co/demo, where you can type your own category — wireless earbuds, electric SUVs, family holidays Italy — and get a structured view of your market in ninety seconds. Free for the first view. No email.
Why this Substack exists
I have spent a lot of my career in rooms where senior marketing people — very clever, very experienced — were asked to make decisions with incomplete information and the wrong kind of evidence. A lot of instinct. A lot of qualitative colour. Not enough structure.
Theia is an attempt to make the structure as easy to see as the instinct. Or at least to make them argue on equal terms.
I will publish a piece every two weeks. Always grounded in real data from a real market. Sometimes client work, more often published research. The core argument will not change much — the market is a graph, the structure is already there, the tools to see it have caught up. But the specifics, the data, the examples, and the occasional firm opinion about what the industry has got wrong — those will keep coming.
If the idea sounds right to you, subscribe. If it sounds wrong, write to me — pascal@theia-digital.co — and tell me why. I read everything.
Welcome to Theia.
Theia Digital builds structured market intelligence for consumer brands. Clients include Canon, Bose, and Best at Travel. Published research covers UK skincare, UK gift market, UK financial services, and UK opticians. Try it on your market at theia-digital.co/demo. Pascal Moyon founded Theia in 2021 after fifteen years in digital strategy at Lastminute.com, Hertz, HSBC, GoCompare, and Canon.