Catalog intelligence

For twenty years, ecommerce strategy was about channels. You picked a platform. You bought traffic on Google and Meta. You built lists on Klaviyo. You optimised conversion on landing pages. The product page sat in the middle of this stack, mostly ignored, filled with copy pasted from suppliers or written in a hurry by the founder at 2 a.m.

That stack is breaking.

ChatGPT processes 2.5 billion queries a day and routes shopping intent through an answer-first interface where the product page is summarised, compared, and often never visited at all. Google AI Mode triggers on roughly half of searches and collapses ten blue links into a single recommendation. Microsoft Copilot ships completed checkouts inside the conversation. On 24 March 2026, Shopify activated Agentic Storefronts for every eligible merchant by default — 5.6 million catalogs, instantly discoverable across ChatGPT, Google AI Mode, Gemini, and Copilot. No opt-in. No configuration.

The merchants who benefit from this are the ones whose catalogs were ready. Most weren't. Peec AI studied 43,000 enrolled product listings in March 2026 and found 41% had title issues and 34% had incomplete feed data. The distribution pipe is open. The catalogs going through it are, on the whole, not good enough.

This is why the channel stack is breaking: no amount of ad spend fixes a catalog the AI channels can't confidently recommend. No SEO campaign ranks a product page with a 12-word description and missing schema. No email list generates repeat purchase if the product page, when re-visited, looks unconvincing. Every surface that sends traffic, in the end, sends it to the catalog. The catalog is the atomic unit. If it's broken, nothing above it compounds.

We call the work of fixing this catalog intelligence.

What catalog intelligence means

Catalog intelligence is the practice of making every product in a merchant's catalog the best possible version of itself — scored, structured, enriched, and ready for every surface where commerce now happens. It is a superset of SEO. It is more rigorous than content generation. It is distinct from feed management. It operates at the atomic unit of commerce — the individual product page — and it operates continuously.

Three things make it a real category, not a rebrand of something older.

First: the atomic unit. SEO operates at the level of pages and keywords. Feed management operates at the level of syndication destinations. Content marketing operates at the level of articles and topics. Catalog intelligence operates at the level of the product — every product, all of them, measured and remediated as individual assets with their own scores, structured data, descriptions, images, and metadata. The question is not “how does my store rank for this keyword?” but “how does product #1,847 score across the signals that decide whether AI channels recommend it, and what is the highest-impact change we can make to that product today?”

Second: the agentic-commerce-era standard. Traditional SEO assessed whether a page had a meta description. Catalog intelligence asks whether the meta description is unique, benefit-led, and under 155 characters. Traditional schema tools confirmed whether Product schema was present. Catalog intelligence scores it against the 20+ contextual properties AI agents actually consume — differentiators, use-case mapping, specification depth, review signals, availability freshness — versus the legacy 5–10 properties Google's SERP engine relied on. Traditional content tools generated a product description. Catalog intelligence tests whether the title of that product matches how shoppers ask ChatGPT for it, because the ChatGPT shopping model uses 7-word natural-language query fan-outs and “The Luna” loses while “Women's A-Line Cotton Midi Dress in Cherry Red” wins.

Third: continuous, not project-based. Agencies have been doing catalog transformation work for years. They do it as a project — three to six months of engagement, a final report, and then the catalog decays. New products launch without metadata. Schema breaks silently. Content grows stale. The agency's work was valuable and also unsustainable. Catalog intelligence is an ongoing state, monitored and maintained, not a one-time deliverable.

The five phases of catalog quality

Every catalog, regardless of vertical, moves through five phases. Merchants without a catalog intelligence layer cycle unevenly through them. Merchants with one compound progress.

Audit and intelligence is the first phase — a full, honest assessment of catalog state. Per-product scores across titles, descriptions, images, alt text, SEO metadata, structured data, vendor consistency, and taxonomy. Per-collection scores on depth and internal linking. Vertical benchmarks — “your catalog scores 52, the footwear category average is 71, the top decile scores 88+.” Without this foundation, every subsequent action is guesswork.

Infrastructure is the second phase — the technical baseline. Product, Organization, LocalBusiness, BreadcrumbList, Article, and FAQPage schemas deployed correctly across the site. Unique SEO metadata for every product and collection, written programmatically to scale. Product data cleaned: standardised titles, corrected vendors, proper taxonomies, typo removal. This phase is tedious, high-volume, and invisible to shoppers — which is why it is almost always skipped, and why every catalog ignoring it loses to every catalog executing it.

Content and enrichment is the third phase — the visible work. Product descriptions rewritten to 150+ words, structured with benefits, use cases, and trust signals, unique per product rather than pasted from a supplier. Alt text generated for every image with meaningful context rather than file names. Blog articles published in topical clusters with embedded Article and FAQPage schema, linked internally to products and collections. Supporting content — sizing guides, care guides, FAQ pages — built once and referenced everywhere. This is where AI content generation earns its keep, but only if the audit and infrastructure phases precede it. Generated content on top of broken infrastructure is noise.

Continuous quality assurance is the fourth phase — the part most merchants never reach. New products launched without metadata. Schema breaking after a theme update. Score drift as the catalog expands. AI citations declining as competitors ship. Catalog intelligence is the layer that notices, alerts, and offers remediation before the merchant knows to ask.

Intelligence and attribution is the fifth phase — proof that catalog quality drives revenue. Graduated attribution backlinks tracked to referral traffic. Organic search lift measured quarter over quarter. Category benchmark position over time. AI citation rate by channel. Without this, catalog quality remains a cost centre. With it, it becomes the budget line nobody questions.

Why this matters now, specifically

Three conditions converged in early 2026 that made catalog intelligence an urgent category, not an academic one.

AI collapsed the cost of the content production layer. What took an agency team three months now takes a well-engineered pipeline five hours. The bottleneck shifted from labour to prompt quality, validation discipline, and methodology. Catalogs that previously required £15k–£40k of agency spend now require software — but only if the software is built around quality standards, not just generation volume.

AI became the new distribution layer. Merchants whose catalogs are semantically thin or structurally absent disappear from Google AI Overviews, ChatGPT, Perplexity, Gemini, and Copilot — the surfaces where intent increasingly lands. Catalog quality moved from SEO nice-to-have to distribution survival.

Shopify made the distribution the default. On 24 March 2026, Shopify enrolled every eligible merchant in Agentic Storefronts by default. The platform did not wait for merchants to be ready. The pipe is open. The merchants whose catalogs are ready will compound their position for years. The merchants whose catalogs are not will spend the next twelve months wondering why AI referral traffic never materialised.

The principle underneath

Every design decision inside Geopad resolves to a single proposition: commerce is won by catalog quality.

The principle is deliberately restrictive. It says email marketing is not catalog intelligence. It says paid ads are not catalog intelligence. It says CRO, loyalty, reviews, referrals, and customer support are not catalog intelligence. All of those are valuable. None of them replace the work of making every product in the catalog the best possible version of itself.

The principle is also deliberately durable. Whatever comes after ChatGPT — whatever agent protocols get standardised, whatever surfaces emerge for commerce to happen on — catalog quality remains the asset being surfaced. Platforms will change. The catalog will not. The merchant who invests in catalog intelligence in 2026 is investing in an asset that outlasts every specific platform that currently mediates its discoverability.

This is why we built Geopad, and why we believe catalog intelligence becomes the default language of Shopify merchant success by 2028. The first generation of AI commerce tooling focused on generation — make more descriptions, more articles, more alt text. That generation commoditised quickly and built very little defence. The next generation is about intelligence — score the catalog against the standards that matter, remediate at the level of the asset, monitor continuously, and prove revenue attribution over time.

Geopad is the first product built natively for this thesis. BritBoot is the proof it works. Shopify's Agentic Storefronts rollout is the reason it matters now.

Install the app. Run the audit. See the score. Fix what matters.

Geopad is built by Apex Orbis. For press, partnerships, or advisory conversations: hello@geopad.ai.