At a mid-size distributor, you are the search function. Not a team. You. So “should we do SEO or AI search” is a question that never lands — both are your job and there is one of you. This path is the hybrid version of the work: category pages ranking on Google, and the catalog cited inside ChatGPT, Perplexity, Gemini and AI Overviews, run as one program by one person. The field hasn’t agreed on a title yet (AI SEO Specialist, AI Search Specialist, SEO & AI Search Specialist all name the same role). The job is real. 51% of B2B buyers now start research in an AI chatbot (G2 Buyer Behavior Report, Apr 2025).
One hire, one program
Treat this as one program, not two teams. You don’t have the headcount to split it, and splitting it is the wrong instinct anyway — the technical groundwork is the same data and the same templates either way. A buyer hunting a Char-Lynn 104 seal kit might land on you through a Google result, a ChatGPT answer, or an Amazon Business listing that scraped your specs. Those aren’t separate channels with separate owners. Same product, found in different places, and your job is to be in all of them. The title hasn’t converged. The work has.
Don’t skip classic SEO
There is no AI-search shortcut around SEO fundamentals. Every GEO and AEO posting wants 3–10 years of prior SEO for a reason. Category pages still have to rank. Internal links still have to route authority. A site that crawls badly is invisible to everyone, human or machine. AI search is a layer on top of solid SEO, not a swap for it. The on-ramp is honest about this: there are essentially no entry-level GEO/AEO jobs you get hired into cold, so “entry” means a junior SEO who picked up AI-tool fluency, not a separate discipline you skip ahead to. Put the rankings fire out before you touch the AI work.
Produce at catalog scale without shipping slop
AI is genuinely useful at catalog scale: drafting, attribute cleanup, building spec tables out of messy supplier data. The trap is distributor-specific. Most of your product copy came straight from the manufacturer, so the same paragraph sits on your page, the manufacturer’s page, and every competitor who pulled the same feed. That is sitewide duplicate content, and an answer engine has nothing to choose between. Dedup is the deliverable here. Not volume. And clean PIM attributes sit upstream of all of it — if the attributes are a mess, you can’t even build the facet pages, let alone make them different from everyone else’s.
Win the part-number long-tail
This is the most distributor-distinctive lane you have. “Gates equivalent of Parker 387 hose.” “1756-L61 replacement.” “What replaces the discontinued PowerFlex 4?” “Imperial equivalent of Class 10.9.” Near-zero reported search volume, near-100% buyer intent, and answer engines have no good source for them today. Rockwell answers obsolescence questions only inside gated PDFs. Parker buries cross-references in a JavaScript lookup tool. So publish your interchange and cross-reference data as flat, crawlable HTML — not a PDF, not a JS widget. A small hydraulics distributor that does this gets cited exactly where the manufacturer’s own tool can’t be read. That is the whole play.
Make the catalog readable to engines and crawlers
You can do everything above and still be invisible. Two blockers do it. First, login-walled pricing and specs: if the data only renders after a sign-in, an engine never sees it. Second, aggressive bot protection that blocks the retrieval crawlers — OAI-SearchBot, PerplexityBot — that fetch pages to ground a cited answer. Block those and you are out of the answers, full stop, however good the page is. Add Product schema with MPN and GTIN, and do it at the template, not the SKU. One fix lands across 100,000+ products. The unit of work is always the template.
Report blended visibility, and survive the AI-Overviews dip
Put Google rankings and traffic next to AI mention and citation share in one report. You can’t proxy AI visibility with rankings — only about 12% of AI-cited URLs rank in Google’s top 10 for the same prompt. A page can be quoted by an engine without ranking, and the reverse. So fix a set of real buyer prompts, run it on a schedule, and track citation share against named competitors. Then there’s the hard conversation: AI Overviews eating clicks. Traffic drops, leadership wants to know what happened, and “AI Overviews” is not an answer anyone can act on. Your job is to separate what is actually lost from what got replaced by a citation, and brief that plainly instead of letting the traffic chart do the talking.