What does the AI-mediated sourcing journey look like?
The classic funnel — inquiry, quotation, negotiation — now starts long before your inbox. A buyer's journey runs through five AI-mediated stages, each with its own prompt patterns and content requirements. The commercial consequence is structural: by the time an RFQ arrives, the buyer has already been educated (possibly by your competitor's content), built a longlist, verified the candidates and framed the comparison. Suppliers invisible in stages 1-4 experience this as mysteriously declining inquiry quality and win rates.
Stage 1: Problem education — what solution exists?
Prompts: how to solve X, what technology for Y, options for Z. Engines answer from educational content across the web. Your play: genuinely useful category education on your domain — application guides, technology explainers — that earns citation without pitching. This stage seeds Semantic Coverage in the four-dimension standard.
Stage 2: Category and longlist — who does this?
Prompts: best manufacturers of X, suppliers of Y in China, alternatives to Z. Engines answer from directories, trade press, review platforms and listicles. Your play: the source graph — directory completeness, trade-media presence, review-layer health. This is the stage where absence is fatal and invisible; the verification queries playbook covers the adjacent trust prompts.
Stage 3: Verification — is this supplier real?
Prompts: is X legitimate, X reviews, X complaints, where is X registered. Engines answer from registries, reviews, news and community. Your play: entity consistency, claimed review profiles, the legitimacy page, negative-context content where needed.
Stage 4: Comparison — which one fits my requirements?
Prompts: X versus Y, X MOQ and lead time, does X hold certification Z. Engines answer from spec tables, certification documentation and comparison content. Your play: quotable commercial content — MOQ and pricing tables, certification pages, spec comparisons — in the buyer's language.
Stage 5: Contact — the RFQ you finally see
By now the buyer's requirements are shaped. Your play: make contact frictionless and the first response instant — and feed the questions buyers ask at this stage back into Stage 1-4 content, closing the loop. Your sales inbox is the best prompt-research instrument you own.
Measuring the journey
Tag your fixed 50-100 question set by stage and track the four dimensions per stage monthly. The diagnostic value is immediate: strong Visibility at Stage 2 with weak Citation Rate at Stage 4 means the source graph works but your commercial content does not. Run the 90-day POC per journey stage: Week-1 baseline, Weeks 2-10 intervention on the weakest stage, Week-12 re-measurement.
One organizational implication: this journey cuts across marketing and sales, so the measurement set must be shared. When the Stage 2 longlist prompts live with marketing and the Stage 5 RFQ data lives with sales, no one owns the gap between them. The four-dimension dashboard, reviewed monthly by both teams, is the simplest governance fix — one question set, one baseline, shared accountability for the stages neither team sees directly. The country guides show how the stage mix shifts by market.