CASE STUDY · E-COMMERCE BRANDS
Compatibility data structured so an agent can answer 'will this fit?'
A B2B distributor's buyers were starting to arrive through AI assistants asking compatibility and lead-time questions the catalog could not answer. We made the fitment and availability data machine-readable and exposed it through a governed interface.
How to read these numbers
This client asked not to be named. The engagement shape is real; the figures below are directional estimates pending client sign-off, not audited results. (Directional estimate · client not named)
Top 20%
Of part families remediated first
Ranked by quote volume
~2x
Coverage of answerable fitment questions
Directional; pending client sign-off
0 claims
Published without human review
Compatibility is a liability surface
Half our value was in reps knowing what fits what. Putting that in the data did not replace them, it stopped them retyping the same answer twenty times a day.
Paraphrased from the engagement and shown without a named person while we wait on client sign-off. We will not attach a name or a title to a quote the client has not approved.
Context
- A distributor carrying tens of thousands of parts across multiple manufacturer lines.
- Compatibility knowledge lived with inside-sales reps and in PDF spec sheets, not in the product record.
The problem
- Agent-mediated buyers asked fitment, minimum-order, and lead-time questions that returned nothing usable.
- Quoting still required a human to open two systems, so after-hours demand simply went elsewhere.
What we did
01 · Define
Ranked part families by quote volume and scored each on the Agent Readiness Ladder.
02 · Design
A fitment and availability schema per family, plus a read-only agent interface with rate limits and an audit trail.
03 · Build
Agent-assisted extraction from spec sheets with human review on every compatibility claim, feeding a validated product feed.
04 · Measure
Tracked answerable-question coverage on the top quoting families and after-hours request capture.
Tooling shown is what was used in this engagement, not a required stack. We build to the platforms and systems you already run.
What we would do differently
We would set the review SLA with inside sales before extraction started. The review queue, not the extraction, was the bottleneck.
NEXT STEP
Recognise the pattern?
Tell us where your signals live and we will tell you whether this shape fits.