Technical content buyers can check line by line

Enrich Layer sells B2B data APIs to developers and data teams, readers who check every claim. Its own pages didn't agree with each other.

At a glance

Client
Enrich Layer, B2B data APIs for developers and data teams
Work
Claims register, automated claim checks, technical content, and archive cleanup
Engagement
November 2025 to present
System
Wordful

Before

The turning point

Instead of fixing pages one at a time, every claim went into a register with a status: approved, provisional, disputed, or disallowed. Then the register became checkable by machine. A claim check now runs automatically when content changes, and its first sweep surfaced problems already on published pages.

When two sources disagreed, the answer came from evidence: testing the product, reading the code, and asking the team that owns it.

What we built

After

Read the work

Read the full story

The problem

The company sells data to technical buyers who test what they read against the API. That makes precision the product. A loose line about coverage, sourcing, or pricing gives a careful reader a reason to doubt everything else.

The company's own pages had drifted apart. Different pages described sourcing, coverage, and pricing in different ways, and a writer or an AI agent had no single place to find the current answer.

A register with four states

Every product claim went into one register with a status. Approved claims can ship. Provisional claims ship with care. Disputed claims wait for an answer. Unknown claims never ship. Each claim links to its evidence, so anyone can check why it says what it says.

Checks that run automatically

The register is machine-readable, so a claim check runs automatically when content changes. Its first sweep of published pages surfaced five kinds of problems no one had caught by hand, including help articles that disagreed with each other on billing terms. Four were fixed within two days. The fifth needed a decision about the data itself.

Evidence, not taste

When sources conflicted, the team tested the product, read the code, and confirmed with the people who own it. One help article's billing claim turned out to be false and was corrected. Another false claim came from the company's own approved-claims list: testing against the live product caught it, and the live page was corrected within two days. A disputed sourcing sentence was removed rather than softened, and one sentence that holds true on every page replaced the variations.

Cleaning up the archive

About nine in ten posts in a large legacy archive were retired, and the rest were kept to update and reuse.

What the team keeps

  • A claims register with evidence for every claim
  • An automated check that runs when content changes
  • A pipeline a small team can run, with review focused on the claims that need a person

If this sounds like you

Your readers check every claim? Tell us what you need. No prep needed.

Send a message

Prefer to talk? Book a free scoping call.

Tell us what you need

We'll only use your email to reply. Privacy policy