Blog ICP-bleed test

Contents

• What the data shows
• The finding: the bleed is real, but "zero signups" is the wrong cut axis
• Caveats (read before acting)

What the data shows

The human audience is overwhelmingly off-ICP. Top blog posts by unique human readers (last 90 days) are clinical / patient-education content:

• Therapeutic activities for teens & kids: 220 (Portuguese), 146 (Spanish), plus EN
• 17 questions to ask teens in therapy: 152
• SOAP / DAR / progress note guides, group-therapy games, "porn addiction," "othello syndrome"
• The one clear buyer-intent post in the top ranks: insurance-credentialing-101 (54)

~82% of blog readers are on non-English translated pages. June: 7,733 unique human readers, but only 1,427 on English /blog. The rest is /pt/, /es/, /ko/, /de/, /it/, /fr/, /ja/. For a core ICP of English-speaking US/ANZ practitioners, that's a large non-core audience.

Nothing converts to signup — including the content we want.

Reader cohort	Unique readers	Signups (14-day)	Rate
ICP — credentialing posts	275	0	0%
Off-ICP — clinical/patient posts	3,927	8	0.2%

The finding: the bleed is real, but "zero signups" is the wrong cut axis

The prune matrix cuts pages that "drive zero signups." But the credentialing posts — ICP content tied to the RCM push — also drive zero signups. So that criterion would delete the exact content we're investing in.

A blog is top-of-funnel; the signup happens later, on another subdomain, and almost never attributes back. "Zero signups" mostly means "we can't see it," not "worthless."

The call

Cut / deprioritise on intent + ICP-fit + search value + whether non-core-language content is worth maintaining — not on signups. Re-anchor the scoring on Search Console impressions/position for ICP queries before anything gets deleted.

Caveats (read before acting)

• Only ~1 month of first-party blog data.
• Blog to app is cross-domain, so signups undercount. The 8 stitched conversions prove tracking isn't fully broken, but absolute rates are understated.
• n = 275 on the ICP cohort is small.
• Non-English does not definitively mean off-ICP (emerging markets are a deliberate Tier-3 segment) — but it's clearly non-core and non-converting.

Directionally solid enough to settle one decision: don't prune on signups, and protect the credentialing posts.