A headline KPI whose measurement basis was widened by the very transaction under legal attack

Claim

Audience KPIs are measured over a defined set of properties. That set is a disclosure choice, it changes, and the change is usually buried in a definitional footnote rather than flagged at the metric.

The dangerous case is when the perimeter widens because of a transaction — and the sharpest version is when that transaction is the subject of an enforcement action, because then the metric is inflated on the way in and will deflate on the way out, for reasons that have nothing to do with the business's own performance.

Evidence — Zillow Group, Rentals

Zillow's FY2025 10-K defines "average monthly rentals unique visitors" as a Comscore measure covering Zillow, Trulia and HotPads —

"…and on Realtor.com and beginning in February 2025, Redfin and its sites, including Rent.com and ApartmentGuide.com."

February 2025 is exactly when the Zillow–Redfin rentals agreement took effect — the ~$100M, nine-year deal under which Redfin shut its internet-listing-services business, and the deal the FTC and five state AGs sued over on 2025-09-30.

Two consequences, and the second is the one that bites:

The headline Rentals audience metric was expanded by the transaction, so growth across the February-2025 boundary mixes organic audience with a definitional widening.
The per-visitor monetization ratio inherits the corrupted denominator. Zillow reported rentals revenue per unique visitor of $4.77, +31% — a ratio whose numerator is Zillow's own revenue and whose denominator now includes an acquired competitor's audience. Any cross-boundary comparison of that ratio is meaningless.

And the basis is now set to break a second time. The FTC settlement of 2026-08-24 requires Redfin to re-enter the rental listings market within six months, hire staff, and invest to become a stronger competitor — while continuing to syndicate Zillow's listings through 2030. Redfin moving from "Zillow's exclusive syndication partner" back to "independent competitor" is another perimeter question, and it will land in the middle of the comparison base.

Detection

Read the KPI definition, not the KPI. In a 10-K these sit in Item 7 or a "Key Metrics" section, and the perimeter is often a subordinate clause with a date in it.
Grep the definition for a month-year. A date inside a metric definition is nearly always a perimeter change. Then check what happened at the company on that date.
Any deal that brings a rival's audience under the same measurement umbrella — acquisition, syndication, exclusive partnership, white-label — is a candidate. Ask "whose properties are in this number, since when?"
Cross-check the audience KPI against a metric with a fixed perimeter — owned-app downloads, logged-in users, or the revenue line itself.

Rule

Before quoting an audience KPI or any per-user ratio built on it, establish the measurement perimeter and its effective date, and state them. If the perimeter changed inside the comparison window, either restate both periods on a common basis or decline to compute the growth rate — do not report a rate across a basis break.

Note on how this interacts with the legal risk

The general lesson generalises past antitrust: a KPI expanded by a deal is a KPI exposed to that deal's reversal. Where a company's growth metric and its legal jeopardy share a single root cause, the metric is not independent evidence about the business — it is partly a reading of the deal. That is worth pricing separately from operating performance.

Zillow is also a live example of the mirror-image trap: the same agreement drove a −450bp gross margin drag through lead-acquisition costs, so the deal inflated the audience metric while deflating the margin. Reading either number without the deal in view gets the business wrong in opposite directions.

What would falsify this

A company restating its KPI history on a consistent perimeter when the perimeter changes, or disclosing the metric both ways across the transition. That is the correct practice and some issuers follow it; the pitfall exists because many do not.

[[pattern-non-gaap-definition-change-leaves-a-restatement-fingerprint]] — the same failure applied to a non-GAAP financial metric rather than an operating KPI; both fired on Zillow in 2026 · [[pitfall-logo-retention-masks-dollar-retention]] · [[pitfall-third-party-panel-contradicts-issuer-disclosure]] · [[pitfall-growth-ratio-against-collapsing-base]] · [[principle-primary-source-beats-vendor]]

History

2026-08-26 — found in the Zillow analysis while mapping the Rentals revenue stream. Caught by reading the 10-K KPI definition rather than the KPI, and noticing the February-2025 date matched the effective date of the agreement the FTC was suing over.