Flat margins during share loss mean the damage is ahead, not that the moat held

Claim

analysis_notes.md §2 sets the quantitative moat test as "ROIC trend and gross-profit-margin trend over time. Rising/stable high ROIC = a real moat throwing off returns. Margin compression = moat erosion."

Run literally, that test passes a company in the middle of losing its franchise, because it is a one-variable test on a two-variable problem.

An incumbent facing a credible new entrant has exactly two responses:

Response	Volume	Margin	What the moat test sees
Cut price to defend share	holds	falls	✅ correctly flags erosion
Hold price and cede share	falls	holds	❌ reads as an intact moat

The second row is the trap. It is also the default corporate response, because cutting price is irreversible, visibly damages the P&L this quarter, and admits the competitor is real.

Evidence — ZTS, 2026-08-05

	Q1 2025	Q1 2026
Gross margin	71.9%	71.7%
Operating margin	36.9%	36.3%
ROIC (annual, 2022→2025)	18.8% → 23.1% (rising)
US companion animal revenue	—	−11%
Total organic revenue growth	—	0%

On margin and ROIC alone, Zoetis passes the §2 test outright — ROIC rose over the period. Yet in the same quarter it lost 11% of its highest-margin revenue line to Elanco's Zenrelia, Befrena and Credelio Quattro, with generic oclacitinib still to arrive.

The 20bp of gross-margin decline is not evidence the moat held. It is evidence the price war has not started. Management's own Q1 language — "price has played a larger role in the decision process" — signals the move to row 1 is coming.

Why this matters for valuation, not just for scoring

The error is not academic; it propagates straight into the fair-value range. If margin is read as intact, the forward earnings base looks safe and the only question is the multiple. Reading the pair correctly reframes it: the earnings base itself is the uncertain variable, which widens the fair-value range and lowers the weight on any multiple-based scenario.

On ZTS this was the difference between "cheap quality" and "cheap, with the largest risk still unpriced." Sizing it: at 71.7% gross margin on ~$9.5B of revenue, 100bp of gross margin ≈ $95M of operating income ≈ 3.5% of net income. A retreat from 71.7% to the mid-60s — ordinary when a monopoly becomes a duopoly against a generic — is a ~20% earnings event that no margin-trend screen would have flagged in advance.

How to apply

1. Never score the moat on margin alone. Always pull the volume/unit/organic-growth series alongside it and read them as a pair.
2. Diagnostic: margin flat and volume falling → the company is ceding share. Ask "how long can they hold price?", not "is the moat intact?"
3. Write the fork explicitly into the report. Name both branches (cede volume / cut price) and state that neither is benign. A moat section that reports "margins stable" without the fork is incomplete.
4. Set the tripwire on margin, not on revenue. Revenue decline is the lagging confirmation that price was held; the first gross-margin break below the historical band is the signal that the company switched branches. On ZTS the tripwire was set at gross margin <70%.
5. Applies symmetrically to the bull case. A company whose margin holds and whose volume holds is genuinely defending. The pair is the signal; either alone is noise.

What would falsify this

An incumbent that holds both price and volume through a credible entrant's launch — which would mean the entrant is not credible, or the market is expanding fast enough to absorb both. Zoetis management explicitly ruled the second out: "competition has not yet translated into overall market expansion" while "the derm market has declining patient volume in the clinic."

Related

• [[pattern-shortage-pricing-masks-share-loss]] — the same failure with the price variable doing the masking instead of the margin variable. These two are siblings.
• [[principle-story-vs-revenue]] · [[pattern-buyback-manufactured-eps-screens-as-growth]]
• [[pitfall-gate-cleared-by-adjective-not-arithmetic]]