The Growth Persistence Ratio: Why ServiceTitan Lost 30% in a Day (Founder’s KPI Brief, September 17, 2026)

ServiceTitan beat its quarter yesterday. Revenue came in at $292.8M against an 18% guide — 21% actual growth. Non-GAAP EPS hit $0.40 versus $0.36 expected. At $1.14 billion in revenue, that is a clean beat on both lines.

The stock fell 30% in a single session.

The reason was one sentence in the guide: back-half growth decelerating to mid-teens. That’s the entire story. And it’s the most important KPI lesson of 2026 for anyone operating a SaaS company above $10M ARR.

The Metric: Growth Persistence Ratio

Growth Persistence Ratio (GPR) measures how much of your current growth rate carries into the next period:

GPR = Forward-period growth rate ÷ Current-period growth rate

ServiceTitan’s math: ~15% forward ÷ 21% current = 0.71.

That number — not the beat, not the EPS — is what got repriced. Public markets have converged on a rough band, and private investors are now applying the same lens in diligence:

• GPR above 0.85 — premium multiple. Growth is structural, not campaign-driven.
• GPR 0.75–0.85 — market rate. Acceptable maturation.
• GPR below 0.75 — derating zone. The market assumes your growth was pulled forward, not compounded.

The uncomfortable implication: you can beat every quarter and still lose 30% of your valuation if your forward number tells a decay story. Beating the quarter no longer buys forgiveness for a decelerating forward curve.

Why This Happened in 2026 Specifically

Two forces collided this year.

1. AI collapsed the cost of the feature layer

OpenAI shipped GPT-6 Astra today with explicit AGI-onset framing. Separately, Microsoft’s MAI-Transcribe-2 is now undercutting OpenAI, Google, and ElevenLabs on both price and speed in speech. Speech AI became a commodity input in roughly 18 months.

When capability gets cheap, the growth that came from having the feature evaporates. Growth that comes from owning the workflow persists. GPR is the cleanest way to tell which kind you have — because it shows up in the forward guide before it shows up in churn.

2. Capital got selective about the shape of growth

Look at what actually cleared the bar in the last 24 hours. Fresha — beauty and wellness booking — raised $80M from KKR’s next-generation technology arm at a $1B valuation. Lio, an AI procurement play, raised a $30M Series A led by a16z. Neither is a thin model wrapper. Fresha owns transaction rails inside a vertical. Lio touches a budget line, not a productivity habit.

The thesis that died in 2026: GPT wrappers with no proprietary data loop. The thesis that got funded: software that sits on a cash outflow the buyer cannot stop approving.

How to Instrument This Monday

You don’t need a public comp set. Build GPR from your own pipeline in three steps:

1. Segment new ARR by durability class. Split last four quarters of net new ARR into (a) expansion within existing workflows, (b) new-logo core product, (c) feature-led / AI-novelty attach. Category (c) is your decay-prone bucket.
2. Compute forward-weighted growth. Re-forecast next two quarters assuming category (c) growth halves. If your GPR drops below 0.75 under that assumption, your board deck is currently overstating durability.
3. Pair it with CAC payback. GPR below 0.80 plus CAC payback above 18 months is the combination that produces flat or down rounds. One is survivable. Both together is a structural problem.

Two supporting KPIs worth adding to the same dashboard this quarter: net revenue retention split by cohort age (durability hides in cohorts 24+ months old) and model-portability coverage — what percentage of your AI features can failover to a second provider. That last one moved from engineering hygiene to commercial requirement after the Pentagon designated Anthropic a supply-chain risk over a collapsed $200M contract. Enterprise buyers will start writing multi-model fallback clauses into MSAs.

The India Read-Through

Commodity inference is the single biggest gross margin lever available to Indian AI SaaS this quarter. The chronic critique of India-built SaaS has been gross margin — 55–65% versus 75%+ for US peers. MAI-Transcribe-2’s price floor directly attacks that gap for anyone selling AI-inflected product to global mid-market.

But margin relief doesn’t fix GPR. India’s vertical SaaS cohort — the “own the trade” playbook ServiceTitan wrote — should now underwrite to mid-teens terminal growth, not 20%+. That changes hiring plans, burn tolerance, and the ARR level at which you can credibly raise.

The offsetting opportunity: procurement AI. Lio’s round validates a wedge where India has genuine structural advantage — GST-linked invoice data, e-invoicing mandates, and ONDC rails create a compliance moat US startups cannot replicate. Expect a Bengaluru fast-follower inside two quarters.