Actionable insights — ServiceNow (NOW): the narrative-overshoot screen

1. The beat-that-fell-flat — separate the self-set guide from the "whisper"

1. When a stock drops hard on an earnings "miss," check what it actually missed: the company's own guidance, the published consensus, or an unofficial "whisper number" the most aggressive bulls priced in.
2. A name that beat every self-set metric but fell short of a whisper has been sold on expectations, not fundamentals — a sentiment dislocation, not a thesis break.
3. Confirm the franchise is intact: no lost marquee customers, renewal rate steady, backlog (RPO/cRPO) still growing.

• Big post-print drops where the company met its own guide; whether the franchise metrics (renewal, RPO) actually deteriorated.

2. Invert the disruption narrative — find where the feared trend is a tailwind

1. Take the consensus disruption story (here: "AI agents make enterprise software unnecessary") and test it against what the company architecturally does.
2. Ask whether the company is the layer the new technology must route through — a system of record AI must read before acting and write after — rather than a layer it replaces.
3. If so, the feared trend increases usage; quantify the new usage driver (decisions/queries scaling from hundreds to tens of thousands per day).

• Companies sold on a disruption theme that are actually the mandatory control point for that disruption; usage/query-volume growth as confirmation.

3. Spot the consumption-revenue meter — usage-based pricing on a fixed cost base

1. Look for a shift from seat-based to consumption-based pricing (tokens/packs charged per AI action) — it decouples revenue from headcount.
2. Pair the autonomous-resolution rate (here 90% of IT tickets) with the consumption model: every automated action rings the register against a fixed infrastructure cost = operating leverage.
3. Track the new AI line's net-new ACV growth as the leading signal of the consumption ramp.

• Net-new ACV of the AI module; autonomous-resolution percentages climbing as the consumption layer compounds.

4. Cash-adjust the multiple, then cross-rank against slower growers

1. Strip net cash out of the market cap before judging "expensive" — a big cash pile lowers the true EV/sales and EV/FCF.
2. Anchor the multiple to the company's own 5–10-year range (NOW's 8× P/S is its lowest since 2016) and to the precedent that followed (a 400%+ move).
3. Sanity-check by ranking it against slower-growing names on the same metric — if a 20%-grower trades below a low-growth industrial, the mispricing is explicit.

• A high-grower at a decade-low multiple after cash adjustment; a faster grower priced under a slower one on the same yardstick.

5. The double insider tell — open-market buy plus cancelled 10b5-1 plans

1. Note open-market purchases by the CEO (cash, not grants) at the depressed price as a confidence signal.
2. Layer in the stronger, rarer tell: executives simultaneously cancelling pre-planned 10b5-1 sale programs — they don't want to be seen selling.
3. Use the combination as corroboration of the mispricing, not a standalone thesis.

• Form 4 open-market CEO buys at lows; cancellations of scheduled 10b5-1 sale programs.

6. Wait for the base, not the knife — volume signature + a dated catalyst

1. Identify the structural long-term support level and accept the downtrend may not yet be broken — don't "catch a falling knife."
2. Look for a constructive volume signature (heavier buying than selling on down days) that precedes a base rather than a continuation lower.
3. Time conviction to a dated catalyst (next earnings) that can flip the narrative, given a setup (prior beat + accelerating backlog) that favors another beat.

• Down-day buying volume building a base; the next earnings date as the entry-confirmation trigger.