1. Weigh backlog against the capex + financing needed to deliver it
The repeatable method
- When a company reports an exploding order book (RPO / backlog / bookings), don't price it as value yet — find the spending and financing required to fulfill it.
- Pair the backlog with capex guidance, the funding plan (debt/equity raises), and free cash flow — a backlog that requires negative FCF and fresh raises is a cost, not yet a win.
- Reframe the stock as a bet on execution: can it build and operate the capacity profitably before the financing strain bites?
Here: ORCL's RPO jumped +363% to $638B, yet the stock fell 12% — the market "stopped paying for backlog and started pricing the cost to deliver it" (~$70B capex, ~$40B raise, FY26 FCF −$24B). "Now a bet on execution."
Watch for
- A record backlog announced alongside rising capex, a new raise, and negative FCF — the tell that delivery cost, not demand, is the new question.
2. Make the gross-margin step-down the fault line in a mix shift
The repeatable method
- When a company grows by scaling a lower-margin line (e.g. renting raw compute vs selling software), track the blended gross margin direction, not just revenue growth.
- Listen for management explicitly guiding margins to "step down" — that admission, plus the mix math, tells you growth is being bought with margin.
- Decide whether the volume more than offsets the margin give-up (gross-profit dollars still rising) or whether you're funding growth that dilutes returns.
Here: ORCL's gross margin fell 5pp to 65% as lower-margin IaaS scaled, and the new CFO guided FY27 margins to step down further — the structural cost of the AI-cloud mix the backlog implies.
Watch for
- "Margins will step down" in guidance paired with a fast-growing low-margin segment — quantify the gross-profit-dollar trade before cheering the revenue.
3. Judge a freemium pivot by paid conversion, not free-user counts
The repeatable method
- When a company gives away a product to build an audience, separate the vanity metric (free MAU) from the one that pays the bills (paid conversion / ARR).
- Recognize the playbook (give the basic tier away to make your paid tier the standard) and the cost: revenue can stall while you wait for conversion.
- Set a clock — give the funnel a few quarters and watch whether free users actually convert; until they do, treat the strategy as unproven.
Here: ADBE grew freemium creative MAU from 50M to 90M (the "Acrobat Reader playbook applied to AI") and shifted to credit-based pricing — but with the stock −40% YTD at 9x fwd EBITDA, the open question is visible paid conversion within a few quarters.
Watch for
- Soaring free-user counts with flat paid ARR — and leadership turnover (a departing CFO) during the transition, which raises execution risk.
4. Normalize a weather/cyclical shock against its long-run baseline
The repeatable method
- For a business driven by an exogenous, mean-reverting input (snow, weather, commodity prices), measure the current period against a multi-decade baseline to size how abnormal it is.
- Separate the one-off shock from any structural demand change — check forward indicators (advance bookings / pre-sales) for whether customers are truly leaving.
- Gauge the buffer: cost-savings programs, dividend coverage, and a "plan for normal next year" tell you how much the company can bridge until conditions revert.
Here: MTN had snowfall 55% below the 30-yr average (worst season on record) and skier visits −16% — but it leaned on a $106M cost-savings plan, held the $2.22 dividend, and planned a "normal" next season. The caution flag: 2026/27 pass pre-sales already −10% units.
Watch for
- Advance pass/booking sales falling even as management blames weather — that's the line between a one-season shock and softening underlying demand.
5. Use the recurring-revenue mix as the macro-defense gauge
The repeatable method
- Split revenue into recurring/subscription versus discretionary/one-off, and track the recurring share — the higher it is, the more the business resists a consumer slowdown.
- When guidance is trimmed, check where: a cut concentrated in discretionary lines while recurring holds (and margins expand) signals "stretched but steady," not broken.
- Pair the recurring mix with free-cash-flow growth to confirm the defensive base is actually converting to cash.
Here: CHWY's Autoship hit 84.4% of net sales (+220 bps) with FCF +45%, so even as it trimmed FY guidance on soft discretionary pet spend, EBITDA margin still expanded ~130 bps — the recurring core carried the print.
Watch for
- A rising recurring-mix percentage holding the line through a guidance cut — the recurring base, not the headline, is the durability signal.
6. Read a pricing-model shift as the real growth lever
The repeatable method
- When a company moves from a fixed per-unit price (per signature, per seat) to usage/credit/outcome-based pricing, ask whether it expands or caps revenue per customer.
- Track the newer, stickier product line's share of recurring revenue (ARR) and its trajectory toward management's target — that's where the re-rating case lives.
- Treat ecosystem integrations (outside AI models, partners) as evidence the platform — not a point product — is the unit being sold.
Here: DOCU shifted IAM to credit-based, outcome-tied pricing (away from per-signature), grew IAM to 12.6% of ARR (→18% target), and wired in Anthropic/OpenAI/Harvey/Thomson Reuters — the platform-vs-point-product pitch (Deloitte: ~3% vs ~30% AI ROI).
Watch for
- A usage/credit pricing switch alongside the new product line crossing into double-digit ARR share — the combination that signals durable revenue-per-customer expansion.