Fit

Fit

Outside the framework's universe (U2 not met); contested: P5

Docusign misses the framework's scale line: a market capitalization of $9.84B against the greater-than-$10B universe test. Confidence is low, on the tally's computed basis — a load-bearing criterion was contested or a load-bearing probability spread exceeded 0.25 (P5, spread 0.26). No exclusion was hit, the China sensitivity flag was not raised, the watchlist-only flag is not set, and the name-mask probe raised no prior-driven risk.

Universe and exclusions

Here is the decisive point. The miss is not a judgment on the business; it is a size test, and Docusign is 1.6% under it.

Derived from fit_features.market_cap: price $47.04 on 2026-07-23 times 209,118,000 shares = $9,836,910,720. The break-even price ($10.0B / 209.118M = $47.82) is derived from those same feature inputs and is not stated in any filing. Share count from the FY2026 Form 10-K [1].

The counter-fact belongs in the same breath: this is the narrowest possible miss, and it is a price-of-the-day fact rather than a finding about the company. The skeptic pass weakened the underlying claim in one respect — the external-tracker corroboration ($8.6–9.0B) was struck from the corpus-only record, leaving the deterministic feature as the sole basis. The listing test is clean: Docusign is a Delaware corporation whose common stock trades on the Nasdaq Global Select Market under the symbol DOCU, not an ADR and not a Chinese issuer [2]. The full treatment is in Business.

The exclusion screen is clean. All four disqualifying exclusions were checked and none was hit; the tally's exclusion_hits list is empty. Docusign manufactures no vehicles — its named competition is Adobe Acrobat Sign plus software vendors [3], so X1 does not apply. X2 (promotion pattern) requires both prongs: insider economic ownership is thin — directors and executive officers hold under 1% of the 195,945,841 shares outstanding, with no open-market purchases [4] — but the delivery record is mostly met or beaten (FY2026 billings $3.4B against a $3.30–3.354B guide, non-GAAP operating margin 30% against 27.8–28.8%) [5], with one timing-driven Q1 FY2026 billings miss, so the repeated promise-versus-delivery gap is not established. X3 (structural decline) does not fire mechanically: revenue has risen in every fiscal year on record and revenue_trajectory.three_year_hsd_decline reads false. X4 (consensus-saturated story) does not fit a name at roughly 3.1x sales, 61% below its peak, with sixteen of eighteen ratings at hold. S1 (China dependence) is absent: China is not mentioned across the FY2022–FY2026 annual reports, and non-US long-lived assets are $89.9M of $528.6M in total [6].

Pattern match

Docusign fits none of the framework's four setups. It is not a cyclical at the bottom and not a large bank. It pays no dividend and does not expect to [7], so the yield-plus-dividend pattern cannot apply. It is not an insurer mis-forecasting a cost trend, and there is no premium-repricing mechanism to underwrite.

The nearest neighbour is the fourth setup — a quality technology franchise on a fear dip — and it fails that pattern's two specific checks. The structure check asks for a monopoly or duopoly; the filings describe one primary global competitor (Adobe) alongside content-cloud bundlers, providers of large language models, and customers' own homegrown alternatives, in a category whose enabling laws (ESIGN, UETA, eIDAS) are vendor-neutral rather than protective [3]. The fear check asks whether the specific fear is testable; the AI-substitution fear is specific but shows no signature yet in the usage data — dollar net retention was over 102% in Q1 FY2027, up more than a point year over year and sequentially higher for seven quarters, with consumption at multi-year highs and customers spending over $300,000 in annual contract value up 12% [8]. An untestable fear is neither confirmed nor cleared, which is why the diagnosis criterion ends contested rather than decided.

The pillar ledger

Source: the run's deterministic fit tally (per-criterion aggregates, votes and spreads); the arithmetic behind each row is cited in the sections that follow.

Year-10 durability — P1

Not met, at probability 0.575 with a spread of 0.10; all four seats agreed and both model families agreed, and the name-masked seat returned the same result. The gate asks for very high conviction that year-10 revenue and adjusted free cash flow are both higher than today's. The conviction sources grade poorly: capital intensity is absent — capex of $106.4M [9] on revenue of $3,219.5M [1], 3.3%; the regulatory regime enables the category for any compliant vendor rather than gating entry; and the market is fragmented rather than a duopoly [3]. The company's own risk factor is prominent on the point: advances in AI "have made and may continue to make some foundational capabilities for agreements … cheaper and easier to replicate" [10]. IAM was 12.6% of total annual recurring revenue at 30 April 2026, guided to about 18% at fiscal year end, which leaves roughly 82% of the recurring base in the exposed legacy capability [8].

The strongest surviving counter-fact sits inside the same treatment: decline is not the base case. Revenue has risen in every fiscal year on record, reported free cash flow rose to $1,058.6M in FY2026, the balance sheet is net cash, and consensus models free cash flow rising to about $1,328M by FY2029. What the gate withholds is conviction, not a forecast of shrinkage — and by construction, genuine doubt resolves to not met. Full treatment in Durability.

FCF consistency — P2

Not met, four seats unanimous, no probability recorded. The deterministic fit_features.fcf_stability series is not_computable because stock-based compensation is absent from the structured cash-flow feed, so the series below is reconstructed from the filed cash-flow statements and is an analyst derivation rather than the feature.

Adjusted FCF = reported FCF − stock-based compensation − trailing five-year average acquisition spend; derived from the filed cash-flow statements [9]. The deterministic feature is not_computable for this series.

Three positive years on the adjusted basis is a rising ramp, not the stable rolling five-year average the criterion asks for, and the two negative years are early-stage compensation drag rather than a business-model-inherent cycle recurring every five to eight years. The counter-fact in the same treatment: reported free cash flow has risen every year from $445.1M to $1,058.6M, subscription cash flows recur rather than cycle, and the objection is the length of the adjusted record and its dependence on compensation compression — not volatility. Full treatment in Durability.

Dislocation and yield — P3a, P3b, P3c, P3d

P3a — identifiable event: met, unanimous. The dated trigger is the 5 June 2025 first-quarter report: the full-year billings guidance midpoint was trimmed from $3,327M to $3,312M, about 0.45%, on a self-inflicted early-renewal and sales-compensation change [11], and the stock fell 18.97% the next session on 22.6M shares, 8.9 times the 50-day average. The counter-fact: revenue guidance was raised $22M the same day, and the two larger later legs (30 September 2025 and 3 February 2026) carried no company numbers cut at all.

P3b — capitulation: met, unanimous. The deterministic gauge measures sustained volume in the fall leg at 2.24 times the 180-day pre-peak median, clearing the 2x reference line. The counter-fact: 2.24x is well short of the three-to-five-times flood of a single climactic washout, and the two heaviest windows sit about eight months apart — episodic selling rather than one exhaustion point. Full treatment in Dislocation.

P3c — yield versus the bar: not met, unanimous, position 437 bps below. The balance sheet is net cash — $866.5M of cash and short-term investments plus $208.4M of long-term investments, no funded debt [12] — which selects the most generous reference band available, 8–9%. FY2026 adjusted free cash flow of $406.2M on a $9,836.9M market cap is a 4.13% yield; the three-year average is 2.99%. The counter-fact in the same breath: reported free cash flow of $1,058.6M yields 10.76% and clears the 10% default bar outright — the entire difference is the $622.3M of stock-based compensation the adjusted basis removes [9].

P3d — forward path: not met, probability 0.34 with a spread of 0.04, unanimous.

Adjusted-implied path = consensus reported FCF (FY2027 $1,133.0M, FY2028 $1,232.9M, FY2029 $1,328.0M) less stock-based compensation assumed at 17%, 15% and 12% of consensus revenue, less about $30M of normalized acquisition spend, divided by the $9,836.9M market cap. The SBC assumption is a modeled scenario — consensus publishes no SBC-adjusted free cash flow. Bar = the 8.5% mid-point of the fortress band.

The path reaches the fortress band only around FY2029, and only if compensation compresses from about 19% to 12% of revenue. The counter-fact is that the mechanism is already running: stock compensation fell to 17% of revenue in Q1 FY2027 from 19%, and diluted weighted-average shares fell 8% year over year to 196.5M [13]. Full treatment in Yield.

Balance sheet and self-help — P4a, P4b, P4c

P4a — outlast and allocation headroom: met, unanimous, at probability 0.92. There is no funded debt, about $1.07B of cash and investments, and a $750M revolving facility undrawn and covenant-compliant to May 2030 [12]. No principal falls due in any year, so nothing forces capital allocation toward debt paydown at the moment repurchases would matter most. The counter-fact: the same cash flow is heavily consumed by $622.3M of annual stock compensation, so the surplus available for genuine share shrinkage is far smaller than headline free cash flow implies.

P4b — repurchase engine: not met, unanimous. The hard fail keys on the deterministic share_count_trend.rising = true, a five-year weighted share-count CAGR of +2.4%. Executed repurchases are real and scaling — $1.76B across FY2023–FY2026, of which $869.1M in FY2026 [9], plus a record $318M in Q1 FY2027 against a remaining $2.4B authorization [13]. The skeptic pass weakened the claim to exactly this reading: recent dilution-offset and turning-point evidence, not a clean engine. The arithmetic that decides it: FY2026 buybacks of $869.1M against $622.3M of compensation is 1.40x, so net repurchase beyond the dilution offset is roughly $246.8M. On the framework's absurdity check, retiring the entire float takes 9.3 years of reported free cash flow or about 24 years of adjusted free cash flow — distant from the roughly three years that marks a price making a claim it cannot support.

P4c — dividend cover: not applicable. No dividend has been declared and none is expected [7], so the return case rests on repurchases alone. Full treatment in Self-Help.

Diagnosis — P5

Contested at p_temporary = 0.63 with a spread of 0.26. All four jury seats recorded the criterion as contested, carrying the adversarial trial's ruling rather than re-eliciting it. The two readings and the split are set out in the contested section below.

The arithmetic that frames it: enterprise value fell about 59%, from roughly $21.4B to roughly $8.8B, while the near-term numbers barely moved — FY2026 revenue and EPS both beat the consensus that stood before the trigger. On reported free cash flow and a +4% perpetuity that leaves a gap of roughly $10B between price and value. The counter-fact, in the same treatment and on the basis the framework actually uses: on full adjusted free cash flow of about $406M (about $436M before acquisition normalization) the +4% perpetuity value is about $6.8–7.3B, below the current enterprise value of about $8.8B — so the gap is absent on Ruchir's own owner-earnings basis. Full treatment in Damage Math.

Instrument context — I1

Not verifiable, unanimous. Listed options with expiries of twelve months or more are plausible for a standard-cycle Nasdaq name, and dated implied volatility of 53.9% at 30 days and 56.3% at 180 days (as of 2026-07-23) would sit at the top of the acceptable band and just into the elevated zone. Aggregate open interest at the longest tenors could not be confirmed to any dated corpus source, and the skeptic pass marked the whole instrument claim unverifiable — so the facts are recorded as unconfirmed rather than as evidence. I1 never blocks a pillar result, and because the framework's watchlist-only overlay applies only to a fits or lean-fit outcome, that flag is not set here. Context in Clock.

What a 3x-in-3-years would require

The tally's re-rating arithmetic is unavailable: Re-rating math unavailable because the applicable bar or normalized adjusted FCF is missing. Stock-based compensation is absent from the structured cash-flow feed, so fit_features.adjusted_fcf, adjusted_fcf_yield and yield_baseline are all not_computable, and the deterministic price at bar-yield could not be computed.

What the framework's own test requires is still statable as arithmetic. To put the adjusted yield at 8% on today's $9,836.9M market cap, adjusted free cash flow has to be about $787M against $406.2M in FY2026 — close to a doubling. Consensus does not concede that: consensus reported free cash flow of $1,133.0M in FY2027 rising to $1,328.0M in FY2029 produces an adjusted-implied yield of about 8.3% only in FY2029, and only if compensation falls to about 12% of revenue. The concession consensus would have to make is not to the cash-flow line but to the compensation line beneath it.

The base rates come from this name's own price history (Clock): the 2018 episode fell 44.9% over 82 days and round-tripped 350 days from the trough; the 2023 episode fell 43.2% over 267 days and round-tripped in 348 days; the 2021–22 episode fell 87.2% over 432 days and has never recovered its peak. The current episode is 61.0% deep, 444 days from peak to trough, and 12.7% off the low. Two of three comparable episodes recovered inside roughly a year; the deepest did not recover at all.

Contested and undetermined

P5 is contested. The trial produced two cited briefs, read by three independent judges in alternating order.

The temporary reading (judges at 0.68 and 0.63): through the drawdown window earning power grew rather than fell. FY2026 revenue was $3.2B, up 8%; ARR grew 8%; billings reached $3.4B, up 10%, with fourth-quarter billings above $1B for the first time in the company's history; and it was the first year of free cash flow above $1B [5]. The triggering billings miss was a disclosed early-renewal timing artifact that reversed inside the same fiscal year [11].

The permanent reading (judge at 0.42): the expansion engine reset durably and has stayed reset. Dollar net retention fell from 119–123% in FY2021–FY2022 to about 102% and has not returned; revenue growth compressed from 45.0% to 8.2%; remaining performance obligations stalled at $2.4B; FY2027 ARR is guided to only 8.25–8.75% growth [13]; and the AI-replication risk is verified verbatim in the current quarterly filing [14].

The split: judges at 0.68, 0.42 and 0.63; mean 0.577; the tally's trimmed figure 0.63; spread 0.26. Reading order moved the panel — the temporary-first mean was 0.68 against a permanent-first mean of 0.525, a gap of 0.155. That order sensitivity, together with the 0.26 spread, is why the result is recorded as contested rather than decided, and it is what drives the confidence tier to low.

Undetermined. One criterion is unresolved rather than contested: I1, where the named missing datapoint is a dated corpus options-chain and open-interest source (with an implied volatility surface) at the longest tenors. No criterion was returned as cannot-determine.

Provenance

Source: the run's deterministic fit tally, provenance block, and the skeptic refutation ledger.

Two independent model families voted the checklist blind to the reader's framework, and a fifth seat voted the same docket with the company name masked to test whether the result was being driven by prior knowledge of Docusign rather than by the evidence; it returned the same gate results with no probability divergence. Separately, every verdict-critical claim was re-derived from its cited pages by a skeptic — none was refuted, seven were narrowed (most consequentially, the buyback claim was cut back from a working repurchase engine to turning-point evidence, and the perpetuity gap was cut back to nothing on the adjusted basis), and one, the options-market evidence, could not be verified at all.

The falsifier ledger

Fifteen conditions stand as the what-would-change-this ledger, each reproduced as the tally records it, with its window where the text defines one and the direction it points.

The framework's own templates. These are the standing conditions the reader's system carries into every name.

• Cash flow. adjusted FCF or EBITDA declines where flat-or-better was underwritten — no window defined; standing. Direction: cuts against fit.
• Revenue. revenue declines for a third consecutive year — three consecutive fiscal years, rolling. Direction: cuts against fit.
• Allocation. capital allocation pivots to debt paydown over repurchases — no window defined; standing. Direction: cuts against fit.
• Share count. share count inflects upward — no window defined; standing. Direction: cuts against fit.
• Sector. the industry repricing cycle fails to materialize where industry-wide mean reversion was underwritten — no window defined; standing. Direction: cuts against fit.

The trial's flip conditions. These are the name-specific tripwires the panel wrote for the diagnosis (P5) — three from the first judge, three from the second, four from the third, carried into the tally as written.

• Consumption/envelopes-sent turn negative YoY or DNR breaks below 101% in FY2027 — the first observable signature of AI substitution actually impairing the eSignature base. (FY2027. Direction: cuts against fit.)
• FY2027 FCF falls below ~$1.0B or buybacks drop well below ~70% of FCF — the per-share compounding that carries the 8x valuation stops. (FY2027. Direction: cuts against fit.)
• IAM ARR growth stalls (fails to reach the guided ~18%) while total ARR decelerates below 8% — the replacement engine proves to be eSignature migration, not incremental demand. (no window stated. Direction: cuts against fit.)
• By Q4 FY2027, total ARR grows at least 11% organically, DNR exits above 108%, and revenue growth exceeds 10% excluding FX. (by Q4 FY2027. Direction: supports fit.)
• By FY2028, IAM exceeds 25% of ARR while total company ARR and RPO accelerate, showing IAM is incremental rather than mainly migrated eSignature dollars. (by FY2028. Direction: supports fit.)
• Management discloses large IAM renewal cohorts with materially above-company gross retention and DNR, replacing the current small-sample evidence. (no window stated. Direction: supports fit.)
• DNR breaks below 100% or consumption/envelopes-sent turn negative YoY — signals genuine AI substitution rather than timing, → permanent (no window stated. Direction: cuts against fit — the tally marks this one permanent.)
• FY2027 FCF falls below ~$1.0B or forward FCF consensus begins to be cut — absolute earning power actually eroding, → permanent (FY2027. Direction: cuts against fit — the tally marks this one permanent.)
• IAM fails to reach guided ~18% ARR and total ARR growth slips below 8% — growth engine fading not stabilizing (no window stated. Direction: cuts against fit.)
• IAM exceeds 25% of ARR with total ARR/RPO accelerating — confirms incremental, not migration, → strongly temporary (no window stated. Direction: supports fit — the tally marks this one strongly temporary.)

Data gaps

• Stock-based compensation is absent from data/financials/cash_flow.json for FY2017–FY2026, so fit_features.adjusted_fcf, adjusted_fcf_yield, yield_baseline, fcf_stability and float_retirement_years are all not_computable. Every adjusted-FCF figure on this page is an analyst derivation from the filed cash-flow statements (FY2026 p.106, FY2024 p.107, FY2022 p.112), not the deterministic feature; the feed should be repaired to ingest the SBC and acquisitions lines.
• fit_features.balance_sheet_class reads 'unknown' (debt and cash flagged missing for FY2026); the fortress classification here is derived from the filed balance sheet (p.102) and liquidity note (p.86), which show no funded debt and about $1.07B of cash and investments.
• fit_features.market_cap uses $10B as the reference; the exact break-even share price (about $47.8) is derived from feature inputs and is not stated in any filing.
• No company-reported e-signature market share exists. Only methodology-dependent third-party estimates (roughly one-third to two-thirds for Docusign) are available, so the P1 share evidence cannot be pinned to a filed primary source and is directional only.
• The year-10 dollar impact of AI substitution cannot be quantified from the corpus; it is bounded here by the exposed share of ARR (roughly 80% legacy) rather than a point estimate.
• Consensus publishes no SBC-adjusted free cash flow. The adjusted forward path uses assumed compensation of 17%, 15% and 12% of revenue for FY2027–FY2029 — a modeled scenario, not a vendor figure.
• CapIQ estimate-revision momentum reaches back only 180 days (to 2026-01-24), so the vendor record cannot show the consensus reaction in the days immediately after the June 2025 trigger.
• The two AI-driven de-rating legs (30 September 2025 and 3 February 2026) have no corresponding corpus filing. The run's web-research budget was exhausted, so attribution rests on harness web search plus the daily price feed.
• balance_sheet.json exposes only cash and equivalents ($602.4M at FY2026), not short-term or long-term investments or debt, so the net cash used for enterprise value was taken from the Q1 FY2027 quarterly liquidity note ($814.2M cash and short-term investments plus $209.9M long-term investments, no debt).
• Reported short interest, days-to-cover and borrow pressure are unavailable for DOCU (the feed returned no position rows), so the crowded-short and forced-buyer channel cannot be assessed.
• Aggregate and long-dated (18-month) option open interest could not be confirmed to a single dated citable source; only the 30-day and 180-day mean implied-volatility points are dated (2026-07-23).
• Exact future earnings dates for Q2–Q4 FY2027 are inferred from the reporting cadence (Q4 FY2026 reported 2026-03-17; Q1 FY2027 reported 2026-06-04), not from a published calendar.
• The absence of open-market insider purchases and the net-seller pattern come from structured Form 4 data rather than a citable filing PDF; the ownership levels are cited to the proxy.
• share_count_trend.cagr_10y_pct is not_computable — share counts exactly ten fiscal years apart are unavailable.
• P4c (dividend cover) is not_applicable: Docusign pays no dividend.