Fit

Does not fit the framework (P1 not met); contested: P2, P3a

Does not fit the framework. The year-10 durability gate (P1) resolves to not met, and under the framework that gate alone forces the verdict — nothing downstream offsets it. Confidence is medium: the P1 probability spread was 0.22 (at or below the 0.25 medium threshold) and two non-load-bearing criteria (P2 and P3a) came back contested across the two model families. No exclusion fired, the watchlist-only overlay does not apply, and the name-mask probe raised no prior-driven-risk flag. The business clears the universe screen and most of the self-help and yield-context checks; it fails on the one criterion the framework treats as decisive.

Universe and exclusions

The universe screen is clean. Accenture is an Irish-domiciled company whose Class A ordinary shares list directly on the NYSE under ACN — a primary US listing rather than the ADR wording the European route names, but a qualifying listing either way, and not a China-domiciled issuer [1]. Market capitalization at the 2026-07-31 close of $165.92 on 632.4M shares is about $104.93B, roughly 10.5x the $10B universe line and an order of magnitude above it even after the 17.97% single-session drop on 2026-06-18 [2]. Both universe criteria (U1, U2) are met with cross-family agreement.

No exclusion hits. The four disqualifiers were each checked and none fired:

Auto OEM (X1) — not met. Accenture sells consulting, technology and managed services to more than 9,000 clients; automotive is one of six sub-industries inside the Products segment, not a manufacturing business [3].
Consensus-saturated story (X4) — not met. The stock trades at about 1.51x FY2025 sales and roughly a 10.4% trailing FCF yield after the June-2026 fall — the opposite of an extreme multiple-to-sales darling; the counter is that forward consensus still embeds ~4–5% annual revenue growth that is largely acquisition-driven [4].
Promotion pattern (X2) — not met. Insider economic ownership is thin — the whole director-and-officer group holds ~0.02% of shares and the CEO about $0.8M outright against $19.9M realized on vesting in FY2025 alone [5] — and the committee lowered the relative-TSR entry threshold from the 40th to the 30th percentile and swapped a ~16-name peer group for the S&P 500 for Jan-2026 grants [6]. But an X2 hit requires both prongs, and the second — a repeated promise-versus-delivery gap — is not robustly evidenced: the stated shareholder-return dollar floor was met or beaten every year on record and compensation-actually-paid fell from $37.4M (FY2021) to $16.6M (FY2025) as two relative-TSR grants paid 0% [7]. One prong only, so no hit.
Structural decline (X3) — not met. Revenue rose every year through FY2025 (to $69.67B, +7.4%), with zero consecutive declining years; the drawdown was a price move, not a revenue one. A genuinely structural mechanism exists — the FY2025 annual report names clients' in-house global capability centres as a competitor category with no capital barrier, and HCLTech quantifies 2–3% portfolio-level AI price deflation [8] [9] — but it is so-far forward-looking, not a realized non-mean-reverting loss, so the exclusion does not fire. This mechanism overlaps the diagnosis (P5) trial and the skeptic weakened the strongest version of the claim (see Provenance).

China dependence (S1) is a sensitivity flag only and is immaterial here: the US is 45% of FY2025 revenue, no other single country reaches 10%, and Ireland is ~1%, so any Greater China exposure sits below the 10% line — the caveat being that Accenture does not separately quantify China, so the figure can only be bounded, not pinned [10].

Pattern match

The framework's reader contract carries four recognition setups: cyclicals at the bottom (large banks); a high dividend-plus-FCF yield where the business is not going away; healthcare/insurance forecasting errors that reprice; and quality tech monopolies or duopolies on a fear dip. Accenture most resembles the fourth — a large, high-quality technology-services name sold off on a specific, testable fear (AI disruption of the consulting labour pyramid, plus a federal-procurement slowdown). It also brushes the second, on a ~10.4% unadjusted trailing FCF yield.

But it fits none of the four cleanly, and this framing lines up with the tally rather than against it. The fourth pattern's defining check is a monopoly or duopoly structure; Accenture's own 10-K describes "the markets in which we operate are highly competitive" with no monopoly and no regulatory entry barrier — the fragmented labour-arbitrage structure is exactly what the year-10 gate turns on [11]. The second pattern turns on a high dividend that will not be cut; here the dividend yield is only ~3.6% trailing (~3.9% forward), below the ~4% materiality line, and the framework's own adjusted FCF yield sits well below the bar (below). This is a quality-tech-on-a-fear-dip in silhouette that fails the pattern's moat check.

The pillar ledger

Year-10 durability (P1) — not met · the gate

This is the criterion the verdict turns on, and it split across the model families: the two Claude seats and one Codex seat read it not met, one Codex seat read it met (3 not met / 1 met), trimmed-mean probability 0.62, spread 0.22, cross-family agreement false, masked verdict not met. Genuine doubt on this gate resolves to not met by the framework's own rule.

The supporting record is strong on its face: revenue compounded from $34.8B in FY2016 to $69.7B in FY2025 (+100.2%, ~8.0% CAGR) with no consecutive-decline years and the high-single-digit-decline disqualifier false, on capex of only ~0.86% of revenue and a managed-services book near half of revenue [12] [13]. The strongest surviving counter-fact, carried in the same claim and skeptic-survived: the model rests on human-hour labour arbitrage in a highly competitive, fragmented market with no monopoly structure, management explicitly "does not track standard measures of unit or rate volume," FY2026 local-currency guidance was trimmed (3–5% to 3–4%) and new bookings fell ~1% — so GenAI-driven delivery productivity could compress the revenue base even as it lifts near-term margin [14]. Because the year-10 adjusted-FCF leg cannot be quantified from features (adjusted FCF is not_computable — see Data gaps), the FCF half of the gate rests on disclosed near-zero capital intensity and ~139% net-income-to-FCF conversion rather than a computed series, which is part of why the gate does not clear with very high conviction. Full treatment: The 800,000-Person Payroll and Who Sets the Unit Price.

FCF consistency (P2) — contested

Contested across families: the two Claude seats read it met, the two Codex seats cannot determine (masked verdict met). On raw FCF the series is exceptionally stable — ten years, no negative years, the rolling five-year average climbing monotonically from ~$5.54B (FY2016–20) to ~$9.14B (FY2021–25) and never declining in any window [15]. The counter, and the reason two jurors could not determine: the framework's measure is adjusted FCF (FCF minus SBC minus a trailing-5-year acquisition average), which the deterministic feature returns not_computable because SBC is absent for every year in the data feed; once the ~$2.1B FY2025 SBC and a lumpy ~$3.6B/yr acquisition average (ranging $1.5B FY2025 to $6.6B FY2024) are subtracted, the level roughly halves and inherits acquisition lumpiness the raw line masks [16] [17]. Full treatment: Spending 1.8x Free Cash Flow.

Dislocation and yield (P3) — mixed; P3a contested, yield below bar

Identifiable event (P3a) — contested. Claude seats met, Codex seats not met (2/2). There is a dated adverse mechanism but no clean company trigger: the nearest event to the June-30 trough was the 18-Jun Q3 FY2026 print, an EPS beat ($3.80 vs $3.70) with a slight revenue miss, a Q4 range widened to 1–5% on a ~$100M Middle East impact and two managed-services deals slipping — with management "pleased with how we are executing" [18]. The fall reads as a fear-driven multiple de-rating against roughly-flat fundamentals (consensus EPS held; only price targets were cut) tied to a named mechanism — AI disruption plus federal exposure — which management explicitly rejects, framing AI as "a tailwind for us and our industry as it scales" [19]. The upstream federal slowdown was first disclosed on the Mar-2025 Q2 FY2025 call and is guided to sunset and return to growth in Q4 FY2026 [20] [21].

Capitulation (P3b) — met. Reported as met on the drawdown depth (-38.2% peak-to-trough on the run's price file). The counter is that the capitulation volume multiple was refuted by the skeptic: the feature's 2.36x-vs-2x reference reproduces only against a single pre-peak row on a partial 84-session file and falls to ~1.94x on all available rows, so P3b rests on the price fall rather than a verified 2x volume spike (see Provenance).

Yield vs bar (P3c) — not met. All four seats not met, cross-family agreement. The balance sheet is net cash (~$6.34B: $11.48B cash and equivalents against $5.15B total debt), which sets the fortress reference line at 8.5% [22] [23]. Adjusted FCF yield computes to 4.90% on FY2025 figures (adjusted FCF $5.14B = FCF $10.87B − SBC $2.09B − $3.64B trailing acquisition average, over the $104.93B market cap), about 360 bps below the bar; the three-year-average adjusted yield is 4.02%, about 448 bps below, with no jump toward the bar [24] [25]. The counter-fact, in the same treatment: on an unadjusted basis FCF yield is ~10.4% and clears the bar — the entire shortfall is the SBC and the near-$3.6B/yr acquisition drag the definition mandates.

Forward path (P3d) — not met. All four seats not met, trimmed-mean probability 0.245, spread 0.08. The consensus_forward_yield anchor shows street FCF yield rising from 10.68% (FY2026) to 12.15% (FY2029), above the 8.5% bar — but that anchor is unadjusted; re-deriving it on the playbook definition (subtracting ~$2.2B SBC and a trailing acquisition average that FY2026's guided ~$9B in ventures and acquisitions pushes toward ~$4.6B) pulls the adjusted forward yield to only ~4.2%, so consensus does not underwrite the adjusted bar within three years [26]. Full treatment: Priced at $165.92.

Balance sheet and self-help (P4) — met across the board

All three self-help criteria are met with cross-family agreement.

Outlast and allocation headroom (P4a) — met. The balance sheet is net cash (~$6.33B) with covenant-free senior notes laddered 2027–2034, so no refinancing wall forces debt paydown ahead of buybacks [27] [28]. Counter-fact: FY2026 is the first year the company plans to spend well above one year's cash — the skeptic weakened the claim's ~1.8x figure to at least 1.66x of the $11.15B guided FCF midpoint (~$18.5B of M&A plus shareholder returns), with cash drawn from $11.5B to $10.2B over nine months and a second debt raise flagged [29]. Headroom is being consumed by choice, not distress.

Repurchase engine (P4b) — met, no hard fail. Diluted share count fell from 645.91M (FY2021) to 632.44M (FY2025), −2.1% (feature share_count_trend.rising = false), so the rising-share-count hard fail does not trip; FY2025 executed purchases were $4,619M [30]. Counter-fact: the base shrank only 2.1% despite $21.3B of gross purchases because ~$8.97B of SBC and employee-plan issuance ran the other way, and $776M of the FY2025 outlay was tax-withholding that does not reduce authorization — so the buyback functions largely as a dilution offset [31] [32].

Dividend cover (P4c) — met. The dividend (~3.6% trailing / ~3.9% forward, just under the ~4% materiality line) is covered ~2.9x by FY2025 FCF and ~47% of net income, raised ~10% a year on an unbroken record [33]. Counter: the FY2025 FCF anchor was lifted ~65% by a $1,519M non-recurring accrued-payroll swing that management guides back toward 1.2x conversion, so normalized cover is thinner though still above 1x. Full treatment: Spending 1.8x Free Cash Flow.

Diagnosis (P5) — met; carried from the adversarial trial

P5 is met, on a trial-ruled probability that the impairment is temporary of 0.63 (mean 0.613, inter-judge spread 0.07, not contested; three judges at 0.64 / 0.57 / 0.63). The temporal signature is split: the federal drag (~8% of revenue, ~1pt FY2026 growth drag) is procurement-driven and management guides it to anniversary and return to growth in Q4 FY2026 — clearly temporary [34] [35]. Against it sits AI price deflation on a fixed-price book that is over 60% and rising — a candidate-permanent erosion with no comparably sized cost pool to absorb a client-side concession [36]. The strongest counter to the temporary read, carried in the same treatment: the permanent leg is unmeasured — management cannot separate price from volume and does not disclose the fixed-price book's duration — and the trial's own quote-check caught the temporary side overstating that management "raised FCF guidance," when FY2026 FCF was held flat at $10.8–11.5B and it was capital return, not FCF, that was raised [37]. A 0.63 temporary reading is above coin-flip but well short of the ~0.65–0.70 the framework's tiered fit thresholds ask for. Full treatment: Priced at $165.92.

Instrument context (I1) — not verifiable

All four seats returned not verifiable. Listed ACN LEAPS beyond the 18-month target do appear to exist (a Jan-2028 and a Dec-2028 expiry alongside a 2027 chain) and 30-day implied volatility of ~52.4% would sit within the ~55 acceptable reference line — but both readings come only from external option-chain and volatility URLs the corpus phase cannot verify, and no dated citable local source exists (see Data gaps). Because I1 is fact-only and never blocks the pillar verdicts, this does not change the outcome; it would have driven the watchlist overlay only had the verdict been fits or lean_fit, which it is not.

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

The tally records no re-rating arithmetic: "Re-rating math unavailable because the applicable bar or normalized adjusted FCF is missing." The applicable bar is 8.5% (fortress) and adjusted FCF is not_computable from features, so the deterministic price-at-bar figure the framework would normally publish is absent.

What can be stated, from the surviving yield claims rather than the tally: at the FY2025 adjusted FCF of $5.14B, the market cap consistent with the 8.5% bar is $5.14B ÷ 0.085 ≈ $60.5B — about 42% below the current $104.93B ($165.92/share) — so on today's adjusted cash flow the framework's yield bar is cleared by price falling, not rising. To instead clear the bar at the current market cap, adjusted FCF would have to reach ~$8.92B (from $5.14B), which needs the ~$3.6B/yr acquisition drag to fall toward zero while FCF holds near $11B — the opposite of the raised FY2026 ~$9B acquisition guidance. Consensus underwrites neither path on the adjusted definition. The episodic base-rate context the framework draws from prior dislocation recoveries is not computed in this run (the tally's re_rating_math is null); the valuation treatment is in Priced at $165.92.

Contested and undetermined

Two criteria were contested; nothing was returned as cannot-determine.

P2 (FCF consistency) — contested. Claude seats met on the stable raw-FCF series; Codex seats could not determine because the framework's adjusted-FCF measure is not_computable (SBC absent for every year). Both readings rest on the same fact — raw FCF is stable, adjusted FCF cannot be built deterministically.
P3a (identifiable event) — contested. Claude seats met (a dated, named macro-fear mechanism — AI plus federal — with a fear-driven de-rating against flat fundamentals); Codex seats not met (the nearest dated company event was an EPS beat, not a guidance or earnings cut of the size the fall implies).

Provenance

Item	Value
Jury seats	a = claude, b = claude, c = codex, d = codex; masked = claude
Families	claude, codex (2 seats each)
Trial (P5) order stability	temporary-first mean 0.64, permanent-first mean 0.60, gap 0.04
Name-mask probe	gate criteria differ: none; max probability gap 0.075; prior_driven_risk: false
Skeptic protocol (full)	20 claims taken to full protocol — survived 10, weakened 7, refuted 1, unverifiable 2
Skeptic triage	18 further claims passed cheap triage (38 claims seen in total)

Source: derived from the run's fit tally, jury dockets and refutations ledger.

The verdict was pressed hard. Two independent model families each held two seats and reached the same gate call on P1 (not met) and the same masked-name outcome, with no gate flip when the company name was masked and only a 0.075 probability gap — so the answer is not an artifact of model priors. One claim was outright refuted (the capitulation volume multiple, which failed to reproduce off the partial price file) and seven were weakened rather than accepted as written, which is why several supporting figures here carry the skeptic's trimmed version rather than the pillar author's original.

The falsifier ledger

These are the standing what-would-change-this conditions carried from the tally; thresholds, directions and test windows are stated where the framework defines them.

Framework templates:

adjusted FCF or EBITDA declines where flat-or-better was underwritten
revenue declines for a third consecutive year
capital allocation pivots to debt paydown over repurchases
share count inflects upward
the industry repricing cycle fails to materialize where industry-wide mean reversion was underwritten

Name-specific conditions:

Bookings/book-to-bill stay below 1.0 for 2+ consecutive quarters with the >=$100M deal count falling YoY (next test Q4 FY26).
ACN's own price realization / revenue-per-unit visibly declines, or peers' 2-3% AI deflation shows through in ACN organic pricing at renewal.
Consensus forward FCF ($11-13B FY26-FY29) is cut materially rather than held/raised.
Federal fails to return to growth in Q4 or ex-federal organic LC growth turns negative, showing the shock spread beyond the 8% pool.
Two consecutive quarters of book-to-bill below 1.0 with $100M-plus client bookings down year over year.
FY2027 organic local-currency growth excluding acquisitions and federal anniversary effects is guided below 2% or revised down after Q1.
Disclosed or peer-implied AI price/revenue-per-delivery-unit deflation exceeds productivity capture and drives margin compression.
AFS does not return to growth in Q4 FY2026 and further price/scope cuts or terminations persist through FY2027.
Bookings/book-to-bill roll over for 2+ quarters with the $100M-deal count declining YoY, showing demand (not just federal) is breaking.
Organic growth ex-federal AND ex-inorganic (i.e., stripping the ~1.5-2% acquisition contribution) stalls near or below zero, exposing bought growth masking a dead legacy engine.
Price realization/margin compresses as peers' 2-3% AI deflation shows up in Accenture's numbers, rather than being captured as productivity (margin already +20bps argues against this today).
Consensus/managed forward FCF is cut materially instead of held or raised.

Data gaps

What the run could not answer, from the tally's list:

Adjusted FCF and adjusted FCF yield are not_computable in features.json (SBC missing for FY2016–FY2025 and no complete 5-year acquisition window), so the year-10 adjusted-FCF leg of P1 cannot be quantified from features; the FCF-leg durability case rests on disclosed near-zero capital intensity and ~139% net-income-to-FCF conversion rather than a computed adjusted-FCF series.
Accenture does not separately disclose a Greater China revenue or asset figure, so S1 China exposure can only be bounded below 10%, not quantified exactly.
balance_sheet_class is 'unknown' (debt/cash missing for FY2025 in the feed), so no leverage-based durability cross-check is available from features; the fortress classification was reconstructed from the filed 10-K (cash $11,478.7M, total debt $5,148.7M → net cash +$6.33B).
The features engine recorded acq_5y_avg as 0.0 for FY2020–2025, contradicting the primary filings' $1.5B–$6.6B of annual business acquisitions; any downstream use of that field would overstate adjusted FCF.
X3 realization: no third-party market-share or price/realization series for Accenture exists in the corpus, and the company does not disclose the duration profile of its >60% fixed-price book, so the timing and magnitude of any AI-driven repricing cannot be confirmed or refuted in Accenture's own realized numbers.
Terminal growth — the decisive input splitting temporary from permanent — is unmeasured: the corpus carries no third-party organic-growth or Accenture-specific AI-price-deflation series, so the permanent-case FCF haircut is an assumption band, not a measured value.
Price history is partial: the run's price file holds only 84 sessions (2026-04-01 to 2026-07-31), so the peak used by the capitulation gauge is a four-month high ($201.33), not a true multi-year peak; the 52-week high $291.09 and the −43.0% damage figure are sourced from the estimates feed, and the 2.36x volume multiple was refuted against this partial file.
I1 liquidity: per-contract open interest and bid/ask spreads for the qualifying ACN 2027/2028 LEAPS, and a dated implied-volatility reading, could not be obtained from a citable local source (corpus/web-search credit was exhausted mid-pass); liquidity and IV adequacy are inferred, not measured.
No published target leverage ratio: management commits only qualitatively to "a low net leverage ratio," so the ceiling on FY2026 debt-funded buybacks/M&A is not quantified.
The X2 promise-versus-delivery prong is not robustly evidenced: the shareholder-return dollar floor was met or beaten every year on record, so the exclusion's second prong rests only on comp-design resets, not a documented pattern of guided-then-missed results.

Checked and unremarkable

No scout memo returned a routine or empty verdict — all eight (accounting and cash quality, business economics, capital allocation, competition and moat, history and track record, industry, people and governance, valuation and expectations) were load-bearing and their evidence is threaded into the pillar treatments above.

Playbook version

Framework: fcf-dislocation, version 4 (spec fit-spec version 2), as frozen for this run.

Generated Aug 1, 2026 13:32 UTC