Financebotresearch desk研究台

ADBE › analyze

ADBE · Analyze from before

WATCH Technology

Date: 2026-09-10 | Price: $248.83 close / $243.50 after-hours post-print | Sector: Technology / Software — Application Baseline: Output/Stocks/Technology/ADBE/analyze-2026-08-04.md — WATCH [6.5], FV $290–360, entry $225–260, trim 13x fwd Second-prior file read: analyze-2026-06-04.md (2026-06-04, ACCUMULATE [7.5]) — used only to see which claims have survived two consecutive tests.

Verdict unchanged in label, up modestly in conviction: WATCH [6.8] (was [6.5])

Both events the baseline pre-registered fired today, but neither fired cleanly. Adobe named an internal CEO (Anil Chakravarthy, effective Dec 1) — the literal condition the baseline set for an upgrade toward [7]. But the pick came bundled with something the baseline did not model: David Wadhwani — the OTHER internal name the baseline flagged, and the head of the ~75%-of-revenue Creative business, and "the public face of the AI strategy" in the August file's own words — is leaving Adobe entirely, with no successor named. The market's read was not calm continuity; the stock fell 6–7% on the CEO news alone. Separately, Q3 beat on revenue and non-GAAP EPS, and total ARR growth (11.2% YoY) and AI-first ARR growth (>150% YoY) both cleared every numeric break-trigger the baseline wrote for today — so the bear scenario ("revenue beat, ARR miss → [5.5]") also did not fire. Two genuine positives, one genuine new negative, net a small conviction gain, not the full move to [7.0] the simple binary implied.


What this updates

Baseline: analyze-2026-08-04.md, WATCH [6.5], written the day before Q3 FY26 was scheduled and five months into an open CEO search. It pre-registered two explicit gates for today: (1) "An internal pick — Wadhwani/Chakravarthy — is the opposite [of an external cost-cutter] signal and would be an upgrade trigger [toward 7]"; (2) "a revenue beat paired with an ARR miss is the specific pattern that would confirm the bear case and take this to [5.5]." It also rewrote five numeric break-triggers keyed to Q3/Q4 data (total ARR growth, AI-ARR growth, gross margin, Creative & Marketing subscription growth) after finding the previously-recorded trigger metric ("Digital Media ARR") had been discontinued.

Event list since 2026-08-04: 1. 2026-09-03 (after close) — Adobe names Anil Chakravarthy (President, Customer Experience Orchestration; joined Adobe Jan 2020) as President & CEO effective 2026-12-01; Narayen moves to Executive Chair. David Wadhwani (President, Creative & Productivity — ~75% of Adobe revenue) announces he is leaving the company, effective unspecified, no successor named. Stock fell ~6–7% on 2026-09-04. 2. 2026-09-10 (after close) — Q3 FY2026 reported. Revenue $6.76B (+13% YoY), non-GAAP EPS $6.13 (+15%), GAAP EPS $4.62. Total Adobe ending ARR $27.50B, +11.2% YoY. AI-first ARR >$650M, +150%+ YoY. FY26 guide raised (revenue, both EPS lines); Q4 revenue guide $6.80–6.85B came in below the ~$6.96B sell-side consensus, and the stock fell a further ~2.1% after-hours on top of a ~2.3% regular-session decline (partly broad-market, per same-day coverage of a market-wide oil/rate-hike selloff). 3. No new development on the permanent CFO search — Steven Day remains interim; sources describe the company as having "more than a year" to find a successor, i.e. still open, now three months longer.

Both events were fully within the baseline's own recheck window (recheck: 2026-09-11, triggers: [Q3 FY26 print Sep 10, CEO successor named, permanent CFO named]) — this is exactly the scheduled test, not an off-cycle re-analysis.


The delta ledger

Lead with the claims that moved. Sources for every new figure: Adobe's own Q3 FY26 earnings release, SEC 8-K Ex-99.1 (sec.gov/Archives/edgar/data/796343/000079634326000147/adbeex991q326.htm — fetched directly, primary source, not vendor-derived); Adobe/CNBC/Bloomberg on the CEO transition; Investing.com, TipRanks, Jefferies-sourced coverage on the market reaction; fin.py --news for the live snapshot.

🔄 SUPERSEDED

# Claim (baseline, 2026-08-04) New fact What replaced it
1 CEO succession: no successor named, month 5. Front-runners per press: Wadhwani or Chakravarthy. Chakravarthy named 2026-09-03, internal, effective 2026-12-01. Narayen → Executive Chair. The "who" question is answered. But see the close-call section below — this claim's replacement is more complicated than "trigger fired."
2 FY26 ARR growth guide: 10.2%, cut from a 12.5% Q2 exit rate — "the single most important new fact since June." Guide held at 10.2% for FY26; Q3 actual ending ARR ran slightly ahead of that pace at +11.2% YoY. No further cut. The bear scenario ("ARR miss") did not materialize. The guide did not move down again — it was reaffirmed against an actual print that beat it. This is the single most important fact in this file.

🆕 NEW (no baseline counterpart)

# Finding Detail Direction
3 Wadhwani's departure Head of Creative & Productivity (~75% of Adobe revenue), explicitly named in the August file as "the public face of the AI strategy," is leaving Adobe entirely. No successor named for Creative Cloud leadership. Undercuts the "internal pick = continuity" read the baseline assumed
4 Market reaction to the CEO news: −6% to −7% (2026-09-04), not the calm de-risking the baseline anticipated Jefferies: expected Wadhwani as "the rational choice" given his revenue scope; flags "more departures and org changes are likely" Directly contradicts the baseline's assumption that any internal pick would read as continuity
5 Q4 FY26 revenue guide ($6.80–6.85B) below sell-side consensus (~$6.96B) The proximate cause of the post-earnings share-price decline despite a clean beat-and-raise on the quarter just reported Negative, near-term
6 Net-new ARR / RPO deceleration flagged by analysts on the call. RPO $22.16B, +8% YoY — running below total ARR's +11.2%. Management attributed the pattern to the freemium strategy and normal Q4-weighted seasonality. (Secondary-source claim of "$560M net-new ARR ex-Semrush, −3% YoY" could not be corroborated from the primary release and is flagged, not adopted — see gaps.) A stock-measure (total ARR, +11.2%) can still show YoY growth while the flow measure (net-new ARR, RPO) decelerates, because the growing base compounds even as new bookings slow. This is a genuine forward-looking yellow flag the baseline's threshold-based triggers were not built to catch. Negative, needs the Q4/FY27 print to resolve
7 Gross margin's first quarter of measured compression: 88.7% GAAP (Q3) vs 89.3% (FY25 annual) Still above the 88% break-trigger, by 0.7pt — one quarter of cushion, not two Mildly negative, watch the next print
8 Barclays (Kalia) flags FY27 guide risk from a double leadership reset — a new CEO and a still-unnamed CFO both setting first-year targets is a classic kitchen-sinking setup Raised PT to $295 (Hold) anyway Two-sided: a lower bar for FY27 beats, but a real guide-cut risk at the Dec 9 print

🔁 REFRESHED

# Claim Old (Aug 4) New (Sep 10) Note
9 AI-first ARR >$500M, "tripled YoY" (~200%+) >$650M, "+150%+ YoY" Dollar base grew ~30% QoQ; YoY rate decelerated 200%→150%+. Still clears the baseline's own 100%-YoY break-trigger with room to spare — not a break, but the first deceleration in this metric since Adobe began reporting it
10 GAAP/non-GAAP EPS gap FY26 guide $17.90–18.00 GAAP vs $24.35–24.45 non-GAAP (+36%) FY26 raised guide $18.12–18.17 GAAP vs $24.45–24.50 non-GAAP (~+35%) Same finding, same magnitude — the honest multiple is still the GAAP one
11 Analyst consensus / mean target Mean $269.61, median $250 — below spot ($253.81) fin.py live pull: mean $277.75, Recommendation HOLD. Barclays raised to $295 (Hold) post-print; RBC $315 (Outperform, pre-print); Stifel raised $200→$225 (Hold); Mizuho raised $245→$260 (Neutral) Consensus PT drifted up and is now roughly in line with spot rather than meaningfully below it — a mild positive shift in sell-side positioning, still capped at Hold
12 Q3 FY26 guide (was forward-looking in Aug) Revenue guide $6.67–6.72B; non-GAAP EPS $6.05–6.10 Actual: $6.76B revenue (beat), $6.13 non-GAAP EPS (beat) Clean beat on both lines, as guided
13 FY26 full-year guide Revenue $26.50–26.60B; non-GAAP EPS $24.35–24.45; GAAP EPS $17.90–18.00 Raised: $26.576–26.626B; $24.45–24.50 non-GAAP; $18.12–18.17 GAAP Second consecutive raise; both above the prior guide
14 Creative & Marketing Professionals subscription growth 11–13% (sources conflicted, as-reported vs cc) $4.65B, +13% YoY / +12% cc — cleanly reported, no conflict this quarter Clears the baseline's own <8% cc break-trigger comfortably
15 CFO vacancy Steven Day interim since 2026-06-15; assessed BENIGN Still interim, three months on; press describes "more than a year" runway to find a successor Not worse in kind, but the vacancy has now persisted a full extra quarter with no narrowing — see close-call discussion

✅ CARRIED (re-tested or unchallenged, compact form)

  • Elite unit economics, largely intact. Q3 GAAP gross margin 88.7%, GAAP operating margin 34.8%, non-GAAP operating margin 44.0%, GAAP net income $1.83B, non-GAAP net income $2.42B, OCF $2.52B (a Q3 record). Directionally consistent with the FY25 89.3%/36.6%/41.4% profile the baseline scored EXCELLENT, with the gross-margin data point trending down rather than up for the first time (see REFRESHED #7 in spirit, ledgered there).
  • Buyback pace continued. 9.5M shares repurchased in Q3 (vs ~8.5M in Q2). Consistent with the $25B authorization running through April 2030.
  • Per-share metrics compounding faster than the total company (rev/share, FCF/share CAGRs ~13–14% vs ~10.5% revenue CAGR) — no new annual data this pass (FY26 closes at Q4); mechanism unchanged.
  • Moat structural map — Document Cloud/Acrobat no credible threat; Creative professional stable-eroding; Creative prosumer/SMB heavily attacked (Canva/Affinity free-forever); Firefly Foundry + GenStudio the one genuinely strengthening moat source. Not independently re-swept this pass — no new competitive intelligence gathered on Canva/Affinity/Figma — carried on the strength of the baseline's own qualitative work, which this pass had no evidence to contradict.
  • Insider signal: director David Ricks' $1.95M open-market buy at $194.51 (2026-06-25) — historical fact, unchanged. No new insider transaction data pulled this pass around either the Sep 3 CEO news or the Sep 10 print — flagged under gaps below, not silently assumed unchanged.
  • FTC/DOJ settlement closed, Semrush ARR contribution (~$480M) — both closed, non-recurring facts, no new developments expected or found.
  • Sector-wide AI-fear de-rating context (NOW, CRM, INTU also down) — not re-verified this pass; carried as an unexamined assumption from the baseline.

⏳ UNTESTED

  • TTM free cash flow / FCF-per-share, refreshed for the three FY26 quarters now reported. fin.py only carries annual statements through FY25; Q1–Q3 FY26 quarterly cash-flow detail was not pulled this pass. The baseline's FCF-multiple valuation model (Model 2, the co-primary valuation basis) could not be honestly re-derived and is carried forward rather than recomputed — see fair-value discussion below.
  • Diluted/basic share count. fin.py's live SharesOut field reads 397.50M — identical to the figure quoted in the 2026-08-04 baseline, despite ~18M shares repurchased across the two intervening quarters (8.5M Q2 + 9.5M Q3). This looks like a stale or infrequently-refreshed vendor field, not a genuine flat count. Not corrected this pass (no primary 10-Q share count pulled) — do not quote per-share figures built on this field without an independent check next pass.
  • The secondary-source claim of "$560M net-new ARR ex-Semrush, −3% YoY." Surfaced by one low-confidence secondary aggregator, not corroborated in the primary SEC release or in the Investing.com transcript excerpt obtained. Recorded as a direction (net-new ARR/RPO decelerating is corroborated by the primary RPO figure), not adopted as a number.
  • Competitive landscape refresh (Canva/Affinity/Figma) — no new data pulled; the Aug 4 moat section stands untested, not reconfirmed.

How the close calls were decided

The central call: does the Chakravarthy appointment satisfy the baseline's upgrade trigger?

The literal test says yes. The baseline wrote: "An internal pick — Wadhwani/Chakravarthy — is the opposite [of an external cost-cutter] signal and would be an upgrade trigger." Chakravarthy is one of the two names literally written down, and he is internal.

The substantive test says only partially. The baseline's underlying reasoning for why an internal pick would reassure was continuity — someone who already knows the AI pivot executing it, rather than an outsider who might reset strategy or cut costs. That reasoning assumed the other half of the leadership bench stayed in place. Instead, Wadhwani — who ran the segment carrying 75% of revenue and who the baseline itself called "the public face of the AI strategy" — is leaving entirely, with no successor named. The market did not treat this as continuity: the stock fell 6–7%, and the sell-side reaction (Jefferies expecting more departures) reads as increased organizational uncertainty, not decreased. Weighing the two: the CEO seat itself is resolved, which removes one specific unknown (who ultimately sets strategy) and gives a firm date (Dec 1). But the operational leadership of the moat's most-attacked, largest segment is now more open than it was on Aug 4, not less — a new vacancy replaced an old one rather than the old one simply closing. Net: a real but partial de-risking, not the clean upgrade signal the binary trigger implied. This is why conviction moves to [6.8] rather than the pre-registered [7.0].

The second call: did the ARR test pass or fail?

By the letter of the baseline's own break-triggers, it passed cleanly. Total ARR growth 11.2% YoY is above the 10% floor and above the 10.2% full-year guide itself; AI-first ARR growth 150%+ is well above the 100% floor; gross margin 88.7% is above 88%; Creative & Marketing subscription growth 12% cc is above 8%. None of the five numeric triggers the baseline wrote specifically to adjudicate this day fired.

But a new, more granular concern surfaced that the triggers were not built to catch: RPO growth (+8% YoY) running below total ARR growth (+11.2%), and analyst questions on the call about slower net-new ARR, both point at deceleration in the flow of new business even as the stock of ARR keeps compounding on a larger base. Management's explanation (freemium routing + normal Q4-weighted seasonality) is plausible and matches Adobe's own stated strategy, but it is management's explanation, not an independently verified one. Resolution: score the print as a genuine pass against the pre-registered thresholds — that is what those thresholds were for, and gaming them after the fact by inventing new ones would defeat the purpose of pre-registration — but name the net-new ARR deceleration explicitly as the thing to watch at Q4, since it is the leading indicator that would show up before the total-ARR trigger ever breaches.

The third call: does the gross-margin data point matter yet?

One quarter at 88.7%, down 0.6pt from the FY25 annual figure, with 0.7pt of cushion before the break-trigger fires. Not enough to call it drift with any confidence — a single quarter's margin can move for revenue mix reasons alone — but it is the first quarter in this file's history where the number moved down instead of up or sideways, and it deserves to be named rather than folded silently into "carried."


Thesis persistence and conviction delta

Structural + Trend claims: of the claims ledgered above, essentially all Structural claims (moat map, cash-machine unit economics, capital allocation discipline) and Trend claims (per-share compounding, buyback pace, FCF-margin profile) survived as CARRIED or REFRESHED-in-direction. Zero were RETRACTED. High persistence — consistent with a business whose reported quarter did not break anything, which is the same conclusion the baseline reached about the prior period.

State and Judgment claims did the moving, exactly where the baseline expected them to: the CEO question (SUPERSEDED), the ARR guide question (SUPERSEDED, in the thesis's favor), and one new State-level concern (net-new ARR/RPO deceleration) that neither report anticipated in its current form.

Conviction: [6.5] → [6.8]. The rows that drove it, explicitly:

Direction Row(s) Weight
⬆ CEO uncertainty resolved with a named, dated, internal successor (row 1) Moderate positive
⬆ FY26 ARR guide held rather than cut again; Q3 actual ran ahead of guide (row 2) Moderate positive
⬆ All five numeric break-triggers cleared, none breached (rows 9, 14, and the untriggered gross-margin/ARR rows) Moderate positive
⬇ Wadhwani's exit removes the operating head of 75% of revenue with no successor named (row 3) Moderate negative — offsets most of the CEO-resolution credit
⬇ Market's own read of the CEO news was negative, not reassuring (row 4) Mild negative
⬇ Q4 guide below street consensus; net-new ARR/RPO deceleration (rows 5–6) Mild negative
⬇ CFO vacancy now a full quarter longer with no narrowing (row 15) Mild negative, unchanged in kind

The positives are more thesis-critical (they were the specific, pre-registered tests), so the net move is positive — but the new Wadhwani/Creative-succession risk is real enough that this is not the clean [7.0] the simple binary predicted. A small, well-supported move, not a large one.


What is genuinely new

  • A third leadership vacancy where the baseline tracked two. On Aug 4 the open seats were CEO and CFO. Today the CEO seat has a name and a date, but Creative Cloud/Productivity leadership — arguably the more consequential seat, given that segment carries the moat's actual attack surface — is now open with no name attached, alongside a CFO search that has not narrowed.
  • The market's verdict on the CEO pick is already in, and it disagrees with the baseline's binary framing. A −6-7% reaction to an internal appointment is not what "internal pick = upgrade trigger" was written to describe.
  • A flow-vs-stock ARR distinction the baseline's trigger design did not anticipate. Total ARR (a stock measure) can keep clearing a growth floor even while net-new ARR and RPO (flow measures) decelerate, because a large compounding base absorbs slower bookings for several quarters before the total growth rate visibly bends. This is worth carrying into the next pass as the earlier warning indicator, ahead of the total-ARR trigger itself.

Updated verdict — WATCH, conviction [6.8]

Run the first-principles table:

Question Answer
Is this a good business? Yes, essentially unchanged from August. 88.7% gross margin (mild first-quarter compression, not yet a trend), 34.8% GAAP / 44.0% non-GAAP operating margin, record Q3 OCF, continued buybacks. Moat map unchanged and not re-tested this pass.
Has the market priced it? About the same as August. Price is flat-to-slightly-down ($253.81 → $243.50–248.83) while FY26 EPS guidance rose modestly, so the stock is very slightly cheaper on an as-guided basis than five weeks ago, not more expensive.
Verdict cell Unchanged: good company, moderately cheap, watch-not-act — the CEO answer removed one specific overhang but replaced it with a different, more segment-specific one.

Valuation — re-derived, not carried

FCF-multiple model (Model 2 in the baseline) could not be honestly re-run this pass — TTM FCF/share needs Q1–Q3 FY26 quarterly cash-flow data not pulled (see gaps). What follows leans on the forward-earnings model, cross-checked against the baseline's FCF-derived range, which is not contradicted by anything found this pass.

Model 1 — forward earnings, updated for the new guide:

Basis EPS Multiple at $248.83
FY2026 guided non-GAAP (raised) $24.475 10.2x
FY2026 guided GAAP (raised) $18.145 13.7x
FY2027E consensus (implied by Yahoo forwardPE 9.05 — not independently reconstructed from quarters this pass, treat as indicative only per pitfall-vendor-forward-eps-is-the-wrong-fiscal-year) ~$27.49 9.05x
Scenario FY27E EPS P/E Implied price
Bear — multiple holds ~10x, growth fades to 8% $26.00 10x $260
Base — modest re-rate to 13x (CEO resolved, thresholds cleared) $27.50 13x $358
Bull — freemium converts, AI-ARR keeps compounding $28.50 16x $456

Fair value: $295–365 (was $290–360) — a small upward nudge, driven entirely by the raised FY26/implied-FY27 EPS guide against a roughly flat price, not by any re-rating of the multiple itself. The 13–16x band is unchanged from August: a quality-but-not-compounder multiple, held rather than raised, because the Wadhwani/Creative-succession risk offsets the credit the CEO resolution would otherwise earn.

Entry: $225–260 (unchanged). Current price ($243.50–248.83) sits inside the zone, closer to its middle than its top — a marginally better entry than August's $253.81, not a materially different one.

Trim: 13x fwd (unchanged). Held rather than raised for the same reason the FV multiple band was held: the case for paying up further should wait for the Creative successor to be named and for the Dec 9 Q4/FY27-guide print under (or just before) the new CEO, not be granted on the strength of a partial resolution.

Break triggers — carried forward, with the flow-vs-stock addition

The five numeric triggers from the baseline all cleared this quarter and remain the standing test:

  1. Total Adobe ARR growth below 10.0% for two consecutive quarters — cleared this quarter (11.2% YoY), still live as a forward test.
  2. AI-first ARR growth below 100% YoY — cleared this quarter (150%+), still live.
  3. Gross margin below 88% for two consecutive quarters — not yet triggered (88.7%), but only 0.7pt of cushion remains after the first quarter of measured compression.
  4. Creative & Marketing Professionals subscription revenue growth below 8% cc — cleared this quarter (12% cc), still live.
  5. (Superseded — CEO named) Replaced with: an external hire or a further reshuffle installs a cost-cutting-oriented head of the Creative business — would be the same negative signal the original CEO trigger was designed to catch, now relocated to the seat that actually matters.

New trigger, added this pass: net-new ARR / RPO growth continuing to run below total ARR growth for a second consecutive quarter — the leading indicator identified above, ahead of the total-ARR trigger itself.

Upgrade conditions: a credible internal or external Creative Cloud leader named with a growth (not cost) mandate; a permanent CFO named; Q4 print holds total ARR growth ≥10% with RPO/net-new ARR growth converging back toward the total-ARR rate.

Downgrade conditions: Creative Cloud goes into Q4 with no leader named at all; the Dec 9 print or FY27 initial guide comes in below the FY26 exit trajectory under new-CEO "kitchen-sinking" (Barclays' explicit risk); gross margin prints below 88% a second consecutive quarter.

Risk register — carried, with two additions

Risk Prob. Impact Change from Aug 4
Creative Cloud leadership vacuum drags into 2027 with no credible successor 30% (new) −8 to −12% New risk this pass
FY27 initial guide (Dec 9) is reset low under incoming CEO/CFO ("kitchen sink") 25% (new, Barclays-flagged) −10 to −15% New risk this pass
Freemium fails to convert; net-new ARR keeps decelerating 30% (down from 35%) Multiple stays 10–11x Total-ARR threshold cleared this quarter, so probability trimmed slightly
BofA's genAI-TAM-compression thesis is right 25% (unchanged) Terminal, multiple → 7–8x Not re-tested this pass
Permanent CFO named externally with a reset agenda 20% (unchanged) −5 to −10% short term Still open, no new information

What this pass did NOT test

  • FCF/share and the FCF-multiple valuation model — needs Q1–Q3 FY26 quarterly cash-flow statements, not pulled this pass. This was the baseline's co-primary valuation model; this file leans on the forward-earnings model alone and flags that as a real gap, not a stylistic choice.
  • Share count — fin.py's live field is suspiciously identical to the five-week-old baseline figure despite ~18M shares of buybacks in between. Not independently corrected against a primary 10-Q this pass.
  • Competitive landscape (Canva, Affinity, Figma) — carried unchanged from August on the strength of the baseline's own work; no new intelligence gathered, so "unchanged" here means "not contradicted," not "reconfirmed."
  • The secondary-source "$560M net-new ARR ex-Semrush, −3% YoY" figure — could not be corroborated against the primary SEC release. Recorded as an unverified single-source claim, not adopted as fact. If it holds up on independent confirmation next pass, it would meaningfully sharpen the flow-vs-stock ARR concern raised above.
  • Insider transactions since the baseline — no new pull around either the Sep 3 CEO news or the Sep 10 print. Given the CEO announcement is exactly the kind of event that tends to move insider activity (vesting schedules, 10b5-1 plans, discretionary buys/sells), this is a real gap for the next pass to close, not a "nothing happened" assumption.
  • Yahoo's forwardPE (9.05x) was not reconstructed from reported quarters per pitfall-vendor-forward-eps-is-the-wrong-fiscal-year, and Adobe reported today, which per that pitfall's Amendment (5) means even the current-year vendor field can lag a same-day print. Used here only as an indicative FY27E cross-check, consistent with how the baseline used it — not re-verified independently this pass.

Sources: Adobe Q3 FY2026 earnings release, SEC 8-K Ex-99.1 (sec.gov/Archives/edgar/data/796343/000079634326000147/adbeex991q326.htm — fetched directly); Adobe newsroom and SEC 8-K on the CEO transition (sec.gov/Archives/edgar/data/0000796343/000079634326000144/); CNBC, Bloomberg, Business Standard on the Chakravarthy appointment; Yahoo Finance, TipRanks, Invezz, 24/7 Wall St., RTTNews on the market reaction to the CEO news; Investing.com (earnings-call transcript excerpt, slides) and StockTitan on the Q3 print; .mcp/fin.py for the live snapshot; Output/Stocks/Technology/ADBE/analyze-2026-08-04.md and analyze-2026-06-04.md as baseline and second-prior file; Knowledge/Playbook/pitfall-vendor-forward-eps-is-the-wrong-fiscal-year, pitfall-yahoo-insider-purchases-counts-rsu-grants, principle-primary-source-beats-vendor.