Day91 Brief — Issue 004

A · read — 8-K exhibit opened. Both backlog figures are on the same page; the split is arithmetic.

FACT. Workday's Q2 FY27 results, filed 27 August 2026 for the quarter ended 31 July 2026, disclose a 12-month subscription revenue backlog of $9.034 billion, up 14.2%, and a total subscription revenue backlog of $27.403 billion, up 8.0%. The prior-year release disclosed $7.910 billion and $25.370 billion, up 16.4% and 17.6% respectively. Subtracting the near portion from the total leaves $18.369 billion of backlog beyond twelve months this year against $17.460 billion last year — growth of 5.2%. On the prior year's disclosed growth rates, the same long portion grew about 18%. Operating cash flow fell to $520 million from $616 million and free cash flow to $460 million from $588 million, on revenue up 12.8%.

READ. Total backlog growth halving while the 12-month figure barely moves is not a demand signal. It is a duration signal. The same dollars of annual commitment are being written over fewer years, which lifts the near-term number and starves the far one. Gross retention held at about 97%, so customers are not leaving — they are committing for shorter periods. Cash flow going backwards in a quarter with 12.8% revenue growth is the same fact seen from the collections side: shorter contracts bill smaller and later.

SO WHAT. Pull your own backlog and split it at the twelve-month line, then run the same split on last year's. If the far half is decelerating faster than the near half, your ACV is holding and your term is compressing — which means next year's renewal book is bigger, earlier, and more contested than your pipeline model assumes. Reforecast renewal capacity, not new logo capacity.

→ Workday 8-K Ex. 99.1, Q2 FY27 — SEC → Workday Q2 FY26 release — prior-year base

A · read — Both releases opened. Same quarter end, 31 July 2026 — a clean like-for-like comparison.

FACT. Okta's Q2 FY27 release, also for the quarter ended 31 July 2026, reports total revenue of $805 million, up 11%, with RPO of $4.858 billion, up 17%, and cRPO of $2.585 billion, up 14%. A year earlier RPO grew 18% and cRPO 13%. In the same release, Okta guides Q3 cRPO to $2.590–$2.600 billion, growth of 11–12%. Its posted commentary discloses that $100,000-plus ACV customers grew 6% to 5,255 while $1 million-plus ACV customers grew 22% to 605, and that trailing-twelve-month dollar-based net retention was 107%.

READ. Okta's backlog is growing six points faster than recognised revenue — the inverse of Workday's shape in the same eight-week window. That is what lengthening or holding duration looks like. But the guidance line matters more than the print: management is telling you cRPO decelerates to 11–12% next quarter, in the same document that reports 14%. And the customer split is doing real work — 6% growth in the six-figure band against 22% in the seven-figure band means the base is concentrating upward rather than broadening.

SO WHAT. Two companies, same quarter end, opposite duration direction. Before you accept any sector-level claim about contract terms this quarter, check which of the two your own segment resembles. Then check whether your top-decile accounts are growing faster than your mid-band — if they are by a wide margin, your net retention is being carried by a handful of accounts and your concentration risk is higher than the blended number shows.

→ Okta 8-K Ex. 99.1, Q2 FY27 — SEC → Okta Q2 FY26 release — prior-year base → Okta Q2 FY27 posted commentary (PDF)

A · read — Release read in full. The conditionality is stated by the company; the organic figure is derived and marked.

FACT. Salesforce's Q2 FY27 release, 26 August 2026, reports revenue of $11.345 billion, up 11%, subscription and support revenue of $10.8 billion, up 12%, and cRPO of $33.5 billion, up 14%. The release states that subscription revenue includes a $440 million Informatica contribution and total revenue a $456 million contribution. Full-year guidance was raised to $46.1–$46.4 billion. The company states that the pending Contentful and Fin acquisitions have been incorporated into all guidance metrics except cRPO, and that the guidance is conditional upon those transactions closing. Derived: backing out the disclosed $440 million from the disclosed 12% subscription growth puts organic subscription growth near 7%.

READ. Three separate things are being counted here and only one of them is organic. The 12% subscription growth includes an acquisition that closed; the raised guidance includes two that have not. cRPO — the one number Salesforce excluded from the acquisition adjustment — grew 14%, and that is the cleanest read in the document precisely because it is the one management could not dress. The Agentforce figures ($1.5 billion ARR, work units up 97% sequentially) are real but sit outside the audited revenue lines.

SO WHAT. When a vendor raises guidance, separate the raise into organic, closed-acquisition, and pending-acquisition dollars before you use it as a market datapoint. Then apply the same test to your own board deck: if your number moves because of something that has not signed, label it. The Salesforce release does. Most do not.

→ Salesforce Q2 FY27 results — Investor Relations

A · read — Press release figures read directly. The disclosure change is from the earnings call transcript and is marked inline.

FACT. Autodesk reported Q2 FY27 revenue of $2.046 billion, up 16%, against billings of $1.854 billion, up 10% — a $192 million gap in the quarter between what was recognised and what was invoiced. Free cash flow was $561 million. The MaintainX acquisition, agreed at approximately $3.575 billion and closed 3 August 2026, is guided to add roughly $60 million of second-half revenue and $70 million of second-half billings. On the earnings call, CFO Janesh Moorjani stated that Autodesk will stop disclosing Design and Make revenue going forward, describing it as a simplification consistent with converging workflows into operations.

READ. Revenue outrunning billings by six points means deferred revenue is being drawn down rather than built — the mirror image of Workday's problem, arrived at from the other side. Autodesk has been migrating multi-year contracts from up-front to annual billing, which mechanically produces exactly this gap, and the company has said so before. That is a defensible explanation. It is also unfalsifiable from the outside once the product-family split disappears. A $3.575 billion acquisition contributing $60 million in its first half lands inside a disclosure line that is being retired in the same quarter.

SO WHAT. Log every disclosure a vendor stops publishing, with the date and the quarter it last appeared. A withdrawn metric is not neutral — it removes the denominator for every future claim about that segment. If you are renewing with Autodesk, price the multi-year against the annual and ask which one their billings model prefers this year; the six-point gap tells you they have a preference.

→ Autodesk Q2 FY27 results — Investor Relations → Autodesk Q2 FY27 earnings call transcript → Autodesk Q1 FY27 10-Q, Note 19 — MaintainX consideration

A · read — 10-Q opened. The supplemental table gives five quarters of both ARR lines.

FACT. Veritone's Q2 2026 10-Q discloses a supplemental table with five quarters of software metrics. SaaS annual recurring revenue fell from $50.910 million at 30 June 2025 to $43.245 million at 30 June 2026. Consumption ARR rose from $10.957 million to $18.757 million over the same period. Total of the two lines moved from $61.867 million to $62.002 million — flat to within 0.2%. Consumption went from 17.7% of the combined figure to 30.3%. Total software customers fell from 3,066 to 2,829. New bookings fell from $14.965 million to $13.949 million. Non-GAAP gross margin fell 891 basis points year over year, which the filing attributes to declines in higher-margin consumption revenue.

READ. This is the substitution argument happening inside one income statement, and the headline ARR number conceals it completely. A company reporting flat recurring revenue is running off its subscription base at 15% a year and replacing it with consumption dollars at lower predictability. The customer count falling 8% while total ARR holds means the remaining accounts are spending more — which reads as expansion until you notice the margin line moved the wrong way with it.

SO WHAT. If any part of your revenue is metered, report the two lines separately to your board and never as a blended ARR. A flat total made of a shrinking subscription and a growing meter has a different forecast, a different gross margin, and a different churn definition than a flat total that is genuinely flat. Ask your finance team which of the two you are, in writing, this quarter.

→ Veritone 10-Q, Q2 2026 — supplemental software metrics

A · read — Release opened. The methodology restatement is in the company's own footnotes.

FACT. Nutanix's Q4 FY26 release, 26 August 2026, reports revenue of $757.1 million, up 16%, ARR of $2.549 billion, up 16%, and net dollar-based retention of 106%. Average contract duration was 3.3 years, up 0.1 years from the prior year. A footnote states that ARR for all prior periods has been adjusted to conform to an updated methodology. FY27 revenue guidance is $3.205 billion plus or minus $25 million, about 12.5% growth at the midpoint.

READ. Duration extending in the same quarter Workday's long backlog stalled is the counterexample that stops this from becoming a sector narrative. Infrastructure software is signing longer; application software in this sample is signing shorter. Separately: any ARR growth rate you computed against a pre-restatement Nutanix figure is now invalid, and the release tells you so in a footnote rather than a headline.

SO WHAT. Duration is the metric almost nobody in your company reports. Add dollar-weighted average term to your monthly pack this quarter — it is the earliest read on renewal pressure you can get, and it moves a full quarter before net retention does.

→ Nutanix Q4 FY26 results — Investor Relations

A · read — Stated in the release's own non-GAAP reconciliation language.

FACT. Okta's Q2 FY27 release reports non-GAAP diluted EPS of $1.05 against $0.91 a year earlier. The same release states that effective 1 February 2026, the start of fiscal 2027, Okta uses a fixed long-term projected non-GAAP tax rate of 21% in computing the non-GAAP income tax provision, against 26% through fiscal 2026.

READ. A five-point cut in the assumed tax rate mechanically lifts non-GAAP net income with no change in operations. The year-over-year EPS comparison is therefore not like-for-like, and the release discloses this plainly in the reconciliation section that most coverage does not reach.

SO WHAT. Before you cite any competitor's non-GAAP EPS growth in a deal, check whether the non-GAAP assumptions changed in the period. Assumption changes are disclosed and are almost never in the headline.

→ Okta 8-K Ex. 99.1, Q2 FY27 — non-GAAP reconciliation

A · read — Subsequent-events note read directly in the 10-Q.

FACT. Klaviyo's Q2 2026 10-Q, Note 14, discloses that on 4 August 2026 the company executed an asset assignment, waiver and release agreement with Agency AI, Inc. to acquire rights to specific proprietary software and intellectual property, and committed to hire Agency's team. Consideration is up to $17.0 million in cash, part of it contingent on transaction expenses. The transaction was expected to close in the quarter ending 30 September 2026.

READ. Structured as an asset purchase with a hiring commitment rather than a share acquisition, which keeps it out of 8-K Item 2.01 and out of most deal trackers. This is where agentic capability is actually being bought right now — in the subsequent-events note, at eight figures, for a team and some code.

SO WHAT. If you track competitive M&A, subsequent-events notes in 10-Qs are a better feed than deal databases for anything under $50 million. Add them to your reading list; they are where small capability buys surface first.

→ Klaviyo 10-Q, Q2 2026 — Note 14, Subsequent Events

A · read — 10-Q opened. Both retention figures are disclosed in the same paragraph.

FACT. zSpace's 10-Q for the quarter ended 31 March 2026 discloses that trailing-twelve-month net dollar retention was 65%, against 97% for the comparable prior period, and that software ACV fell to $10.1 million from $11.6 million. Total revenue fell 22% to $5.3 million. The filing attributes the decline to disruption in K-12 funding sources lengthening sales cycles, and to tariff surcharges being passed into revised quotes.

READ. A 32-point retention collapse in a single year, disclosed without a defined-term footnote softening it. The stated cause is buyer-side budget disruption rather than competitive loss, which is a different problem with a different fix — and the tariff pass-through detail is unusual: pricing changes are being forced through quotes mid-cycle, which is its own churn mechanism.

SO WHAT. If any of your segments depends on a single funding source — public budgets, grants, one channel — model the retention downside separately from the blended number. Concentration on the buyer's funding side does not appear in your customer concentration report.

→ zSpace 10-Q, quarter ended 31 March 2026

A · read — 10-Q opened. Both the ARR figure and the concentration shift are disclosed.

FACT. N-able's Q2 2026 10-Q discloses total ARR of $544.5 million at 30 June 2026 against $513.7 million a year earlier, an increase of 6.0%. Customers with over $50,000 of ARR grew 6.5% to 2,706. That cohort moved from approximately 60% of total ARR to approximately 63%.

READ. Six percent growth with the top cohort absorbing three more points of the total means the long tail is flat or shrinking. Growth is coming from expansion in accounts that already exist rather than from new logos or from the base broadening.

SO WHAT. Run the same cut: what share of your ARR sat in your top cohort a year ago versus now. If the share is climbing while growth is single digit, your expansion motion is working and your acquisition motion is not — and those need different budget decisions, not one blended growth target.

→ N-able 10-Q, Q2 2026 — ARR and customer cohorts

Every claim above carries the state of its own sourcing. Derived figures state the calculation and the inputs. If a link is wrong or a number is misread, reply and it is corrected in the next issue with the correction stated plainly, and logged publicly.

Day91 · Document-sourced revenue signals. AI-drafted, human-approved.