August 24, 2026
Day91 Labs Brief — Issue 003: B2B SaaS
Ten signals. Six are already sitting on a document. Four are not, and they say so.
1. The biggest software story of the month has no document behind it
Evidence tier C — No filing exists. Reported talks only.
On 13 August 2026, Reuters reported that Silver Lake was in talks to take Workday private. The stock closed roughly 18% higher the same day. No 8-K has been filed. There is no Item 1.01 definitive agreement, no merger agreement exhibit, and no proxy.
Talks are not an agreement, and a wire report is not a document. The 18% move is real and measurable; the transaction it is pricing is not yet disclosed. Every other publication ran this as a deal. It is a report of a conversation.
If you are modelling a competitor set off this, mark it unconfirmed and set an alert on the acquirer and target CIKs for an Item 1.01. Day91 will not carry this as a fact until that filing exists — and if it never does, this entry becomes the correction.
Sources
2. Price per account up 12%. Retention on that same base down 12 points.
Evidence tier B — Figures reported from the Q2 filing. Pull the 10-Q before relying on them.
Thryv's second quarter reported SaaS monthly ARPU of $394, against $352 a year earlier. In the same period, Seasoned Net Revenue Retention — clients at least two years on the platform — fell to 90% from 102%. Consolidated gross margin fell to 62.7% from 69.7%.
ARPU up and NRR down, disclosed together, is not expansion. It is a shrinking cohort paying more per head. The gross margin move says the incremental revenue is arriving at a worse cost to serve than the revenue it replaced.
If you are pushing a list-price increase into a soft renewal book this quarter, read your ARPU and your seasoned NRR on the same slide, not on consecutive ones. The pair is the signal; either one alone reads fine.
Sources
3. Five vendors, five incompatible billing units, one buying cycle
Evidence tier B — Vendor-published pricing pages. Re-verify each before send; these move without notice.
As published this month: Atlassian standard plans include 25 Rovo AI credits per user per month, with Virtual Service Agent conversations charged at $0.30 beyond the included limit. HubSpot charges $10 per 1,000 AI credits above tier allotment. Zendesk prices its AI resolution agent per resolved conversation. Salesforce runs Agentforce on three parallel models simultaneously — Flex Credits, priced conversations, and per-user licensing for Agentforce for Sales. SAP has said it will move to charging on AI consumption rather than user access.
Credits, conversations, resolutions, seats and tokens are five different denominators. There is no exchange rate between them, and any vendor comparison that produces one has invented it. This is the Repricing Index's whole premise, now visible in vendor-published documents rather than in commentary.
When you next benchmark your own price against a competitor, check whether you are comparing the same denominator. If you are not, the comparison has no content — and neither does the one your board is looking at.
Sources
4. A document-intelligence company shrinking 13% is worth reading on yourself
Evidence tier B — Company-reported quarterly results. Confirm against the 10-Q.
FiscalNote reported second quarter revenue of $19.6 million. ARR declined roughly 13% year over year. Quarterly net revenue retention recovered to 98% from 89% in the first quarter, while trailing-twelve-month NRR sat at 81%.
Quarterly NRR recovering while the trailing number sits at 81% means the churn is behind them and the base is materially smaller. The recovery is real; it is a recovery on less.
The read for anyone selling information as a broad platform subscription: the subscription decays when it is sold as access to everything. It holds when it is sold as coverage of one question the buyer is accountable for. Check which one your renewal deck is describing.
Sources
5. Private software is now trading above public software
Evidence tier C — Index levels from a third-party publisher. Cite the index owner directly, not a summariser.
The SaaS Capital Index peaked at 16.9x ARR in 2021 and entered 2025 near 7x. It fell to a decade-plus low around 3.2x ARR by mid-2026 as the market priced in AI substituting for per-seat software, then recovered to roughly 3.8x in late July. Private lower-middle-market medians held near 4.5x ARR through the same period.
That is an inversion, not a spread. Public markets reprice daily on a narrative; private deals are negotiated on cash flows six to twelve months behind. For most of software's history the discount ran the other way.
If you are a private company benchmarking against public comps, you are currently benchmarking down. If you are a sponsor, the arbitrage is the entire explanation for signal 1.
Sources
6. An earnout structure worth copying into your own term sheet
Evidence tier A — Filing opened and read.
Core Scientific closed its acquisition of Polaris DS on 13 August 2026. The 8-K discloses an aggregate purchase price of approximately $444.3 million in cash, subject to customary post-closing adjustments, plus up to $40 million more if an additional 40 megawatts of firm electric capacity becomes available before 31 December 2026. A portion was held back in escrow against the seller's indemnification obligations.
The contingent $40 million is tied to a physical, verifiable, third-party-controlled milestone with a hard date. Not a revenue target, not an integration milestone, not anything the buyer controls the measurement of.
Most SaaS earnouts are written against ARR or bookings, both of which the acquirer controls post-close and both of which produce disputes. If you are negotiating one, the structure to argue for is the one whose trigger neither party can move.
Sources
7. A software rollup filing an 8-K signed by an interim CFO
Evidence tier A — Filing opened. Item detail still to be read in full.
Upland Software filed a Form 8-K dated 14 August 2026, executed by an interim Chief Financial Officer.
Upland is an acquisition-assembled software platform. An interim finance chief signing a current report is a governance state, not an accounting one — but it is the state a company is in when integration accounting, debt covenants or a strategic process are being reworked.
For anyone tracking rollup consolidators as acquirers or as competitors, this is a watch flag rather than a conclusion. Read the item, then check whether the next 10-Q restates any segment.
Sources
8. Two AI tuck-ins where the price was not disclosed
Evidence tier C — Deal reported, terms undisclosed, no US filing located.
Asana completed an acquisition of AI workflow company StackAI on 31 July 2026, terms undisclosed. On 30 July, Thoma Bravo took a majority stake in French occupational-health software provider padoa through its European fund, value undisclosed.
Undisclosed terms in a completed acquisition usually means the consideration fell below the acquirer's materiality threshold for Item 2.01 disclosure. That is information: it puts a ceiling on the price even when the price is withheld.
When a public acquirer buys and says nothing about price, you can bound it from their own disclosure thresholds rather than guessing. Read the acquirer's stated significance test in its last 10-K, not the press release.
Sources
9. The pricing-software company went private
Evidence tier B — Closed transaction. Pull the merger 8-K and final proxy for terms.
Thoma Bravo's $1.4 billion all-cash acquisition of PROS Holdings closed in March 2026. PROS sells pricing and revenue-management software.
A sponsor buying the vendor that sells pricing optimisation, in the same year the entire industry is rewriting how it prices, is a statement about the category rather than about the asset.
If you evaluated PROS or a competitor in the last eighteen months, the roadmap you were sold is now subject to a sponsor's return timetable. Ask for the post-close product commitment in writing at renewal.
Sources
10. The list the sell side handed you for free
Evidence tier C — Analyst commentary. Included as a watchlist, not as a valuation.
Following the Workday report, KeyBanc named GitLab, UiPath, PTC and Procore as likely candidates within its coverage, with Adobe and HubSpot from its large-cap coverage. The same note benchmarked EV/FCF across the group: Atlassian at 21x, ServiceNow 19x, SAP 18x, Salesforce 13.5x, Intuit 12.5x, HubSpot 11x, Adobe 10x. Truist separately characterised 2026 software M&A as having been quiet.
None of this is a document and none of it is a forecast worth acting on. It is useful for one thing only: it tells you which names a room full of buy-side analysts will be running screens against next week.
Add the six names to your competitor watchlist now rather than after a deal prints. If one of your vendors is on it, your renewal in six months is a different conversation than the one you planned.
Sources
Every claim links to its source. Spotted something wrong? It will be fixed and logged on the corrections page. Also published on Beehiiv.