August 7, 2026
Day91 Labs Brief — Issue 001: B2B SaaS
Ten signals. Six sit on a filed document. Four do not, and they say so.
1. Airtable cleared $1.285 billion at roughly 2.7x ARR
Evidence tier A — Purchase price and ARR both taken off the filed document.
The acquisition of Airtable closed in the first half of 2026 at aggregate consideration of $1.285 billion in cash and stock, as stated in the acquirer's 8-K and the merger agreement filed as an exhibit. The same filing put Airtable's annualised recurring revenue at approximately $480 million at signing.
$1.285bn against ~$480m of ARR is roughly 2.7x. For a company that raised at $11 billion in 2021, the multiple is the whole story: the ARR held, the growth rate did not, and the price followed the growth rate rather than the revenue base.
If your ARR is flat and your board is still anchored to a 2021 comparable, reprice the internal expectation now. Buyers pay for the second derivative. Pull your last eight quarters of net new ARR and see which curve you are actually on.
Sources
2. Entrata's S-1 disclosed net revenue retention most private peers never publish
Evidence tier A — Retention figures read off the S-1 itself.
Entrata's Form S-1 stated net revenue retention above 110% for the trailing period, alongside gross retention in the high nineties and a customer count concentrated in large property-management operators.
The gap between gross and net retention is where the business model lives. High-nineties gross with 110%+ net means expansion is coming from more modules inside the same accounts, not from new logos — a land-and-expand engine that only works while the platform keeps shipping modules.
Put gross and net retention side by side in the same slide. If gross is high and net is barely above 100, your expansion motion is not working and no amount of new logo spend fixes the shape.
Sources
3. The seat-compression story was started by a company that later contradicted it
Evidence tier B — Reported and consistent with the 10-Q, but the segment table was not pulled.
A large enterprise software vendor spent 2025 telling investors that AI agents would compress seat counts across the industry. Its own most recent 10-Q shows seat-based revenue growing in the same period, with the decline concentrated in a single legacy product line.
The narrative was directionally useful for positioning a new agent SKU and was never true of the company's own book. Seat compression is real in support and QA. It is not yet visible in the filings of the platform vendors calling for it.
Do not rebuild pricing around a compression thesis that the filings do not support yet. Check the seat line in the 10-Qs of the three vendors you benchmark against before you move off per-seat.
Sources
4. Two vendors changed the unit, not the price
Evidence tier A — Rate and unit taken from the published pricing page on the date shown.
Two covered vendors moved from a per-seat rate to a per-resolution and per-conversation rate respectively, with the headline number falling in both cases. The published per-unit rates dropped while the billing denominator changed underneath them.
A lower rate on a unit you cannot forecast is a price rise for any customer with volume growth. The comparison the vendor invites — old seat price versus new unit price — is not a comparison at all.
Before you renew, model twelve months of your actual volume against the new unit. If the vendor will not cap it, ask for the cap in writing; that request alone tells you what they expect your volume to do.
Sources
5. A public filer held logo retention and lost spend
Evidence tier B — Figure reported in the earnings release; the 8-K exhibit was not opened.
A public filer reported customer retention above 95% in the same quarter revenue growth fell to low single digits.
Logo retention above 95% with flat revenue means every account renewed and renewed smaller. The headline metric was doing exactly what it is designed to do: hide contraction inside a renewal.
If logo retention is the number your board sees, replace it with net revenue retention on the same cohort. The gap between them is your contraction rate in dollars.
Sources
6. A nine-figure acquisition closed with no consideration disclosed
Evidence tier C — Undisclosed terms. No filing.
A strategic acquirer announced the close of an acquisition described in the press as nine figures. No 8-K Item 2.01 with terms was filed, and no purchase price appears in any public document.
Reported deal size with no filed consideration is a number someone chose to say. It is carried here because it moves comparables in the category, and marked because there is nothing behind it.
Do not put this in a comps table without a footnote. If you need the number, set an alert on the acquirer's CIK for the next 10-Q — purchase accounting usually surfaces it a quarter later.
Sources
7. Recognised revenue up, contracted backlog down
Evidence tier A — Deferred revenue and RPO read from the balance sheet and the disclosure note.
A platform vendor reported double-digit revenue growth while remaining performance obligations fell against the prior fiscal fourth quarter, with current RPO also down.
Recognised revenue reports contracts already signed. RPO is the contracted future. When they diverge, the growth on the income statement is being drawn from a backlog that is not being replaced at the same pace.
Put cRPO next to ARR on the same axis in the board deck. The quarter they diverge is the quarter to change the plan, not the quarter after the miss.
Sources
8. A Form D showed a raise the company never announced
Evidence tier B — Form D filed; the use-of-proceeds detail is not in it.
A venture-backed vendor in the covered set filed a Form D disclosing a new equity round. There was no press release, no blog post and no change to the pricing page.
Form D is the quietest signal in the corpus and the one competitors miss. Money in with no announcement usually means either a down round or a bridge — both are useful to know before a competitive deal.
Add Form D to your competitive alerts, not just funding-news feeds. A quiet raise is a live one.
Sources
9. The category multiple everyone quotes has no filing behind it
Evidence tier C — Analyst commentary. No document.
A widely circulated benchmark multiple for vertical SaaS this year traces back to a single analyst note aggregating private transactions with undisclosed terms.
A median built from undisclosed numbers is an opinion with a decimal point. It is carried here because boards are using it, and marked because it cannot be checked.
When someone quotes a category multiple at you, ask which filed transactions are in the set. If the answer is none, price off the deals that filed.
Sources
10. Gross margin fell while the headline metric improved
Evidence tier A — Both figures read off the same 10-Q.
A covered filer reported an improving efficiency metric in its shareholder letter while the 10-Q showed gross margin down year over year on higher inference and hosting cost of revenue.
The metric that improved is the one the company chose. The one that fell is the one in the audited statement. AI features are showing up in cost of revenue before they show up in price.
If you shipped AI features this year without changing price, pull your own gross margin by product line. The margin is where the cost lands first.
Sources
Every claim links to its source. Spotted something wrong? It will be fixed and logged on the corrections page. Also published on Beehiiv.