August 17, 2026
Day91 Labs Brief — Issue 002: B2B SaaS
Ten signals, every one from a filing or a live price page
1. A seat-priced company just cleared ~26x ARR
Autodesk closed its acquisition of MaintainX on 3 August 2026. The merger agreement dated 28 May put aggregate consideration at $3.575 billion; Autodesk's own announcement described the deal as approximately $3.6 billion, all cash. Autodesk stated MaintainX expects to exceed $135 million of annualised recurring revenue for calendar 2026, with growth in excess of 50 percent. At $3.575 billion against $135 million, the implied multiple is roughly 26x forward ARR — and because $135 million is stated as a floor, the true multiple is somewhat lower. Autodesk also committed $150 million in RSUs to continuing MaintainX employees after closing.
Issue 001 covered Airtable clearing $1.285 billion at 2.7x ARR. Same year, same buyer market, roughly ten times the multiple. The variable is not the pricing model — MaintainX sells per user. It is the growth rate and the position: 50%+ growth in an operations workflow the acquirer's customers already run and cannot easily rip out. Autodesk was not buying revenue at that price. It was buying the "operate" half of design-make-operate, and paying for a category position it could not build fast enough.
Growth rate and adjacency set the exit multiple. Pricing model does not. A per-seat business growing 50% in a defensible workflow cleared ~26x in the same year a plateaued platform cleared 2.7x. Before you rebuild pricing to look more AI-native, check whether your growth rate is the actual constraint. And if you sell into Autodesk accounts, expect an operations-bundle conversation inside two quarters.
Sources
2. Autodesk's own backlog went the other way
Autodesk's 10-Q for the quarter ended 30 April 2026 reported net revenue up 18% to $1.93 billion, with recurring revenue at 97%. In the same filing: deferred revenue of $4.46 billion, down 5% against the prior fiscal fourth quarter; remaining performance obligations of $7.81 billion, down 6%; current RPO of $5.38 billion, down 2%. Net revenue retention was slightly above the 100–110% range on a constant-currency basis, unchanged year over year.
Recognised revenue reports on contracts already signed. RPO is the contracted future. When one is up 18% and the other is down 6% against the prior quarter, the growth on the income statement is being drawn from a backlog not being replaced at the same pace. Sequential RPO comparisons are noisy — Q4 is Autodesk's heaviest booking quarter, so a Q1 decline is partly seasonal. Current RPO, down 2%, is the harder number, because that is the twelve-month contracted book.
If your board deck leads with ARR or recognised revenue, put cRPO next to it on the same axis. The quarter they diverge is the quarter to change the plan — not the quarter after the number misses. Ask finance for cRPO by segment; if the divergence sits in one segment, that is a renewal problem, not a demand problem, and the fixes are different.
Sources
3. 95% retention, 12% decline
The Trade Desk reported Q2 2026 revenue of $715 million, up 3% year over year. Customer retention stayed above 95%, as the company says it has for over a decade. Adjusted EBITDA was $241 million, a 34% margin against 39% a year earlier. GAAP net income was $64.4 million at a 9% margin, versus 13%. Q3 guidance: revenue of at least $650 million and adjusted EBITDA of approximately $160 million — roughly a 12% year-over-year revenue decline.
Retention above 95% alongside a double-digit revenue decline is not a contradiction. It is the clearest available statement that the customers stayed and the spend did not. Every account renewed; every account renewed smaller. That is a net revenue retention problem wearing a logo retention headline, and the headline actively obscures it.
If logo retention is the retention number your board sees, it is the wrong one. Pull net revenue retention on the same cohort — the gap between the two is your contraction rate, stated in dollars. Then check whether your comp plan pays anything for defending spend inside a renewing account. In most GTM orgs it pays for the logo and it pays for the upsell, and it pays nothing for the flat renewal that quietly came in 15% lighter.
Sources
4. Waystar disclosed why its NRR fell — and it was borrowed
Waystar's 10-Q for the twelve months ended 30 June 2026 put net revenue retention at 108.3%, against 114.6% for the same period in 2025. The filing states that the rates for both periods include the impact of win rates above the company's historical highs and accelerated implementation timelines, both tied to a competitor's February 2024 cybersecurity incident.
The company named its own tailwind inside the filing. The 114.6% was never a structural expansion rate — it contained share taken from a competitor temporarily unable to serve its customers. As that cohort annualises, the number reverts. 108.3% is closer to the underlying rate than last year's figure ever was, which makes the decline a disclosure event rather than a performance event.
Decompose your NRR into structural expansion and event-driven expansion before you set next year's plan on it. If a competitor outage, a regulatory deadline, or a one-off migration wave sits in the base, the plan built on that base is wrong by the size of the event. The test is a single question: which expansion dollars would have arrived anyway?
Sources
5. GTM software is the cheapest category in horizontal SaaS
As of 17 August 2026, sales and marketing automation software traded at 1.7x NTM revenue and 7.9x NTM EBITDA, against a horizontal SaaS median of 2.2x and 9.7x. Its NTM revenue growth was 11% — one point above the median. Design and engineering software traded at 4.3x on 12% growth; ERP at 3.1x on 10%; BI and analytics at 3.0x on 14%. Only AdTech sat lower, at 0.9x. Source is a vendor comp set built on FactSet consensus and Morningstar historicals — not a filing, and labelled as such.
Same growth, less than half the multiple. The market is not discounting sales and marketing automation for performance. It is discounting it for the risk that the workflow gets replaced rather than augmented. Design software puts AI inside a loop that still needs a designer. The CRM loop — capture, route, sequence, log — is the one most exposed to an agent that never opens the interface it is priced against.
If you sell GTM software, growth alone will not re-rate you, and your board should stop expecting it to. What moves the multiple is evidence the AI *is* the product rather than a feature layered on a system of record: a metric that moves because of the model, disclosed on the same cadence as ARR. If you buy GTM software, this is leverage — your vendors' category is being repriced and their renewals desks already know it.
Sources
6. ACI's bookings fell 25% while the full-year plan held
ACI Worldwide's Q2 2026 results, filed 6 August, reported net new ARR bookings of $18 million, down 25% from Q2 2025. Trailing-twelve-month net new ARR bookings were $68 million, down 15%. New license and services bookings were $59 million in the quarter, up 2%, with the trailing twelve months at $255 million, down 12%. The company attributed the ARR shortfall to the timing of expected payment software contracts, and stated it expects full-year growth in both new ARR and new license and services bookings.
A 15% trailing-twelve-month decline paired with a full-year growth commitment puts the entire recovery in the back half. "Timing" is a legitimate explanation in enterprise payments, where one contract can move a quarter. It is also the most common phrasing for a slip that has not been called yet. The filing does not distinguish between the two, and neither can a reader.
When a plan depends on second-half recovery, the only thing worth tracking is whether the specific deals labelled "timing" close by the end of Q3. Ask for that list by name. If your own forecast has this shape, the honest version quantifies it: here are the four deals, and here is the number if two of them slip again.
Sources
7. SMB software retention is below 100, and staying there
EverCommerce's 10-Q for the quarter ended 30 June 2026 reported an annualised net revenue retention rate of approximately 94%, against 97% a year earlier. The company reported approximately 95% for the March quarter, against 97%. Approximately 97% of revenue was recurring or re-occurring in the six months ended 30 June 2026 and 2025.
Two consecutive quarters below the prior year, on a business that is almost entirely recurring. Sub-100 NRR in SMB software is structural, not cyclical — small businesses close, downgrade and consolidate tools faster than the survivors expand. The 97% recurring figure is a durability metric, not a growth one. At 94% NRR, that base shrinks every year unless new logos replace it.
If you sell to SMB, your growth model is an acquisition model, and pricing the expansion motion as though it will carry the number is a category error. Budget CAC on that basis, and treat any quarter where new logo volume drops as a revenue event that lands four quarters out — not a pipeline event that lands next month.
Sources
8. ARR up 52%, and a large share of it was bought
Digi International's fiscal Q3 2026 results, filed 5 August, reported ARR of $191 million at quarter end, up 52%. The same release states that fiscal 2026 results include Jolt Software for the full quarter and nine-month period and Particle Industries from its January 2026 acquisition date, while fiscal 2025 results include Jolt only from its August acquisition date and do not include Particle at all.
The 52% is real and the disclosure is clean — the company states exactly which periods contain which assets. What it does not state is the organic rate, and the base effect is large: one acquisition contributed a partial quarter last year and a full quarter this year; another contributed nothing last year. The headline is accurate and the comparison is not like-for-like, both at once.
Any ARR growth figure from an acquisitive company is two numbers until it is disclosed as two numbers. If you benchmark against a serial acquirer, ask for organic ARR growth or rebuild it from the acquisition dates before you put it in a board deck. If you are the acquirer, disclose both — the market assumes the worse of the two when you don't.
Sources
9. New business priced 7% below the installed base
SmartRent's 10-Q for the quarter ended 31 March 2026 reported Units Booked SaaS ARPU of $9.52, against $10.28 a year earlier — a 7% decline. In the same filing, property net revenue retention was 106%, up from 102%, and customer churn on Smart Communities Solutions was 0.05% against 0.02%.
Two directions at once. The installed base expanded: 106% property NRR is a real improvement over 102%. New bookings came in 7% cheaper per unit. Expansion pricing on existing customers is holding while the price required to win new business is falling. NRR reports what the market paid a year ago. Booked ARPU reports what it will pay today.
New-business ACV is the leading indicator; NRR is the lagging one. If your ACV on new logos has been drifting down while NRR holds, you have roughly a year before mix catches up and NRR follows. Track new-business ACV monthly and separate discount from packaging: a 7% decline caused by deeper discounting is a sales problem, the same decline caused by customers starting on a smaller bundle is a packaging problem, and the fixes point in opposite directions.
Sources
10. Two ways to cut sales and marketing, filed the same season
Box's 10-Q for the quarter ended 30 April 2026 reported sales and marketing expense of $101.9 million, up 3%, at 33% of revenue against 36% a year earlier — with sales and marketing headcount up 7%. The filing attributes part of the company's broader employee cost growth to hiring in lower-cost regions. Thryv's 10-Q for the quarter ended 30 June 2026 reported sales and marketing expense down $9.0 million, or 16.1%, driven by $4.6 million less in employee costs on lower sales headcount and $3.2 million less in sales commissions.
Same ratio pressure, opposite mechanics. Box held the ratio down while adding people, by changing where the people sit. Thryv cut the ratio by removing people — and the commission line is the tell. Commissions fell alongside headcount, which means volume came down with the coverage. One company preserved capacity and moved cost. The other reduced capacity and called it cost.
Before cutting sales and marketing to hit a ratio, separate the two levers. Cost-per-head is a location and mix decision that leaves coverage intact. Headcount is a capacity decision that shows up in bookings two quarters later. And watch your own commission line through any cut: if commissions fall in step with heads, you did not get more efficient — you sold less with fewer people at unchanged productivity, and the ratio improvement is arithmetic, not operating leverage.
Sources
Every claim links to its source. Spotted something wrong? It will be fixed and logged on the corrections page. Also published on Beehiiv.