Board memo

Five questions every software board should ask in 2026

May 2026.8 min read

The old growth logic, and where it is straining

The first growth curve in B2B software was, in commercial terms, a contract between buyer and vendor about access. The buyer paid for seats. The vendor sold for seats. Renewals were re-priced on seats. Expansion was modelled on seats. The compounding asset was the size of the user base inside the customer, and the long-running unit economics of multi-tenant SaaS were strong enough to carry an 80-point gross margin standard for over a decade.

That contract is breaking in 2026, and the break is not subtle. The iShares Expanded Tech-Software ETF, the cleanest proxy for the public software basket, was down more than 21 percent year-to-date through Q1 2026, with the sector erasing roughly $2 trillion in market capitalisation as investors repriced agentic AI risk into the cash flows. The selloff was not driven by a macro recession. It was driven by a structural question, and boards are now being asked to answer it: when one autonomous agent can do the work of ten or twenty seats, what exactly is being monetised on renewal?

The five questions that follow are designed to force that answer at the next board meeting. They replace the AI roadmap update that most boards now receive.

The roadmap update tells the board what is being built. These questions tell the board what is being protected.

Question 1. What workflow do we actually monetise today?

The first question is the one most management teams will not write down in clean language. Not “what does our product do.” Not “what is our value proposition.” What human workflow does the customer’s organisation perform inside our software, on which our pricing depends? The honest answer is usually narrower than the product brochure, and it almost always implies a specific seat count of humans repeating that workflow on a defined cadence.

The reason this matters in 2026 is that the answer to question one is the input to questions two and three. If the monetised workflow is “the sales rep updates pipeline notes in our CRM,” that is one risk profile. If it is “the system of record for every customer obligation in a regulated industry,” that is a different one. Boards consistently overestimate the second category and underestimate how much of their own portfolio sits in the first.

Question 2. Could an autonomous agent now remove that workflow?

This is the workflow obsolescence risk question, and the honest answer is uncomfortable. Atlassian disclosed its first ever decline in enterprise seat counts in 2026, attributed to AI agent adoption replacing tasks previously performed by humans inside Jira and Confluence (Death of Per-Seat Pricing, The SaaS CFO). ServiceNow now reports that 50 percent of net new Now Assist business is on non-seat pricing tied to tokens and consumption, on a product tracking to $1.5 billion in ACV for 2026 (Bessemer AI Pricing Playbook). These are not edge cases. They are the lead indicators.

The board test is not whether the agent fully removes the workflow today. It is whether a credible engineering team, with off-the-shelf models, can remove enough of it within twenty-four months to compress the per-customer seat count by more than the contractual price escalator. If yes, the existing growth plan is funded on a base that is contracting underneath it.

Question 3. Does AI raise or lower our switching costs?

Switching costs are the asset that quietly carried SaaS through the last cycle. They are also where AI is doing the most ambiguous damage, because the same technology that can lock customers in can also lower the cost of moving off. Three categories remain structurally defensible in 2026: proprietary datasets that competitors cannot replicate, regulatory and compliance infrastructure with long switching cycles, and integration depth that makes a system of record expensive to remove (HarbourVest, The Software Industry’s Great Reset). Everything else is being repriced.

A survey of 817 enterprise software builders found that 35 percent have already replaced at least one SaaS tool with a custom-built alternative, with workflow automation the leading category under pressure (The SaaS Moat Crisis, Big Ideas DB). The board question is not whether the company has switching costs. It is which of those costs would survive a customer’s AI engineering team being told to evaluate replacement, and which would not.

Question 4. Can our pricing model survive AI-native economics?

The fourth question is the one that hits the P&L hardest. Per-seat pricing assumed two things that are no longer true at the same time: the seat count is stable or growing, and the marginal cost of serving the seat is near zero. AI changes both. Seat counts compress when agents arrive. The marginal cost of serving an AI-enabled feature is not near zero, because inference cost is variable. ICONIQ’s January 2026 snapshot put average AI product gross margin at 52 percent, up from 41 percent in 2024, but still well below the 80-percent SaaS norm (The Economics of AI-First B2B SaaS in 2026, Monetizely). Across Q4 2025 and Q1 2026 earnings, public SaaS vendors disclosing AI-driven margin pressure now name 60-to-70-percent gross margin as the realistic operating range (The AI COGS Problem, SaaSMag).

This is the AI margin illusion. Growth that is not economic quality. The board question is binary. Has the company modelled, against actual inference cost curves and customer usage assumptions, the gross margin profile of its three largest products at five-year out adoption rates? If the answer is no, the pricing model has not yet survived the test. Bessemer is now explicit that consumption- and outcome-based pricing, or hybrid models with a fixed base plus variable consumption, are the dominant transition state for 2026 enterprise renewals.

Question 5. Which part of our go-to-market motion is becoming obsolete?

Most boards see AI in the product roadmap but not in the growth plan. The growth plan still assumes the existing go-to-market motion: outbound that fills a top of funnel sized for human buyers, inbound that runs on search and content, SDR and AE coverage at a particular ratio, and renewal management that defends seat counts. Every one of those assumptions is being tested.

Gartner now forecasts that at least 40 percent of enterprise SaaS spend will move to usage, agent, or outcome pricing by 2030, with seat-based revenue share declining from 21 to 15 percent (Bessemer, AI Pricing Playbook). Thoma Bravo and Vista, the largest specialist software investors, are publicly reassuring LPs that their portfolios are intact while privately reorganising portfolio company go-to-market around agentic delivery (Bloomberg, Thoma Bravo and Vista on AI; Yahoo Finance coverage). The board question is which part of the company’s current go-to-market motion is structurally redundant inside the next three renewal cycles, and what is being put in its place.

What the board should stop funding, and what to test

The five questions converge on a single conclusion. The growth plan a software board approved in 2023 was built on the SaaS playbook isn’t enough premise: seat compounding, 80-point margins, and pricing power resting on switching costs that AI now reaches into. The plan needs to be rewritten, not patched.

Three actions to put on the next board agenda. First, stop funding AI roadmap theatre and require a written link between every AI investment and a change in pricing, margin, or workflow ownership. If a feature does not change any of those, it does not belong in the strategic envelope. Second, commission an outside-in workflow obsolescence audit on the top three monetised workflows, using the same engineering capability the company’s own customers now have. Third, run a parallel pricing model design alongside the existing one for a single product line over the next two renewal cycles, and instrument it against gross margin and net revenue retention.

The board question that should replace the old one is not whether the company is adopting AI. It is whether the company can still earn an 80-point gross margin on the workflow it monetises, two AI-renewal cycles from now. That answer determines the next growth curve. The roadmap update does not.

Sources

boards, AI-native, pricing, switching-costs, go-to-market

Back to writing