In partnership with
Costs are rising almost everywhere you look: benefits, software, automation, hiring. Your customers are looking every bit as hard at what they spend on you, and an annual contract doesn't buy the security it used to. In this week's feature, find out what your customer needs to see within six months of signing, even when they've committed for a year or more.
| In this issue | ||||||
|
The median business spent $11.95 per employee on AI in July, according to Ramp's August AI Index. The top 1% spent a median $7,400. That leaves a lot of room between "we use AI" and "we spend serious money on AI." If your AI costs rise with usage but your customers pay a flat price, model what your heaviest users cost you before you decide how much AI to include.
Health costs to rise 8.2% after plan changes
Mercer's preliminary survey says health benefit costs will rise an average 8.2% in 2027, the biggest increase since 2003. And that's after employers make changes to their plans. Without those changes, they expect costs to rise 11%.
GLP-1 drugs account for about one percentage point of the increase. Fifty-nine percent of employers plan cost-cutting changes, and about two-thirds of employers with 500 or more workers expect employees to pay a larger share of premiums.
Founder read: If you choose a higher deductible to keep benefit costs down, decide how much of that extra risk you want employees to absorb. You can offset some of it through the plan design, employer contributions, or other compensation before the change starts feeling like a pay cut.
Atlassian starts charging automation by the step Dec. 3
Starting Dec. 3, Atlassian will count each step an automation rule runs instead of each completed rule run. Additional automation will cost $0.50 per 1,000 steps, so rules with more conditions, branches and actions will consume more of the allowance (Atlassian, TechRepublic). Founder read: Find the automations with high volume and lots of branches. Those are now candidates for simplification or moving into code. A workflow your team built because Atlassian made it easy can become expensive as the company grows.
GitHub moves Copilot seats to upfront billing Oct. 1
Starting Oct. 1, GitHub will charge for every assigned Copilot Business and Enterprise seat at the beginning of the billing cycle, including seats already assigned. That moves the charge ahead of the month in which the seat gets used (GitHub Changelog). Founder read: Run Copilot pilots with a small group and line them up with the billing cycle. Measure who uses it and where it saves time, then add seats. Upfront billing puts more of the cost of a failed rollout on you.
Professional and business services hiring fell by 188,000 in July
The Bureau of Labor Statistics reported 188,000 fewer hires in professional and business services in July, while the quits rate across the economy held at 1.9%. Employers pulled back on hiring without a corresponding increase in workers leaving their jobs. Founder read: Use the slowdown to tighten your must-have criteria and screen harder at the top of the funnel. Reject weak matches earlier, then move quickly when someone clears the bar.
AI made PMs faster. Multiplayer mode is still broken.
A PM can summarize research, draft a PRD, and mock up a prototype before lunch. The hard part starts when the team has to decide what actually gets built.
Jira Product Discovery gives product teams one place to capture insights, prioritize ideas with consistent frameworks, and build living roadmaps stakeholders can rally around.
And because it’s connected to Jira, the context behind every decision stays with the work—so developers and their agents know not just what to build, but why.
AI helps PMs move faster. Jira Product Discovery helps the whole team build with confidence.
What your customer needs by month six
Buyers expect a return within six months of buying, however long the contract runs.
In the last quarter of 2023, a pharmaceutical company's CIO paid extra for 500 employees to use Microsoft 365 Copilot. Six months later he turned it off.
We know this because he said so on a call with Morgan Stanley analysts, and Business Insider read the transcript. He is identified only as "Greg." He compared the slide decks the tool produced to "middle school presentations," and about the cost he said: "The price is double, and we really just do not see the value." At $30 a seat a month, the 500 seats came to about $180,000 a year.
The company went unnamed, and Microsoft has sold a great many seats since, so one CIO on one analyst call means only so much.
G2 asked nearly 2,000 software buyers when they expect a return: 78% said within six months of buying, and 44% said they got one on that schedule.
The contract runs longer than that. Tropic, which tracks more than $18 billion in software spend, put the average AI-native contract at 22.4 months in 2025.
How often is often?
In May 2026 Madrona asked 150 enterprise decision-makers, all at companies with $500 million or more in revenue, how often they re-evaluate their AI tools. More than three in four said at least every six months, and the largest single group, 29%, answered "rolling or ad hoc," which means whenever. About one in four still reviews once a year or when the contract ends.
The report never defines "re-evaluate." TechCrunch's Julie Bort read the finding to mean that "an enterprise contract no longer secures long-term revenue." The survey records how often buyers say they review, and nothing about what they decide.
In an April 2025 G2 survey of 1,169 decision-makers, half of the buyers at companies with 1,000 to 5,000 employees said they had already switched a vendor to get better AI. At companies above 5,000 employees, 41% had.
For a vendor, the safe assumption is that the customer can reopen the decision while the contract is still running, and some of them do.
Causes of death
Madrona also asked what became of the pilots these buyers ran in the previous year. For 83% of them, fewer than half made it into production.
The reasons pilots failed, in order of frequency, were integration trouble, security and compliance, ROI scrutiny and admin friction. Price came fifth. "Did not work as promised" came sixth, and Madrona's reading of that ranking is that "the product usually works."
The pilots that made it to full deployment had strong end-user adoption, an executive sponsor and clean integration, in that order, with clear and quantifiable ROI fourth and favorable pricing fifth again.
Madrona adds that "by the time the formal ROI analysis happens, the decision has often already been made informally." On both lists the ROI case ranks below integration, so a value report counts for little if the product doesn't integrate.
By month six, I'd want a customer to have evidence for those things without needing a rescue mission from the sales team.
The invoice as a translation problem
The unit on the bill either matches how the customer describes the value or forces them to convert it.
Madrona's buyers were asked which pricing model they prefer and which they get. A third prefer to pay for outcomes; 13% are offered that. Usage-based pricing runs the other way: 21% want it and 45% get it.
(Hybrid and per-seat roughly match their demand; I'll spare you the rows.)
Madrona sums it up this way: "Buyers want to pay for outcomes. Vendors are charging for consumption."
Andreessen Horowitz asked 50 technical AI buyers a related question. Twenty-seven preferred credits tied to recognizable work and 14 preferred tokens. The post doesn't say what the other nine wanted.
The example the post gives is a support leader, who can estimate conversations handled and cannot predict "context length, retrieval volume, retries, reasoning time, or output tokens."
The firm's advice is to show customers "what work was completed, where capacity went, and why certain tasks consumed more than others."
Somebody inside the customer will eventually have to defend the spend, and a count of completed work is easier to justify than a count of tokens.
Pricing can also start the review on its own. Tropic found AI-related increases at renewal running 20% to 37% over the past 12 to 18 months, against a normal annual uplift of 3% to 9%.
A bill that jumps by a fifth or more gives procurement a fresh reason to ask what the company got for it.
Before somebody asks
Madrona tells vendors to build "usage analytics, proactive health checks, expansion triggers." The customer shouldn't have to assemble the case from several systems and a call with your account executive.
Check what evidence they can already pull on their own. Can they see who is using the product? Can they connect that use to recognizable work? Can the economic buyer see what changed after go-live? Can someone explain the bill in the same units the company uses to describe the value?
The contract might run a year or two. Greg gave it six months, and 78% of G2's buyers expect a return on the same schedule.
Give them something concrete to show before the six-month review.
Abel lays out a two-to-three-day pilot with three or four users and success criteria agreed in advance. It's a useful model for keeping a pilot from turning into weeks of activity that nobody knows what to do with. Start around the 49-minute mark.
At 12:32, Walling suggests pricing an AI agent at roughly one-tenth the salary of the role it replaces. If you're pricing an agent, that gives you another number to test yours against: what the work costs the buyer today.
Poyar covers a Profound test of whether ChatGPT and Perplexity could read Cloud 100 pricing pages. Only 57% passed. The piece includes a simple first check for yours: load the page with JavaScript turned off and see whether the pricing survives.
Your next 100 customers are already in Apollo
Find, reach, and close your perfect deals — without juggling five tools or hiring more reps.
Apollo gives you everything you need to build real pipeline, fast. From inbound to outbound, first touch to close.
All in Apollo.
I run The Pricing Reset, a six-week engagement for founders of sales-led B2B SaaS companies whose product has outgrown the pricing they set a few years ago. I rebuild the packaging, price points and discount rules, then help put the new pricing live on new deals so the team can run it without every exception landing on the founder.
The work continues through a 90-day measurement window, with check-ins at days 30, 60 and 90 and a final readout against the company's own starting numbers.
If your pricing has not kept pace with the product, book a 20-minute call at redwoodridge.net/book, or reply with your current pricing page. I'll tell you whether The Pricing Reset applies and, if it doesn't, what I'd look at instead.
Thanks for reading. See you next Wednesday.
— Jason
P.S. If your customer starts questioning the purchase earlier than you expect, April's The Renewal You're Walking Into Blind covers what happens when those questions turn into a decision.
