年度合同未必完美：SaaS定价陷阱与AI时代的计费策略

Annual contracts seem so great. VCs love them, and they do seem to reduce churn. Seem to. In fact, when CROs struggle to hit a plan, they often want to push customers onto annual contracts. That and pricing increases are classic go-to moves.

I’ve long been a vocal proponent of annual contracts. I felt like when I was a B2B CEO and had to go profitable, they helped saved my rear. Close, say, a $125k contract, even after a healthy sales commission, that’s $100k+ in the bank right now! And if your burn rate is low (ours was probably $100k-$150k at the time we started to close bigger annual deals), that’s like a whole extra month of runway just on that deal. A multi-year pre-paid contract can be even more magical, and give you months of extra cash in a start-up.

Annual contracts only hide churn

They don't reduce it

Don't let anyone tell you different

— Jason ✨👾SaaStr.Ai✨ Lemkin (@jasonlk) October 30, 2026

This is still true. But as time has gone by, and I’ve worked with more SaaS companies, I’ve also seen the downside of annual contracts at many start-ups as well:

• Annual deal collections can be tough for start-ups. Bigger companies are good at this, but if you aren’t, you may not get the benefits of an annual contract if it takes you 4-6 months to actually collect the cash. Annual contracts require P.O., bills, “payment terms”, and often, repeated follow-up. Most start-ups don’t have a good enough finance person or firm to be good at collections. Forcing your sales team to do collections is OK in the early days, but doesn’t work perfectly either, and doesn’t scale. Monthly invoices can make things even worse, of course. But getting paid in a simple ACH or credit card payment each month can be magical.
• Annual deals can create unnecessary customer friction. Big customers will want to pay annually. So for sure, do that. It’s how their procurement departments work. But if you have an incentive structure where even SMBs are pushed to go annual … best case, that will add friction to your sales process. Is it worth it? I’m not sure. Let SMBs pick annual or monthly, at least as an experiment. You’ll see what they really prefer.
• Annual deals get sent to legal for review, procurement for negotiation, etc. This makes sense for a $250k+ deal. But for a smaller deal, you are again just making it harder on the customer and slowing down the close. The odds if a deal gets sent to legal that it closes the same month are close to zero. This always at least takes weeks. Avoid it if you can.
• Annual deals, done wrong, can create unnecessary (and even more importantly, perpetual) discounting. You almost always have to provide a 10%-20% discount to incent annual deals. Is it worth it? Maybe not as the account grows over time. In Years 2, 3, and 10, you may wish you had that discount back when you have to honor it, one way or another, in the renewal. You are robbing a bit of a long-term revenue stream if you over-discount to get an annual deal done now. This can compound with multi-year deals. If you tell a loyal customer in Year 2 you are taking away their discount … expect that loyalty to evaporate.
• Annual deals mask churn. This is ultimately bad, not good. You may like annual deals for a second reason — they take a year to churn. But that’s not really true. It’s only true in your financial statements and your numbers. But if a customer never renews in SaaS, or if they stop using your product even if they’ve already paid for it for the year, it’s basically like they never really existed at all. A customer you lose next year doesn’t help you at all to get to $10, $20m, $100m in recurring revenue. A deal has to recur or in the medium and long run, otherwise it just doesn’t matter. Worse, it may mask weak efforts in customer success and retention efforts, and disincent you to track and grow NPS early.
• Annual deals are a tough fit with many API services and variable-usage services. It’s often better with variable revenue to move to an annual contract later in many cases, a few months later or even at renewal. But forcing customers to estimate variable usage before they are even in production, best case, again adds friction to a sales process. Is it worth it? I’d argue — rarely. In fact, MongoDB found revenue went up and sales cycles went down when they stopped forcing customers into annual contracts. More on that here.

Agents Make Forced Annual Contracts Even Harder to Defend

Agents change the annual contract math in three ways.

• Agents re-evaluate vendors every week, not once a year at renewal. A human buyer locked into an annual deal grumbles and waits for renewal. An agent routes around a vendor the day it underperforms. We lived this from the buying side this month. One of our core vendors was fully down for a day, app and API both throwing internal errors, with no acknowledgment. We’d already prepaid the year. 10K, our AI VP of Marketing, stood up an alternative and we switched. The vendor still has our cash for the rest of the year, and their dashboard probably still shows us as a retained customer. They lost us anyway. With agents doing the work, an annual contract buys you time, not loyalty. Benchmarkit’s 2026 data has median gross revenue retention across B2B falling from 88% to 84%, and agents switching vendors this fast is part of the reason.
• Buyers want agent work metered, not committed up front. In G2’s 2026 AI agents report, 57% of buyers preferred usage-based pricing and 9% preferred seats. No one knows yet how much agent work they’ll run 12 months from now, so asking a customer to commit to a year of agent usage before they’re in production is the old variable-usage problem made worse. HubSpot shows where this lands. The seat commitment stays annual, and agent usage floats on top as credits, with Customer Agent driving about 60% of credits consumed in Q4. Salesforce is selling Agentforce per conversation, per action through Flex Credits, and per user, all at the same time, because no one has settled on the right unit yet.
• The deals are bigger, and the commitment should follow results. The agentic B2B companies I’ll invest in now are the ones whose deal sizes are at least 2x what pre-AI competitors get, because they’re tapping budget beyond the software line. That kind of money doesn’t show up because procurement forced a 12-month term on day one. It shows up after the agent proves out in production. Land agentic customers monthly or on usage, show cost-per-outcome in the product, and then ask for the annual commitment. By then the customer has the data to justify a bigger number.

Annual Isn’t Dead, Though. It’s Just Less Often Where Agents Start

The AI leaders didn’t abandon annual contracts. They changed what the customer commits to.

Anthropic is the clearest example. A growing number of its enterprise customers sign non-cancelable annual or multi-year contracts, and those contracts mostly turn into consumption-based revenue. Consumption was $13.1B, or 80% of revenue, in H1 2026, and subscriptions were $3.2B, or 20%. The customer commits to spend for the year, and the bill follows usage. Anthropic now has 11,900 enterprises paying over $100,000 a year, up from 950 at the end of September 2025.

Sierra sells an annual platform contract and then charges for results. Customers pay a pre-negotiated rate per resolved case, escalations to humans are free, and Sierra’s revenue only shows up when the product works. ARR went from $100M to $200M in about two quarters.