夜雨聆风价值计量正在从“有多少人登录”，转向“系统完成了多少工作”。
For most of my career, budgeting for software was one of the easier line items. You knew your headcount, you knew your per-seat rate, you multiplied the two, added an inflation assumption, and moved on to harder problems. It wasn’t exciting, but it was predictable — and predictability is most of what a CFO wants from a vendor relationship.
在我职业生涯的大部分时间里，为软件编制预算一直是较容易处理的项目之一。员工人数已知，每个席位的费率已知，两者相乘，再加上通胀假设，就可以转向更棘手的问题。这件事并不令人兴奋，却足够可预测——而在与供应商的关系中，可预测性正是首席财务官最看重的特质之一。
That predictability is now being tested by AI agent economics. When software’s value came from how many people used it, pricing by seat made sense. When a growing share of the work is done by agents working alongside people, the value model changes. And even if an agent is logging in on its own account, the value isn’t measured in adoption — it’s about impact. If you architect your AI systems effectively, you can have fewer people and more output with AI. Under a pure per-seat model, that would look like a shrinking software bill for a company getting more value — which is obviously not sustainable for anyone building the technology, and arguably not the right signal for the buyer either.
如今，AI 智能体的经济逻辑正在考验这种可预测性。当软件价值取决于使用人数时，按席位定价合情合理；但当越来越多的工作由智能体与人协同完成时，价值衡量方式也随之改变。即使智能体使用独立账号登录，衡量价值的也不再是“采用了多少”，而是“产生了多大影响”。如果 AI 系统设计得当，企业可以用更少的人力获得更多产出。在纯粹的按席位定价模式下，这会呈现为：企业获得的价值越多，软件账单反而越低。对技术提供方而言，这显然不可持续；对采购方而言，这也未必是正确的价值信号。
In a recent conversation with CIO magazine, Workday Chief Technology Officer Gabe Monroy said,
在最近接受《CIO》杂志采访时，Workday 首席技术官 Gabe Monroy 表示：
Value is no longer tied to a fixed factor like employee count. It’s tied to how much use an organization actually gets out of the system.
价值不再与员工人数这样的固定因素挂钩，而是取决于组织实际从系统中获得了多少使用价值。
That reframing is uncomfortable for finance teams who built their forecasting muscle on fixed costs. But the direction of travel isn’t really in question anymore. Gartner’s analysis, cited in Deloitte’s 2026 TMT Predictions, projects at least 40% of enterprise SaaS spend will move to usage, agent, or outcome-based pricing by 2030, with seat-based revenue falling to well under a fifth of market share.
对长期依靠固定成本建立预测能力的财务团队来说，这种重新定义并不轻松。但行业方向已基本没有悬念。德勤《2026 年科技、传媒和电信行业预测》援引 Gartner 的分析指出，到 2030 年，至少 40% 的企业级软件即服务支出将转向按使用量、智能体或业务结果定价；按席位计费带来的收入占比则会降至远低于五分之一。
According to Metronome’s 2025 State of Usage-Based Pricing research, 77% of large software companies have already built usage-based pricing into their revenue model. And 61% of SaaS companies are now running some kind of hybrid subscription and usage pricing, according to research by OpenView Partners. In other words, the shift away from per-seat pricing is already underway across the industry.
根据 Metronome《2025 年按使用量定价现状》研究，每 100 家大型软件公司中，已有约 77 家把按使用量定价纳入收入模式。OpenView Partners 的研究还显示，每 100 家软件即服务公司中，约有 61 家正在采用订阅费与使用费相结合的混合定价。换言之，整个行业已经开始告别单纯的按席位定价。
What a good pricing model owes you
好的定价模式应当为你提供什么
A lot of the commentary on usage-based pricing boils down to one legitimate fear — pay-as-you-go turns your software bill into a variable you can’t control. That fear is well-founded when the model is pure metering with no ceiling — a bad month for usage becomes a bad month for budget.
围绕按使用量定价的许多讨论，最终都指向一个合理的担忧：即用即付会让软件账单变成不可控的变量。如果采用没有上限的纯计量模式，这种担忧完全成立——某个月使用量过高，也就意味着当月预算失控。
What do customers need instead? They need a flexible AI credits approach.
那么，客户真正需要什么？他们需要一种灵活的 AI 点数机制。
A flexible AI credit model is anchored on an annual subscription, not a running meter. You buy a block of credits once per year, the same way you’d buy any other software subscription. This means the purchase is a predictable, one-time payment within the normal budget cycle. Credits should also be fungible — in our case, one universal currency applies across every Workday-built agent and platform solution.
灵活的 AI 点数模式应以年度订阅为基础，而不是让计量表持续累加。客户每年一次性购买一批点数，就像购买其他软件订阅一样。这样，采购仍是正常预算周期内可预测的一次性支出。点数还应能够通用——以 Workday 为例，一种通用点数可用于 Workday 构建的所有智能体和平台解决方案。
In a flexible credit model, usage should be metered against completed actions, not tokens or queries, so the ROI is clear — a contract reviewed, a close task completed, a case resolved — rather than an abstract unit that means nothing to a business stakeholder.
在灵活的点数模式下，使用量应按已经完成的业务动作计量，而不是按 token 或查询次数计量。这样，投资回报就清晰可见：审阅了一份合同、完成了一项结账任务、解决了一个案例，而不是用业务负责人难以理解的抽象单位来计费。
Info-Tech Research Group’s Scott Bickley pointed out that many ERP vendors have built usage models so layered that customers struggle to understand how capacity gets consumed. And it’s even more complicated to prove that capacity is adding value. That’s a real risk in this market. I’d encourage any finance leader to interrogate any vendor’s usage-based pricing, including ours. Ask to see the rate card, and ask what happens the day you exceed your allotment.
Info-Tech Research Group 的 Scott Bickley 指出，许多企业资源计划软件供应商设计的使用量定价层级过多，客户很难弄清额度究竟如何被消耗；要证明这些额度确实创造了价值则更加复杂。这是当前市场中的现实风险。我建议每一位财务负责人都仔细追问任何供应商的按使用量定价方案，Workday 也不例外。要求查看价目表，并明确询问：额度用完的当天会发生什么？
At Workday, there’s a live consumption dashboard with alerts at 80%, 90%, and 100% of balance, so finance and IT see the trend before it becomes a surprise. If a customer does run past their credit balance, the account team schedules a conversation to reconcile usage. This transparency ensures it’s cheap to course-correct while the industry is learning how to operate within the new pricing model.
Workday 提供实时用量看板，并在余额使用达到 80%、90% 和 100% 时发出提醒，让财务和 IT 团队在意外发生前看到趋势。如果客户确实超出点数余额，客户团队会安排沟通，对实际用量进行核对。在整个行业仍在学习如何适应新定价模式的阶段，这种透明度能够降低调整方向的成本。
Beyond the pricing debate - the work agents are doing
不只讨论定价，更要看智能体完成了什么工作
The only way to justify this shift is to prove the outcomes are worth it. Our early access customers have reported that using agents to review contracts cuts execution time by roughly two-thirds, and frontline agents have reduced time spent managing staffing changes by up to 90%. On the finance side, audit agents are saving early users up to 900 hours a year on audit evidence collection, and the agents in planning have cut data exploration and analysis time by around 30% close to 100 hours a month for some teams.
要证明这种转变合理，唯一的办法是证明成果值得付费。参与早期体验的客户报告称，使用智能体审阅合同后，执行时间大约缩短了三分之二；面向一线员工的智能体则把处理人员配置变更所需的时间最多减少了 90%。在财务领域，审计智能体每年可为早期用户节省最多 900 小时的审计证据收集时间；规划智能体把数据探索和分析时间缩短了约 30%，对一些团队而言，相当于每月节省近 100 小时。
AdventHealth has cut agency staffing spend by roughly $68 million in 12 months using our talent mobility tools, and early adopters including Arcis Golf, Mister Spex, and Valvoline have cut weekly scheduling time by up to 67%. Those are the numbers a usage-based model is trying to expose - cost tied to a specific task, set against the hours or errors that task used to cost you.
AdventHealth 使用 Workday 的人才流动工具，在 12 个月内将第三方用工支出减少了约 6,800 万美元；包括 Arcis Golf、Mister Spex 和 Valvoline 在内的早期采用者，则将每周排班时间最多缩短了 67%。按使用量定价希望呈现的正是这些数字：把某项具体任务的成本，与这项任务过去耗费的时间或造成的错误进行对照。
Getting ready for your next renewal - three moves to make now
为下一次续约做好准备：现在就采取三项行动
Here are three ways to prepare for a new pricing model while you evaluate a vendor:
在评估供应商的同时，可以通过以下三种方式为新的定价模式做好准备：
Model your highest-volume, most repetitive processes first. Usage-based costs scale with volume, not headcount, so the tasks that touch your largest employee or transaction population are the ones to price before you switch anything.
首先测算业务量最大、重复度最高的流程。按使用量计算的成本随业务量而非员工人数增长，因此，在切换任何模式之前，应优先为覆盖员工最多或交易量最大的任务测算成本。
Ask for a grace period and use it as a genuine test window. Use that grace period to establish your real usage baseline, rather than assuming the free period means the pricing conversation is deferred.
要求设置宽限期，并把它真正用作测试窗口。利用这段时间建立实际用量基线，而不要因为暂时免费，就认为可以推迟讨论定价。
Treat the consumption dashboard as a budgeting tool. Whatever platform you’re on, insist on real visibility into usage before you’re relying on it to hit a number.
把用量看板当作预算工具。无论使用哪个平台，在依靠它实现预算目标之前，都应要求获得真实、清晰的用量可视性。
None of this makes the transition effortless — moving from a fixed cost to a variable one never is. But the direction is set by the economics of the technology, not by any one company’s pricing strategy. The finance leaders who get ahead of it now, by building the forecasting and governance muscle early, will be in a stronger position.
这些做法并不会让转型变得毫不费力——从固定成本转向可变成本，从来都不轻松。但这一方向由技术本身的经济逻辑决定，而不是由任何一家公司的定价策略决定。现在就提前建立预测和治理能力的财务负责人，未来将处于更有利的位置。
资料来源：diginomica 原文页面 · 2026 年 8 月 18 日
封面摄影：Bohdan K. / Pexels · Pexels License