Summary

  • Workable announced an AI Credits Report on September 10, available within Recruiting reports on all plans without an additional reporting charge, subject to administrative permissions.
  • Visibility into credits is useful accounting evidence. It does not by itself establish total hiring cost, cash savings or the agent's causal return on investment.

A recruiting team can consume fewer credits without spending less cash. It can also buy cheaper credits and leave more of them unused. Those possibilities are not objections to metering. They explain why Workable's new reporting feature is a starting point for financial scrutiny rather than its conclusion.

The September 10 announcement describes a dashboard covering balances, consumption, expirations, activity breakdowns and cost per hire from agent activity. Job-level tables can drill down to candidates, with filters including department and location. Workable presents this as a way to make its AI activity accountable. The report carries no additional charge across plans, but access requires the relevant administrator-level permissions; that does not make agent usage free or the report visible to every employee.

One credit is not one universal cash price

Workable's current pricing page, reviewed on September 11, gives the units a concrete commercial context. It lists one credit for an evaluation, two for sourcing a passive candidate into the pipeline, and ten for a candidate's chat interaction. These are different charged actions, not interchangeable measures of recruiting value. The reviewed page does not separately resolve every re-evaluation charging question.

The displayed bundles range from 5,000 credits for $600 to 50,000 for $4,750; a 10,000-credit bundle costs $1,000. That produces stated unit prices of $0.12, $0.095 and $0.10 respectively, with larger custom packages also offered. A paid account starts with 3,000 free credits to try—not a publicly promised monthly replenishment. Purchased credits expire after a year.

Consequently, the same credit count need not represent the same purchased-dollar allocation for two customers. Nor is cash paid for a bundle identical to the value of credits consumed during a particular reporting period. Free balances, differently priced purchases and unused expiry all complicate a single multiplier. No customer invoice, realised saving or effective cost has been verified for this article.

The date of the hire matters

There is a second boundary in the denominator. A job may incur evaluation activity in one period and produce a hire in another. A buyer would want to understand which events and cohorts the reported cost per hire connects, including reopened roles and periods without a completed hire. Those are questions about interpretation; the announcement does not disclose a formula that this article can reproduce.

History also has an explicit starting point. Workable says report data begins on July 13, 2026, when its credits model migrated. Balances from the previous model appear as one carryover item. That can help reconcile an opening stock without establishing a reconstruction of earlier activity under the new model. It is not evidence that old records were lost or could never be obtained.

This makes a before-and-after efficiency claim more demanding than comparing two dashboard totals. The reader needs to know whether the units, role mix and hiring stage are comparable. A difficult specialist search and an easier recurring role can consume different resources for reasons that a credit counter alone does not explain.

Better visibility, a narrower claim

The report could reduce reliance on ad hoc requests for consumption information. It may help teams notice where activity concentrates and decide where to investigate. Those are practical benefits of a clearer record, even before anyone claims an investment return.

But credit spend attached to a hire is not the entire recruitment bill. Recruiter time, the underlying software subscription and other hiring expenses are separate questions. A lower consumption ratio also does not establish that the agent caused additional hires, improved their quality or released labour that the employer actually saved.

Workable's release describes further Agent Productivity Reports, including token consumption and productivity distribution, as future work. They should not be treated as existing evidence. The present news is a more inspectable usage ledger. Its commercial value will depend on how convincingly that ledger can be connected to money, comparable hiring outcomes and the work that remains outside it.