How is cost per hire calculated?

Cost per hire is a single division: total recruiting spend in the period divided by the number of people who started work in it. The difficulty is not in the formula, it is at both ends of it. If it is not written down which costs enter the numerator, the number changes with whoever calculates it; if the period boundary in the denominator is vague, the same quarter yields two different results.

The visible consequence is the argument in the budget meeting. The invisible one costs more: a number whose line item list was never written down cannot be compared with last quarter, because the two may not be measuring the same thing. The job of a cost metric is not to find an absolute figure, it is to produce a figure that can be repeated the same way every period.

The GoTeam panel does not calculate this metric; there is no such card on the dashboard. What the panel gives you is some of the inputs: the number of people who started in the period, the paperwork behind the software line item, and the evaluation credits spent in the period. Line items like posting budget and team time stay outside the panel. This guide keeps the two apart all the way through.

• The formula is fixed, the line item list is your decision. Take it once and keep it in writing.
• If the numerator and the denominator do not share a period boundary, the number is not wrong, it is meaningless.
• Keep panel sourced inputs and manually tracked inputs in separate columns; repeating the calculation next period gets easy that way.

Write the period down first: the first and last day of the month, or the first and last day of the quarter. That boundary has to hold for the costs and for the hire count alike. Choosing the period is the most frequently skipped decision in this calculation, because it looks self evidently right.

The only way to pull the number of people who started in the period out of the panel is the export screen under Settings. Pick Employees as the source and a Start Date range appears among the filters; set its start and end to exactly your period boundary. Department, position and employment type filters sit on the same screen, so a broken down calculation is narrowed there.

The three columns the calculation needs are Start Date, Department and Position. The employee export can also carry personal data columns, national ID included, so leave every column you do not need unchecked. The screen asks for a permission as well; a user without export rights never sees the tab.

The line item list is the real decision here, and there is no single correct list. What matters is that the list is written down and does not move between periods. Include an item this quarter and drop it the next and the two numbers stop being comparable; that is exactly where the metric loses its value.

The second decision is the tax base. GoTeam list prices include VAT: the amount charged to your card is the amount on the price page. The invoice splits that same amount into net plus 20 percent VAT, so the net behind the 1,499 TL monthly fee is 1,249 TL. Accounting keeps its records on the net figure, so a calculation built on VAT inclusive amounts leaves the two tables disagreeing.

Practical rule: write the base you built on into the heading of your table. The net base gives the right answer when you talk to accounting, the VAT inclusive base when you talk about cash going out. Both are defensible; a mix of the two is not.

The first part of the software line item is the plan fee, and its paperwork sits on the subscription screen under Settings: plan, status, period dates, invoices and payment history in one place. A company billed monthly reads the period fee straight off; on annual billing the yearly fee is divided across the periods, because annual means paying for ten months and the amount per month lands below the monthly list price.

The second part is the extra evaluation packs. When the plan allowance runs short, extra packs are bought on that same screen, and the panel prints the fee divided by the credit count on top of each card as a per candidate unit price: 100 credits at 750 TL make 7.50 TL, 500 credits at 3,250 TL make 6.50 TL, 1,500 credits at 8,250 TL make 5.50 TL. The panel flags the lowest unit price itself, you never have to hunt for it.

The two parts are added together. If an extra pack was bought in the period, the software line item does not end at the plan fee; the pack fee is a cost of that same period, even when its credits spill into later months.

To see how much of the software line item was actually used, look at the credits spent. The AI Evaluation panel on the dashboard shows credits spent in the period, credits remaining (split into plan and extra pack), the operation count and the distribution per operation. When the remaining balance drops below the 10 percent threshold of the monthly allowance the panel warns you; the gauge on the subscription screen opens the extra pack link at the 20 percent threshold. The two thresholds differ on purpose: one is an early warning, the other is the buying moment.

The detail lives on the AI usage screen under Settings. It carries 7 day, 30 day, 90 day and all time ranges, credits spent and refunded, the operation breakdown and a paginated credit ledger. The ledger only appends: period allocation, spend, period end voiding and refund each sit as their own line and never change retroactively. Verifying credit consumption one line at a time is done here.

The unit fits in one sentence: one candidate evaluation is one credit. Every paid AI operation spends exactly one credit, whichever operation it is. There are routes that spend zero credits too, and they are covered by the subscription fee: the integrity gate, natural language search, the requirement check, embeddings and fit recalculation.

• Operations that spend one credit: analysis of an incoming application, scoring a pool candidate against another posting for the first time, AI posting generation, interview questions, answer analysis and the position rubric.
• Fit against the posting a candidate applied to is part of the application analysis; a single application is never charged twice.
• While AI analysis is switched off in company settings the line never runs: no credit is spent, the breakdown stays empty and evaluation consumption inside the software line item is zero.

Divide the line item total by the headcount you found in Step 1. The result is that period's cost per hire, and it is not a target on its own; its meaning appears once you set it beside a previous period calculated with the same line item list.

If you also want the software line item to have its own unit cost, divide by the capacity you paid for, not by the credits you used. When only 40 of a monthly 400 credit allowance were spent, cost per evaluation cannot be found by dividing by 40: the remaining 360 credits expire at period end, and they were paid for as well. Dividing by credits used makes a quiet month look like an efficiency win.

The last step is recording the calculation: period boundary, line item list, tax base and where each number was read from, all on a single page. What keeps the next quarter from turning into an argument is not the formula, it is that record.

Splitting the inputs into two columns ends the hunt for them every period. The panel sourced ones are numbers you can bound with a date range and back with a document. The manually tracked ones are costs the panel holds no record of; in most companies the largest share of the total is posting budget plus team time, so underrating that column breaks the calculation.

The table below says which input comes off which screen and where its limit is. Do not skip the limit column: a number existing does not mean it exists in the shape you need.

Interviewer time is the second largest line item in most calculations and has no measured counterpart in the panel. The interview record keeps the planned duration in minutes, defaulting to 30 minutes, and stores the interviewer emails as a list. What comes out of it is the planned duration; how many minutes the interview actually ran is not in the record.

Here is what you can do: multiply the number of interviews in the period by the planned duration on the record and by the interviewer count, then multiply the result by an internal hourly cost. What you get is an estimate, and the note on your calculation has to say so. Preparation and evaluation time sits outside that product; if you want it in, set a fixed extra amount per interview and write that down as well.

The seven mistakes below are where this calculation breaks most often. What they share: none of them raises an error, every one of them produces a plausible looking number. So you check them once while setting the calculation up, then keep the note next to the line item list.