How to calculate cost per lead

Cost per lead in the report is spend for a period divided by the number of leads in the same period. On a Yandex Direct strategy, “conversion price” is a different number: a limit you set, not what already happened. Mixing them means comparing a plan with a result and then being surprised they disagree.

In the report wizard this historical price sits next to the metrics CR, CPA and Conversions. Below is the same division by hand, on teaching numbers, and two reasons CPA rises even though the buyers did not change.

The division the report is doing

Suppose a month cost 90,000 ₽ and brought 30 leads. Cost per lead is 3,000 ₽. That is an example, so the division is easy to see. It is not a benchmark for a niche.

Take both numbers from one report. In Direct, the click date and the lead date diverge. In the report wizard a lead sits on the day of the click. In the pay-per-conversion report it sits on the day of the lead. How that makes a week look empty is covered in clicks without leads. Until the calendar is the same one, a quotient from two different exports is not a cost per lead.

Few leads make the quotient shaky. Direct writes that a strategy needs at least 10 goal completions a week in order to learn. Two leads in a month produce some price, and the next week can double it or zero the denominator. A budget decision is not made on two leads.

The price on the strategy is a limit, not a fact

On the “Maximum conversions” strategy you set a conversion price, a share of revenue, or a budget. The system tries to bring target visits inside that frame. The report’s CPA can land above or below the price you set. The report divides what happened. The strategy constrains what comes next.

Yandex separately shows how to think about whether the price you set is worth it, and that is not historical CPA either. Profit per customer, in their wording, is conversion value minus conversion price, and total profit is that difference multiplied by the number of conversions. In their example, 100 leads brought 50,000 ₽, so the value of one lead is 500 ₽ (how to choose a conversion price). Value is income from a lead. CPA is spend per lead. You can subtract one from the other. You cannot fold them into a single metric.

Where a rise comes from when no buyers were added

Take the same 90,000 ₽ and 30 leads, a price of 3,000 ₽. Suppose you judge that 20% of spend, 18,000 ₽, went to clicks with no buyer. The remaining 72,000 ₽ on the same 30 leads is 2,400 ₽. The report does not make this split. It still shows 3,000 ₽. The 20% is your estimate, not a field in the account. How not to invent it, and how not to add invalid clicks that were already returned, is covered in the budget-loss calculation. The traces an estimate can even come from are in signs of click fraud.

The same 3,000 ₽ appears if the goal undercounted leads. Suppose there were 40 real leads and 30 reached the report. Spend per real lead is 2,250 ₽, and the report shows 3,000 ₽. The direction is the same as with empty clicks: the price rose, the buyer did not get more expensive. CPA alone cannot tell them apart. First check that a test lead is recorded.

Click-through is not the direct cause. You can receive many cheap clicks that never become leads, and CPA rises while CPC stays where it was. “CTR looks fine, so the advertising is healthy” does not agree with the cost of a lead. How CTR is built is in what CTR is.

A visit check on the site does not calculate CPA and does not know your margin. It can shrink spend on empty visits without touching leads from real people, and then the quotient falls. It will not restore leads the goal failed to record. The denominator stays short, and the price keeps lying. If the check does not answer, the visitor is let through, and spend on that visit does not depend on the filter on the site.