I averaged the CTRs of my two ad sets, so why does it not match the campaign's overall CTR?

Because a simple average treats a small ad set as if it were as big as a large one. The campaign's CTR is all clicks over all impressions, not an average of rates. The right way is to add up impressions, clicks, spend and conversions across the ad sets first, then enter the totals in the calculator, which gives a rate weighted by each ad set's size.

Published · 3 min read

The example that exposes the mistake

A campaign has two ad sets. The first is small: 1,000 impressions, 50 clicks, a spend of 100 and 5 orders. The second is large: 9,000 impressions, 90 clicks, a spend of 400 and 6 orders.

Worked example

Ad set 1: CTR 5%, CPC 2, CPA 20.

Ad set 2: CTR 1%, CPC 4.4444, CPA 66.6667.

Simple average of the two CTRs: 3%.

Totals of 10,000 impressions, 140 clicks, spend of 500 and 11 orders: CTR 1.4%, CPC 3.5714, CPA 45.4545.

The gap between 3% and 1.4% is not a rounding detail: the simple average makes the campaign look more than twice as good as it is. Ad set 1 accounts for only a tenth of the impressions, yet the simple average gave it half the weight. The real campaign rate sits close to the large ad set, because most impressions came from it.

The rule: add raw numbers, never rates

  1. For each ad set, copy the four raw figures for the same period, not the ready-made rates.
  2. Add impressions to impressions, clicks to clicks, spend to spend and conversions to conversions.
  3. Enter the totals in the Campaign performance calculator and all five metrics come out correctly weighted.
  4. Check that the resulting CTR lies between the two ad sets' rates and nearer the larger one. If it does not, recheck your sums.

The same trap with CPA and CPC

Take the CPAs of 20 and 66.6667 and average them, and you get a number that is not 45.4545, the true cost of each order in the campaign. A weighted CPA is total spend over total orders, and that is what the calculator gives you from the totals. The same goes for CPC, conversion rate and CPM: every rate or average has to be recalculated from the totals.

The trap returns when you combine months: the average of three monthly CTRs is not the quarter's CTR if spend differed from month to month. Keep each month's raw figures, add them up for the period you want, then calculate. If spend comes from scattered invoices, the Expense summary generator can total it by month before you enter it.

When a simple average is acceptable

Only when the ad sets have exactly the same denominator, for instance the same number of impressions each. That is rare in practice, and it is easier to always add up the raw figures than to think about the exception. When you move from metrics to profit, work out return from total sales and total spend in the ROAS calculator with the same logic.

Common questions

The platform already shows the campaign's CTR. Why calculate it?

If it is available for the same period and level, use it. Calculating from totals is what you need when you combine ad sets, campaigns or platforms that no single report adds up for you.

Can I add up figures from two different platforms?

Arithmetically, yes, but what counts as a click or a conversion may differ between them, so note that next to the result.