Note: Page views can only be measured on the Plus plan with the page view setting enabled.
Q: In the A/B test CSV report, are page views counted per test?
A: No. Page views in the CSV are counted per variant (one row each), not per test. Each row shows how many times the target page was loaded while that variant was assigned. Because each page view is recorded under exactly one variant of a test, adding up all the variant rows of the same test gives the page views for that page (for tests that do not use a control — see "Tests that use a control" below).
Page views are not the number of times the display was shown. To see how many times the display was actually shown, check the Impressions column.
Details
Example
Three A/B tests (Test 1–3) target the same URL, and each test has two variants (A and B). The page was loaded 700 times in one day.
Test | Variant | Page views |
Test 1 | A | 352 |
Test 1 | B | 348 |
Test 2 | A | 341 |
Test 2 | B | 359 |
Test 3 | A | 355 |
Test 3 | B | 345 |
Page views for the URL = Test 1 A + B = 352 + 348 = 700
Tests 2 and 3 also add up to 700, because all three tests target the same URL.
The page views for that day are 700, not about 350 (one variant row).
The total of all rows (2,100) counts the same page views three times (once per test), so it cannot be used as the total number of page views.
How to read the CSV
Page views of one test: Add up the variant rows of that test only.
Page views when several tests target the same URL: Use the total of any one test. Do not add up the totals of multiple tests.
Number of times the display was shown: Use the Impressions column, not the Page views column.
Tests that use a control
Important: The control (the variant that shows no display) does not appear in the report or the CSV. If your test uses a control, the page views assigned to the control are not included in any row. In that case, the total of the variant rows is lower than the actual page views of the page. You can check the control's count on the A/B test screen.