Conversion rate

How to calculate conversion rate

Pick the action you want to count, and pick the base. A conversion is that action: an order, a lead, a trial. The base is who had a chance to do it: clicks, visits (sessions), or users. Divide conversions by the base. Write both on the slide.

Let's take a shoe workshop in Mexico as an example. They make leather shoes and sell them online to Mexico and the United States. In March the shop had 4,000 visits and 80 people placed an order. A visit is someone opening the shop website. 80 ÷ 4,000 = 0.02. That is a 2% visit-to-order conversion rate. The checkout (the pages where you pay) is the moment that counted. A visit that bounced on a missing size did not.

If tracking changes what counts as an order, the conversion-rate chart will jump even if the page did not. Let's take a shoe workshop in Mexico as an example. They make leather shoes and sell them online to Mexico and the United States. Name the base every time. 80 orders from 4,000 visits is not the same metric as 80 orders from 2,000 ad clicks.

Why conversion rate matters

Conversion rate sits in the middle of the paid funnel (the path from ad to order). CTR (click-through rate: how often a view becomes a click) gets people into the shop. Let's take a shoe workshop in Mexico as an example. They make leather shoes and sell them online to Mexico and the United States. Conversion rate decides whether a visit to the workshop's shop became a pair of shoes. CPA (cost per acquisition: ad spend divided by those orders) and ROAS (return on ad spend: credited sales divided by ad spend) are what happen after.

Conversion rate is also the diagnostic when the click looks cheap and the order does not. Let's take a shoe workshop in Mexico as an example. They make leather shoes and sell them online to Mexico and the United States. The workshop can pay $1 a click and still have a $100 CPA (cost per acquisition) if only 1 in 100 visitors buy. The auction did its job. The visit did not. That split tells you whether to fight the bid or fix the page, the size run, the shipping promise, or the audience.

How to read conversion rate

Always name the base when you quote conversion rate. These are not the same metric:

• Click to purchase measures whether an ad click became an order.
• Visit to purchase measures whether any shop session became an order.
• Impression to conversion measures a view of the ad, with no click required.

Split conversion rate before you average it. Let's take a shoe workshop in Mexico as an example. They make leather shoes and sell them online to Mexico and the United States. The workshop's Mexico traffic and United States traffic will not share a rate. A returning customer reordering a known size will convert higher than someone who landed on a boot that is sold out in size 42. Mobile vs desktop, new vs returning, brand search vs a cold Instagram ad: each is a different product. A blended 2% is a headline. The segment is the decision.

A landing-page test should hold traffic quality still. Let's take a shoe workshop in Mexico as an example. They make leather shoes and sell them online to Mexico and the United States. If the workshop changes who they target in the same week they rewrite checkout, conversion rate will move and nobody will know why.

Common conversion-rate mistakes

• Raising conversion rate by attracting easier, lower-value traffic. Let's take a shoe workshop in Mexico as an example. They make leather shoes and sell them online to Mexico and the United States. A 4% rate on $40 sandals can make less money than a 2% rate on $180 boots.
• Mixing view-through conversions into a click conversion rate. Someone who saw a shoe-workshop ad, did not tap, and bought later is not a click conversion.
• Rewriting the whole checkout because the blended conversion rate dropped, when the drop was United States shipping or a sold-out size.

If conversion rate drops, walk the funnel (the path from ad to order) before you rewrite the page.

• Did click-through rate (CTR: how often a view becomes a click) jump, bringing cheaper, worse traffic that never meant to buy shoes?
• Did the offer, the audience, the inventory, or the tracking change?

A page test only means something if the traffic mix held still. If you need the cost of the order, use CPA (cost per acquisition: ad spend divided by orders). Do not raise bids to fix a checkout problem.