Diagnosing a falling conversion rate, in the right order

A falling conversion rate gets people’s attention fast. Someone senior notices the line going down, and within an hour there are suggestions on the table: change the button colour, redesign the product page, add a discount. Those might help eventually. But in most cases, the page isn’t what changed.

This guide gives you a diagnostic order that finds the real cause quickly and stops you redesigning a page that was working fine.

▸ Definition — Conversion rate: the share of visits (or visitors) that complete a goal, such as a purchase. 50 orders from 2,000 sessions is a 2.5% conversion rate.

Remember it’s a fraction

A conversion rate is a fraction: conversions on top, visits on the bottom. It can fall for two very different reasons:

• Fewer conversions for the same visits (something is putting people off), or
• More visits that were never likely to convert (the mix of visitors changed).

Think of a restaurant. If the share of passers-by who come in and eat drops, maybe the food got worse. Or maybe a bus stop opened outside and hundreds more people now walk past without ever planning to eat. The restaurant is the same; the crowd changed.

So before you touch anything, work through four steps in this order.

Step 1: Is the tracking telling the truth?

A surprising number of conversion “drops” are measurement problems. Check:

1. Real orders vs. reported conversions. Compare your analytics with the shop’s or CRM’s actual orders for the same period. If real orders are steady, the problem is tracking.
2. Recent changes. Was the site updated, a cookie banner changed or a checkout app installed around the date the drop began? Any of those can break conversion tracking.
3. Consent effects. If fewer visitors accept analytics cookies, both visits and conversions can be under-counted, and not always equally.

▸ Margin note — Write down the date the drop started. Almost every diagnosis gets easier when you can line it up against a timeline of site changes, campaigns and promotions.

Step 2: Did the traffic mix change?

If tracking is fine, compare the period before the drop with the period after, broken down by:

Often you’ll find that each segment converts about as well as before, but the mix shifted towards lower-converting segments. That’s not a page problem. It may not even be a problem at all, if those new visitors are cheap and some convert later.

▸ Example — A mix shift. A fashion brand’s conversion rate drops from 2.4% to 1.8%. On closer look, a giveaway on Instagram doubled social traffic, which converts at under 1%. Search and email convert exactly as before. Total orders are actually slightly up. Illustrative numbers.

Step 3: Did something outside the site change?

Some causes sit outside your control but still need to be in your explanation:

• Seasonality: compare with the same weeks last year, not just last month.
• Price or stock: did a best-seller go out of stock, or did prices rise?
• Competitors: is a competitor running a big sale?
• Delivery promises: did delivery times or fees change?

Step 4: Now look at the page and checkout

Only when the first three steps don’t explain the drop should you look for a problem on the site itself. Work from the end of the journey backwards, because problems near checkout cost the most:

1. Checkout: test it yourself on a phone and a desktop. Look for errors, failed payment methods, surprise fees or forced account creation.
2. Cart: has the add-to-cart rate held steady while cart-to-purchase fell? That points to checkout, delivery costs or payment.
3. Product pages: did page speed drop, images break or key information disappear in a recent update?
4. Landing pages: does the page still match what the ad or email promised?

Funnel reports in your analytics tool show where the extra people drop out. A drop concentrated at one step tells you where to look.

Take it to work

When everyone is pointing at the page, being the person who calmly asks “has the traffic mix changed?” is a quiet way to earn a lot of trust.