Conversion rate

Conversion rate is the share of visitors who did the thing you wanted them to do. Divide the people who did it by the people who could have, and that is the whole calculation. Everything difficult about the number happens before the division: deciding what counts, and making sure it is counted properly.

In short

• It is the share of visitors who did the thing you wanted.
• The number is meaningless until you have defined what that thing is.
• Published benchmarks are close to useless, because nobody counts the same way.
• A rising rate on falling traffic is usually a business getting smaller.

The part people skip

The arithmetic is trivial and it is not where the mistakes live. The mistakes live in the definition.

Before there is a rate, somebody has to decide what a conversion is for this business. A restaurant might count a table booked. A clinic might count an appointment request. A shop counts an order. These are all reasonable, and they produce numbers that cannot be compared with each other, because they are measuring different events at different distances from money.

Businesses that skip this step end up with a rate assembled by whoever installed the analytics, which usually means everything that could be counted is being counted. Newsletter signups sit alongside completed orders, and the resulting average describes nothing anybody would act on.

Choosing what to count

A short list works better than a long one, and the test for each item is whether you would be pleased to receive it.

• The thing you actually sell, first and separately. Orders, bookings, submitted quote requests.
• The serious step before it, if there is one. A conversation started, a form part-completed, a brochure requested.
• Nothing else as a headline. Scroll depth and time on page can be recorded, and they should not sit in the same list as an order.

Everything you mark counts towards the rate, so a generous definition produces a flattering number and a useless one. This decision belongs in tracking, and it needs making before a campaign runs rather than afterwards.

Why benchmarks mislead

Published averages for a trade circulate widely and they are worth very little, for a reason that has nothing to do with the trades involved.

Two businesses in the same industry will define a conversion differently, sit at different distances from the buying decision, and draw traffic from different places. One counts newsletter signups from a blog audience. The other counts signed contracts from search traffic. The first will report a rate many times higher while earning far less, and both figures are correct.

The only comparison that carries information is against yourself, on the same page, counting the same event, over a period long enough to see past the noise.

The trap in a rising number

A rate can improve for a reason you would not have chosen, and the direction of travel looks identical either way.

Consider a business that stops advertising broadly and keeps only its brand searches. The rate climbs sharply, because the remaining visitors were people who already knew the name and were ready to act. Nothing has improved. Fewer people arrived, fewer became customers, and the headline number went up.

This is why the rate should never be read on its own. Put it next to the number of visitors and the number of conversions, and read all three together. A rate rising while conversions fall is a business shrinking politely.

Where the number actually moves

Once the definition is honest, the rate responds to a fairly short list of things.

• Whether the page matches what was promised. The largest single factor for paid traffic, covered on the landing page entry.
• How much you are asking for. Every extra form field costs you some share of the people who started.
• Whether the objections are answered. The four or five things people hesitate over, written out rather than avoided.
• Speed on a phone. People leave before they have read anything, and the cause never shows up as a content problem.
• Whether the next step is obvious. A page with three equally weighted actions has effectively offered none.

Reading it at small scale

Most Philippine SMEs are not running the traffic volumes the advice online assumes, and the standard guidance breaks down accordingly.

At low traffic, the rate is dominated by chance. Going from two inquiries to three in a week is a fifty per cent improvement on paper and means nothing at all. Tests that would settle the question need more visitors than the business will see in a month.

The workable approach is to look monthly rather than daily, to compare year on year against the same month last year when the trade is seasonal, and to treat the rate as one input alongside what customers say on the phone and in chat. Judgement is doing more of the work here than statistics, and pretending otherwise leads to a great deal of pointless redesign.

When the sale does not happen on the site

For a large share of Philippine businesses the website introduces the customer and the sale is agreed somewhere else, usually in Messenger.

That changes what the rate is measuring. It is no longer the share of visitors who bought, it is the share who started a conversation, and those two numbers behave differently. A page can be excellent at starting conversations and the business can still be losing most of them at the reply stage, which no analytics tool will show. Count the conversation as the conversion, then measure what happens after it as a separate step with its own number.