Conversion Rates on a Site With Few Visitors

Half the rate, and nothing changed

The monthly report says the site's conversion rate fell by half. The owner wants to know what broke, and whether the home page needs redoing.

On a site with four hundred visits a month, that fall can be the difference between eight enquiries and four.

Four enquiries is not a trend. It is four people, each with their own reasons for getting in touch that month or not.

Small sites need a different way of reading their own numbers, and it starts with seeing how little a monthly rate can carry.

What the rate is made of

Analytics works out a goal conversion rate by dividing the number of goal completions by the number of visits.

On a large online store, hundreds of orders a month keep that figure steady. On a small business site, the top of the fraction is a handful, so each enquiry moves the rate a long way.

A rate can only be as steady as the smaller of its two numbers, and here the smaller one fits on one hand.

Why small counts swing

• Chance, since enquiries arrive unevenly even when nothing changes.
• Seasons, which move visits and enquiries at different times.
• Junk visits from bots and spam referrers, inflating the bottom number.
• Staff visits counted as if they were customers.
• Calls and emails that Analytics never sees at all.
• Duplicates, such as a thank-you page reloaded twice.

The first explains most of it, and owners find it the hardest to accept. A site averaging five enquiries a month, with nothing changing, can easily produce months of two and months of eight in an ordinary year.

Count the enquiries themselves

The rate is a calculation. The enquiries are the real thing, and on a small site they can be counted one by one.

Take them from the inbox, the form's stored entries and the phone, not from a report. Note each one's date and route, and leave out spam and sales pitches.

Counted this way, a bad month often turns out to be a quiet month for forms and a busy one for calls.

Longer periods, fairer comparisons

Add months together. A quarter smooths out much of the chance; a rolling twelve-month total, recalculated each month, smooths out the seasons as well.

Compare a period with the same period last year rather than with the month before. A landscaper's November tells you nothing when set beside its May.

A rolling total moves only when something lasting has changed, which is exactly the kind of change worth acting on.

When a month does mean something

With counts this small, a single month has to fall well outside everything seen over the past year before it means anything on its own.

A run is more telling than a dip. Three months in a row below last year's range deserve a closer look.

A zero is different again. A month with no form enquiries at all usually means something broke: the form itself, the address its notifications go to, or a spam filter catching them.

Test the form before doubting the website.

A worked example

A hydraulic repair shop that rebuilds cylinders, pumps and valves for logging, farm and construction equipment had a site that drew a few hundred visits a month and carried a contact form.

In August the owner opened Analytics and saw July's conversion rate at well under half of June's. He asked his web company to quote for a new home page.

The web company went through the inbox with him first. June had brought seven form enquiries and July three, while the phone log for July was busier than June's.

July's visits were also up, partly from a spam referrer sending fake visits, which made the rate look worse still.

Over the previous twelve months, form enquiries had ranged from two to nine a month, with no trend either way.

The redesign was postponed. The shop filtered out the spam referrer, started a monthly tally of calls and forms, and agreed to look again at the end of the year.

Calls belong in the count

For many trades most enquiries arrive by phone, and urgent ones almost always do. A rate built only on forms misses them entirely.

Each tap on the phone number can be sent to Analytics as an event, covering calls begun on a phone from the site itself. It cannot see the visitor who wrote the number down and rang from the office.

A tally at the phone, with a note of how each new caller found you, fills that gap for nothing.

Decisions without enough numbers

Testing two versions of a page against each other, with a tool such as Google Optimize, needs volume. With a handful of enquiries a month, a fair test could take years to show a real difference.

So on a small site, decisions come from judgement and from customers: watching somebody use the site on a phone, asking callers what they looked for, reading the enquiries themselves.

Some changes need no proof. A phone number that is hard to find, or a form with fourteen questions, is worth fixing because it is plainly wrong, not because of a dip.

The counter-case

Sites with thousands of visits and dozens of enquiries a month can read a monthly rate with more confidence, and an online store with many orders more still.

A sudden fall to nothing is worth checking the same day on any site, and a long, steady decline over a year means something whatever the volume.

But on a site with a few hundred visits, count the enquiries, compare longer periods, and do not redesign because of one month.

The monthly check

1. Count forms, calls and emails each month.
2. Keep a rolling twelve-month total.
3. Compare with the same month last year.
4. Test the form whenever a month shows zero.
5. Filter out spam referrers and staff visits.
6. Note site changes as Analytics annotations.
7. Wait three months before judging any change.

Step two is the one that settles arguments. A single month can say almost anything on a small site, while a twelve-month total only moves when something real has changed.

Why the path from an ad to an enquiry is getting harder to follow in some browsers is covered in Safari 12 and the cookies it blocks.

Frequently asked questions

West Coast Media Solutions Inc. provides web design, web development, hosting, digital marketing, and business consulting to organisations across Canada, drawing on more than twenty-five years in the field.