What to measure on a small site

Low volume changes what is reliable

A conversion rate calculated from four enquiries in a month is not a rate, it is four events. Any percentage built on a small number moves wildly for reasons that have nothing to do with your site or your business.

That rules out most of what a dashboard emphasises. Bounce rate, session duration and engagement percentages need volume to mean anything, and at a few hundred visits a month they simply do not have it.

What remains reliable is counting. How many enquiries, from where, about what. Those are whole numbers rather than ratios, and a whole number of three is genuinely three rather than an estimate.

Start with the outcome

The site exists to produce enquiries, bookings or sales. Everything else is a step toward that, and measuring the steps while ignoring the destination is the commonest mistake small businesses make with analytics.

Count enquiries first, every month, as a plain number. Then count how many were worth having, because twenty enquiries for work you do not take on is a worse month than five for work you do.

Our page on analytics that matter covers the tooling side. This page is about deciding what success means before you go looking for a number to represent it.

Ask, because the data will not tell you

Analytics can say somebody arrived from a search. It cannot say they had already heard your name from a neighbour and searched for you specifically, which is a completely different kind of customer.

So ask every new customer how they found you, and write the answer down. Twenty of those answers teach a small business more than any dashboard, and the exercise costs one sentence per conversation.

Expect the answers to contradict your analytics, and trust the answers. Attribution is genuinely hard at any scale and it is close to meaningless at small volume, where a handful of visits decide everything.

The measurement plan

The list below is a complete measurement plan for a small business site. It fits on one page, takes an afternoon to establish, and then costs twenty minutes a quarter to maintain.

• Enquiries received, counted as a number
• How many were work you wanted
• How each customer says they found you
• Which pages people arrive on
• What they searched to reach you
• Which pages nobody ever reaches
• Whether the contact route is still working

Watch the pages, not the totals

Total visits tell you almost nothing actionable. Which pages people arrive on tells you what your site is actually known for, and it is frequently not what you expected or intended.

Equally useful is the opposite list. Pages nobody ever reaches are either answering a question nobody asks or are invisible, and both are worth knowing before you write another one like them.

Our page on how many pages covers the wider judgement. A page with no arrivals after a year has given you its answer.

Compare against yourself, a year apart

Month against month is noise at this volume. A quarter compared with the same quarter last year removes the seasonality and is long enough that a real change shows through the variation.

Write down the number and one sentence about what you concluded. The conclusion is what you will want in a year, and nobody ever goes back to reread a saved report to reconstruct it.

Resist comparing yourself to published averages. They are computed across businesses nothing like yours, and a figure that means something for a national retailer means nothing for a local trade.

What to do with what you find

Each quarter, pick one thing. A page that gets arrivals and no enquiries, a question customers keep asking that no page answers, or a form that is losing people partway through.

Then change it and leave everything else alone. Changing several things at once means the next quarter cannot tell you which one worked, which is a genuinely useful discipline at low volume.

Our page on updating content covers what those changes usually are. Measurement without a change is a habit rather than a tool.