How much traffic should a small-business website get?

There is no useful benchmark, because it depends entirely on what you sell and how often people need it. The question that does have an answer is whether the traffic you get produces enquiries. A local specialist with modest monthly traffic and steady enquiries is in better shape than a site with ten times the visits and none.

How long before I can judge a change?

Two to three months at typical small-business volumes. Less than that and you are reading noise as signal — which usually leads to reversing a change that was working.

Do I need paid analytics tools?

No. Free analytics plus a note of where each enquiry came from covers almost every decision at this size. Paid tooling starts to earn its cost when you are running paid campaigns and need to know which one produced the customer.

My traffic dropped but enquiries stayed the same. Is that bad?

Usually not, and sometimes it is an improvement — it often means you lost visitors who were never going to buy. Worry when the enquiries move, not when the chart does.

How do you know if your website is working?

One number decides it, and it is not visits. How to set up the minimum measurement, how to tell whether the problem is traffic or the page, and what to change first.

Key takeaways

One number decides it: qualified enquiries. Everything else is diagnosis, not a verdict.
Visits, bounce rate and time on page cannot tell you whether the site works — they tell you why it does or does not.
The minimum setup is three things: analytics, a way to know which enquiries came from the site, and a monthly look.
If traffic is low, the problem is upstream. If traffic is fine and enquiries are not, the problem is on the page.
Judge changes over months, not days. Weekly numbers at small volumes are noise.

What does a working website actually mean?

It produces enquiries you would want to answer. That is the whole definition, and it is deliberately narrow.

The narrowness matters because almost every reporting habit drifts away from it. Traffic went up, so things are improving. Bounce rate went down, so the design is better. Neither statement contains the only fact that pays your invoices. A site with modest traffic and a steady flow of enquiries is working. A site with impressive traffic and none is a magazine.

Vanity numbers and decision numbers

The difference is not that one is fake. It is that only one of them can change what you do next.

Decision numbers: enquiries received, how many were the kind of work you want, where they came from, and how many became customers.
Diagnostic numbers: visits by source, which pages get the visits, how far people scroll, what they do on a phone versus a laptop. These are useful — but only once a decision number has told you something is wrong.
Numbers that decide nothing: total pageviews, average time on page, follower counts. Comforting, unactionable.

The minimum setup, and nothing more

Three things. Resist adding a fourth until these have run for three months.

Analytics on the site, in your own account, with the form submission recorded as an event. Without that last part you are counting visits, not outcomes.
A way to know where an enquiry came from. The cheapest reliable version is asking, in one field or one sentence on the phone. Attribution tooling is worse than a human question at this volume.
A monthly note. Enquiries, source, what changed on the site. Ten minutes. This is the part everyone skips and the part that makes the other two useful.

Is the problem traffic, or the page?

Every website problem resolves into one of these two, and treating the wrong one is how budgets disappear.

Almost nobody arrives. Then no amount of button colour will help. The work is upstream: search visibility, your Google Business Profile, being present in the sources buyers and AI assistants read. Redesigning a page nobody visits is the most expensive way to change nothing.

People arrive and leave without contacting you. Now it is the page. Usually one of four things: the offer is not clear in the first screen, there is no price signal at all, there is no proof a stranger can verify, or contacting you takes more effort than the visitor is willing to spend.

Test which one before spending anything. Open your own site on a phone, on mobile data, and try to work out what you sell and what it costs in twenty seconds.

What to look at monthly — and what to ignore weekly

Once a month, four questions:

How many enquiries, and how many were the work you actually want?
Where did they come from — search, Maps, a recommendation, an AI assistant?
Which pages did people who enquired look at first?
What did we change last month, and did anything move?

Weekly numbers at small-business volume are noise. Three enquiries versus one is not a trend; it is a week. Judging a change before two or three months have passed is how good decisions get reversed for no reason.

The five-minute test

Before any tooling, do this. It finds more problems than most audits.

Open the site on your phone, on mobile data, not office wi-fi. Time how long until something readable appears.
Count the taps from landing to contacting you. More than two is a leak.
Read the first screen out loud. Does it say what you do, for whom, and where?
Send yourself a message through the form. Does it arrive, and to a mailbox someone actually reads?
Ask an AI assistant the question your customer would ask. Does anything of yours come back? If not, that is a visibility problem, not a design one — the mechanics are in how to make your website visible in AI.

What to change first

In order, because the order saves money. Make the offer and the price signal clear before touching design. Make contact take one tap. Add proof a stranger can verify — names, real photos, checkable reviews. Only then work on traffic, and only then consider paying for it.

Paying for traffic to a page that does not convert is the most reliable way to spend a budget with nothing to show. Fix the page, then buy the visitors.