Website Analytics: What to Track

Most small business websites have analytics installed and nobody looking at them. The tracking code went in during the build, a monthly email arrives with a number in it, and nothing about how the business operates has ever changed as a result. That is not a data problem. It is a question problem.

This guide is about the small number of things worth measuring on a small business website, how to set them up, and how to read them without drawing conclusions the data cannot support. Six useful numbers beat sixty you never open.

Start with questions, not tools

The standard sequence is to install analytics and then wonder what to do with it. Reverse it. Write down the three decisions you would make differently if you had better information, then work out what would need measuring to inform them.

For most small businesses the questions are unglamorous and consistent. Is the money spent on advertising producing enquiries? Which services do people actually come looking for? Where do people give up before contacting us? Is the blog producing anything commercial?

Each of those maps to a measurement, and together they require far less setup than a full analytics implementation. A site tracking four things well is more useful than one tracking forty things nobody has interpreted.

The six things worth tracking

Notice what is absent. Total pageviews, average session duration and bounce rate all appear on every dashboard and none of them supports a decision on a small business site. They move for reasons you cannot interpret and they do not tell you whether anyone contacted you.

Setting up conversion tracking

This is the step most small business sites skip and it is the one that makes analytics worth having. Without it you are looking at traffic and guessing, which is how businesses conclude that a page performing well is performing badly.

1. Define what counts as a conversion. Form submission, phone tap, booking completed, email click. Be specific and keep the list short.
2. Set up an event for each one, then mark the important ones as key events in your analytics property. Google documents how to set up Analytics for a website.
3. Track phone taps on mobile, which most small business sites miss entirely despite calls being their main enquiry route.
4. Use a thank-you page where possible, because a page view is the most reliable conversion signal available.
5. Test every one of them yourself before trusting the data. Untested conversion tracking is wrong more often than it is right.
6. Re-test after any site change. Form plugins update, thank-you URLs change, and tracking breaks silently.

That last point matters more than it sounds. Broken conversion tracking does not produce an error; it produces zero, and zero looks like a marketing problem rather than a measurement problem. Businesses have cancelled campaigns that were working because of it.

Search Console versus Analytics

These are different tools answering different questions and small businesses routinely expect one to do the other’s job. Analytics tells you what happened on your site. Search Console tells you what happened in Google search before anyone arrived.

Only Search Console shows the queries people searched, your average position, and how many times you appeared without being clicked. That last figure is the most actionable data available to a small business, because a page with many impressions and few clicks usually has a title and description problem rather than a content problem.

The two will never agree on numbers, and this alarms people unnecessarily. They count different things, over different windows, with different definitions of a session. Use each for what it is good at and stop reconciling them.

“Analytics tells you what visitors did. Search Console tells you what the people who never became visitors were looking for. The second is usually more useful.”

Reading the numbers without fooling yourself

The most common analytics mistake is not a technical one. It is inferring causation from a coincidence, then acting on it. Traffic went up in March, we posted a blog in March, therefore the blog worked. Possibly. Or a competitor went offline, or seasonality, or a single article got shared once.

Protect yourself with three habits. Compare like periods rather than adjacent ones, because most small businesses are seasonal. Look at the trend over months rather than the change since last week. And before believing a change, ask what else happened at the same time.

Segment before concluding. An overall conversion rate that has fallen might be entirely explained by a burst of low-intent social traffic, while your search traffic converts exactly as it did before. The aggregate number told you something was wrong and pointed you in the wrong direction.

The small numbers problem

A site with 400 visits a month and 12 enquiries has a 3 percent conversion rate. Next month it has 9 enquiries and a 2.25 percent rate. That looks like a 25 percent decline and it is three enquiries. On numbers that small, ordinary variation looks like a crisis.

This is the single biggest interpretive trap for small business analytics, and it produces a specific failure: constant changes made in response to noise, none of which is given long enough to be evaluated properly.

Work in quarters rather than months for anything you intend to act on. Look at absolute numbers alongside percentages, because three enquiries is easier to think about honestly than 25 percent. And accept that with small samples you frequently cannot tell whether something worked, which is uncomfortable and true.

Where visitors really came from

Attribution on a small business site is less reliable than the reports imply. Someone hears about you from a neighbour, searches your business name, clicks, leaves, then returns two days later by typing the URL. Analytics will credit that enquiry to direct traffic, and the actual source was a conversation.

Direct traffic in particular is a catch-all. It contains people typing your URL, clicking a bookmark, arriving from an app, following a link in an email client, and anything the browser declined to pass a referrer for. Treating it as brand strength is optimistic.

The cheapest correction available to a small business is to ask. One question on your enquiry form, “how did you hear about us?”, produces information no analytics package can give you, and comparing the answers to your reports tells you how much to trust the reports.

Analytics and Canadian privacy law

Analytics data can constitute personal information, and Canadian businesses have obligations regardless of where the analytics vendor is based. The Office of the Privacy Commissioner publishes guidance on privacy laws in Canada, and it is worth reading before assuming a default configuration is compliant.

Practically, three things matter for most small businesses. Your privacy policy should describe what you collect and why, in plain language. You should not be collecting more than you need, particularly precise data you will never use. And where you use advertising or remarketing features, the disclosure obligations increase.

This is general information rather than legal advice, and obligations vary by province and by what you actually collect. Where you are unsure, confirm it with a qualified advisor rather than relying on a template. We cover the practical side in our guide to the Canadian website privacy policy.

A monthly review that takes fifteen minutes

Analytics only produces value if somebody looks. The realistic answer for a small business is a short, fixed monthly routine rather than an ambitious dashboard nobody opens.

That last line is what turns measurement into management. A review that produces no decision is a report; a review that produces one dated change is a process, and after four quarters you have an actual record of what worked.

Choosing an analytics tool

Google Analytics is the default because it is free, ubiquitous and integrates with Search Console and Google Ads. For most small businesses that combination is decisive and there is no strong reason to look further.

There are real alternatives and they solve a specific problem: simplicity and data handling. Lighter privacy-focused analytics tools collect less, present far fewer numbers, and are considerably easier to interpret. For a business that only wants traffic sources and conversions, that reduction is a feature rather than a limitation.

The trade-off is integration and depth. If you run Google Ads, or you need query-level search data joined to on-site behaviour, the Google stack does things the alternatives do not. If you run no paid advertising and want four honest numbers, a simpler tool will get looked at more often.

Whatever you choose, pick one and stay with it. Running two analytics packages in parallel produces two sets of numbers that disagree, and the time spent explaining the discrepancy exceeds any value either provides.

Campaign tracking without overcomplicating it

If you send email, run ads, or post links anywhere you control, campaign tags let you see exactly which of those produced visits. Without them, an email click and a social click can both land in the same undifferentiated bucket.

The mechanism is simple: add parameters to the end of a link identifying the source, the medium and the campaign. Analytics reads them and reports the visit accordingly. Three parameters cover most small business needs and the rest can be ignored.

Inconsistent tagging is the most common failure and it is silent. A report showing five variations of the same source is not five channels; it is one channel and a naming problem, and it usually means nobody can tell how the campaign performed.

Measuring whether your blog is working

Blog measurement is where small businesses most often conclude that content does not work, usually because they measured the wrong thing over the wrong period. Pageviews on an article tell you nothing about whether it produced business.

Ask three questions instead. Does the article rank for anything, which Search Console answers by query. Does it bring in visitors who go on to a service page, which the landing page and path reports answer. And does it produce enquiries, either directly or as an assisting step earlier in a visit.

Give it time. Content typically takes several months to accumulate search visibility, and an article judged after four weeks is being judged before it has been indexed properly and settled into a position. Review a quarter’s worth of articles at once, two quarters after publishing.

Then act on the pattern rather than the individual. If the articles answering practical customer questions produce enquiries and the general industry commentary does not, that is your content strategy telling you what to write next, which is more valuable than any individual article’s performance.

“An article with 40 visits that produced two enquiries beat one with 4,000 visits that produced none. Both are on the same report and only one is a business result.”

When a page gets traffic but no enquiries

This is the most common and most fixable analytics finding on a small business site. A page brings people in and nothing happens. The causes are a short list and they are diagnosable in about twenty minutes.

First, check intent. A page ranking for an informational query attracts people who are researching, not buying, and the correct outcome may be a newsletter signup or a return visit rather than an immediate enquiry. Judging it by enquiries misreads what it is for.

Second, check whether there is anything to do. A surprising number of high-traffic pages on small business sites contain no call to action at all, or one buried below several screens of text. If the page answers a question and then stops, the visitor leaves satisfied and unconverted.

Third, check the mechanics. Test the form on a phone. Test whether the phone number is tappable. Test what happens on a slow connection. Broken or awkward contact routes on your highest-traffic pages are far more common than anyone expects, which is why we treat form testing as part of routine website maintenance rather than a one-off.

Fourth, check the match. A visitor arriving from a query about repairs, landing on a page about installations, has not found what they were looking for. That is a content and internal linking problem, and the fix is usually to send them somewhere better rather than to change the page they landed on.

Setup errors that quietly corrupt your data

Work through that list once when analytics is set up, then again after any significant site change. Each item is a few minutes to check and each one, left in place, quietly makes every number afterwards less trustworthy.

What good looks like on a small site

Owners frequently ask what a good conversion rate is, and the honest answer is that published benchmarks are close to useless at small business scale. They aggregate across industries, traffic mixes and definitions of conversion that have nothing in common with your site.

Your benchmark is your own site last year. If 2.4 percent of visitors enquired last spring and 3.1 percent enquire this spring, that is a real improvement regardless of what any industry average says. Comparing to yourself removes every variable that makes external benchmarks misleading.

Where external comparison does help is in spotting something obviously wrong. A site converting at 0.1 percent almost certainly has a broken form, a mismatch between traffic and offer, or tracking that is not firing. That is a diagnostic use rather than a target.

Set targets in absolute numbers instead of rates. “Twelve enquiries a month by December” is a goal a business can plan around. “Improve conversion rate by 20 percent” is a number that can move because traffic fell, which is not the outcome anyone wanted.

Deciding who actually looks at it

Analytics fails at small businesses for an organisational reason more often than a technical one: nobody owns it. The agency assumes the owner reads the reports, the owner assumes the agency is watching, and in practice nobody opens anything for eight months.

Name a person and a date. Fifteen minutes, first Tuesday of the month, one named individual. That single decision does more for the value of your analytics than any dashboard configuration, and it is free.

Write the review down somewhere durable, even three lines. What the numbers were, what you changed, what you expect to happen. Six months later that record is the only way to tell whether anything you did worked, and memory is not a substitute.

If you outsource marketing, ask for the numbers in your own account rather than in a supplier’s report. A report prepared by the party being evaluated is a reasonable summary and it is not independent verification, and you should be able to see the source data yourself.

Measuring less and learning more

There is a growing case for collecting less analytics data rather than more, and it is not only a privacy argument. Smaller, simpler datasets get looked at. Comprehensive ones get admired and ignored.

Most small businesses could remove half of what they currently track and lose nothing. Demographic breakdowns nobody acts on, scroll depth on pages with no scroll, engagement metrics that measure attention rather than intent: each of these costs page weight, adds a disclosure obligation and produces no decision.

Ask of every metric whether removing it would change anything you do. Where the answer is no, removing it makes the site marginally faster, your privacy position marginally simpler and your monthly review marginally more likely to happen.

The same logic applies to how long you keep data. Retention set to the longest available option by default is a choice nobody made deliberately, and shortening it reduces both risk and the temptation to run analyses of three-year-old traffic that cannot inform anything current.

Frequently asked questions