More traffic or better enquiries: the website KPIs that matter

Say you run a company that designs and fits kitchens in one region. The figures below are invented for this illustration.

In March a post on your site about small-kitchen storage gets shared widely. Visits jump from 1,500 to 3,400. Your web report shows a steep line going up, and it feels like a good month. But enquiries stay at 14, and when you read them, two are from people in other countries asking where you bought a shelf.

In April you publish a clear page about what a kitchen quote includes, with photos of real measurements and a short form. Visits fall back to 1,600. Enquiries rise to 19, and 11 of them are homeowners in your region with a budget and a date in mind.

On a traffic chart, April looks like a decline. In the business, April is the better month by some distance. If you only watched visits, you would have drawn the wrong conclusion twice: you might have written more viral storage posts, and worried about the quote page.

A visit is someone loading a page. It says nothing about who they are, whether they could ever buy from you, or whether they left satisfied. Traffic rises for good reasons (a useful page starts appearing in search) and for useless ones (a post reaches people who will never be customers, or bots slip through). It also falls for harmless reasons, like a seasonal lull, or because you removed pages that only attracted the wrong visitors.

That does not make traffic worthless. When enquiries drop, traffic by source and by page is usually the first place to look for the reason. The point is the order: start from the outcome, then use traffic to explain it.

Most business websites have the same ladder, whatever they sell. Each rung is a different number, recorded in a different place, and each one means less than the rung above it.

Two things on that ladder are easy to misread.

An engaged visit is a low bar. In Google Analytics, a session counts as engaged if it “lasts longer than 10 seconds”, “has a key event”, or “has 2 or more screen or page views”. That is useful as a quality signal on traffic, and far from a sign of interest in buying.

A tap on your phone number is intent. A click on a phone link, email address or chat button records that someone tried to start a conversation. It does not record whether the call connected, or whether the person was a real prospect. Count these separately from enquiries sent, and never add them together.

A visit can turn into contact intent in a few different ways. Someone might tap your phone number, open their email app or start a chat. Each of these shows interest, but none of them is an enquiry yet. They simply mean the visitor has taken a step towards a conversation.

An accepted form submission is different from a contact-link click. The form system has accepted the visitor’s message or details. Check separately that the message reaches the right inbox or business system; acceptance alone does not prove delivery or suitability.

These two stages sit close together, and it is easy to mix them up. Keep them separate. Count contact intent and enquiries on their own, so you can see what is really happening and where to focus your effort next.

The right numbers depend on what your site is for. A rough guide:

• A service business that wins work by enquiry (a kitchen fitter, a law firm, an accountant): enquiries sent, qualified enquiries, and their self-reported sources. Keep these separate from measured channel totals unless a privacy-safe per-enquiry attribution method has been established. Pages that lead to enquiries are worth watching. Total visits are background.
• A business that takes bookings (a clinic, a hotel, a tutor): completed bookings, cancellations, and the share of bookings that came through the site rather than by phone.
• An online shop: orders, revenue, and the share of visits that end in an order. Shops have more numbers worth watching, covered in our guide to ecommerce KPIs.
• A supplier with a product catalog: quote requests, the products those requests mention, and repeat requests from existing customers.

Then pick no more than five numbers you will read every month. If a number would not change a decision when it moved, leave it off.

• Total page views. They rise when people click around confused, and fall when they find what they need on the first page.
• Time on page. A long time can mean someone was absorbed, or that they could not find the phone number.
• Social followers and likes. Useful for social media work. They are not website outcomes.
• Site-wide average position in Search Console. It averages searches that matter with searches that do not. Our guide to Google Analytics and Search Console explains why position should be read per search and per page.
• Speed scores on their own. Speed matters to people using the site, and Google’s Core Web Vitals give clear targets: Largest Contentful Paint within 2.5 seconds, Interaction to Next Paint of 200 milliseconds or less, and Cumulative Layout Shift of 0.1 or less, measured at the 75th percentile of real visits. They are worth checking, the way you check a building is safe. They are a health check, not a result.

A number is only as good as the moment it records.

• Count an enquiry when it succeeds. Count verified acceptance by the form system once, not the button click. Google’s tutorial uses a confirmation-page visit, but a page view alone can be replayed: protect against refreshes, direct visits and duplicate events. Check inbox delivery separately.
• Use the standard event names. Google Analytics recommends generate_lead for a lead, and has further recommended events for the later rungs, such as qualify_lead when a lead fits your criteria and close_convert_lead when it becomes a customer. Using them keeps your numbers comparable over time and with your own records.
• Mark the enquiry as a key event. Google defines a key event as “an event that measures an action that’s particularly important to the success of your business.” That is exactly what this is.
• Keep personal details out. Google’s policy is that no data should be passed to Analytics that Google “could use or recognize as personally identifiable information.” Names, emails, phone numbers and the text of an enquiry stay in your inbox or customer records, never in analytics events or page addresses.
• Know who is not counted. With basic consent mode, Google tags stay blocked when consent is refused. Advanced consent mode can send cookieless signals. Behavioral modeling requires an eligible property and compatible reporting settings; do not assume a small site qualifies. Check the setup and report before calling totals modeled estimates. Treat recorded totals as incomplete, not an exact count of people.

The practical setup, step by step, is in our guide on how to track leads from your website.

Whether an enquiry was any good, and whether it became a customer, is something only you know. Analytics cannot read your inbox or your bank account.

So the top of the ladder needs its own record. For a small business that can be a spreadsheet with one row per enquiry: date, how they found you, what they asked for, whether it was a fit, whether you quoted, whether you won it. A larger business might keep this in a CRM. Either way, it takes a minute per enquiry and it answers the questions a traffic report never can.

Add one field to your enquiry form or your first reply: “How did you hear about us?” It is imperfect, because people forget and simplify, but it catches what tracking misses, like the customer who heard about you from a neighbor, searched your name and filled in the form.

Many business websites get a modest number of visits and a few enquiries a week. At that scale, numbers jump around for no meaningful reason. Three enquiries one month and six the next may be chance.

• Compare like with like. This month against the same month last year, or a three-month total against the previous three months.
• Read the enquiries themselves. Ten enquiries read carefully tell you more about the site than a percentage calculated from them.
• Do not expect statistical tests to settle small questions. Testing two versions of a page needs more traffic than most small sites have. The heatmaps section below covers what you can learn from visitor behavior without a formal test, and what it cannot prove.
• A new site starts with no history. Record that the measurement works from launch day, then build your first baseline from what you observe. The absence of past data is not evidence of a problem.
• Some reports need traffic before they show anything. Search Console’s Core Web Vitals report, for example, says “No data available” when there is “not enough data” from real visits to say anything meaningful.

A heatmap is a picture of a page with colors laid over it, showing where many visitors clicked, tapped or scrolled. Red means more, blue means fewer. It can show you what happened on a page, which makes it a good way to find questions worth asking. It cannot answer them.

• Why anyone did anything. A heatmap records behavior, not intent. To learn why, watch people use the page or ask them.
• Whether a click was a good outcome. A click on your phone number can be a visitor ready to call or an accidental tap. A cluster of clicks on a menu can mean people are finding their way, or that they are lost.
• That a scroll is a read. A scroll map shows how far down the page people got. A scroll depth of 80 percent fits careful reading and flicking to the bottom equally well.
• What different visitors do. An aggregate hides groups. People on phones and people on desktops see different layouts, so filter by device before you read anything.
• Whether the page makes money, or whether a change worked. A click map says nothing about whether a visitor then sent an enquiry. Two heatmaps from different weeks, with different traffic, are not an experiment. Judge a change by enquiries over a long enough period, and keep a note of what you changed and when.

A map built from a few dozen visits is a few dozen people, and a few of them can create a hot spot on their own. Heatmaps earn their place on a page where you already have an unexplained pattern, not on every page.

If you use one, keep personal details out of it. Mask any names, addresses or order details your pages show, tell visitors in your privacy notice that you use a behavior tool, and, where the law requires consent, pass the consent choice to the tool. Ask a privacy adviser if you are unsure.

Most owners need one page, read in ten minutes:

1. Enquiries sent (or bookings, or orders), this month and the same period last year.
2. Qualified enquiries, and how many became customers.
3. Self-reported sources of qualified enquiries, kept separate from aggregate Analytics channels and any separately verified measured attribution.
4. Pages associated with recorded enquiry events, where the report supports that comparison; this does not establish the page or channel of every qualified enquiry.
5. One thing that looks wrong, and what you will check.

If a line on that page is unchanged for months and never prompts a decision, swap it for something that would.

Choosing better numbers does not produce better enquiries by itself. It shows you where the site is failing so someone can fix it. It also costs a little discipline: the qualified and customer rungs depend on someone recording each enquiry, every time. And no measurement setup sees everything. Some visitors are never tracked, and some customers find you in ways no report records. Aim for numbers good enough to decide with, not for a perfect picture.

If you would like your site set up to record the right moments and report them simply, our analytics service explains what we configure and what you receive. For the tools themselves, read Google Analytics and Search Console, explained simply. If enquiries are low and you want to know why, our guide on why a website is not bringing enquiries is the natural next read.

When you want to talk about your own site, you can discuss your project. More guides are in the Guides section.

Analytics Help, Engagement rate and bounce rate: https://support.google.com/analytics/answer/12195621

Analytics Help, About key events: https://support.google.com/analytics/answer/9267568

Analytics Help, Recommended events (lead generation events): https://support.google.com/analytics/answer/9267735

Analytics Help, Tutorial: Set up a key event: https://support.google.com/analytics/answer/12966437

Analytics Help, Best practices to avoid sending personally identifiable information: https://support.google.com/analytics/answer/6366371

Analytics Help, Consent mode on websites and mobile apps: https://support.google.com/analytics/answer/9976101

Analytics Help, Default channel group: https://support.google.com/analytics/answer/9756891

web.dev, Web Vitals: https://web.dev/articles/vitals

Search Console Help, Core Web Vitals report: https://support.google.com/webmasters/answer/9205520