Website conversion checklist for local service businesses

Decide what counts as a conversion, then record a baseline

Before you check anything, write down the one action that means 'this person wants to hire us', plus one softer action for people who aren't ready yet. For most local service businesses the main action is a phone call or a quote request; for clinics and salons it's usually a booking. The softer action takes less nerve: a callback request, an SMS link or a short question form. Every page you check should make the main action obvious and the softer one easy to find, with nothing else competing for the tap.

Your conversion rate is the share of visits that end in one of those actions: leads divided by visits to the page, over the same period. Google Ads Help uses the same logic for ads, defining conversion rate as conversions divided by ad interactions, with the example of 50 conversions from 1,000 interactions giving 5%. Count calls and forms together. A practice that takes most bookings by phone can show a tiny form rate and still be flat out.

Treat published 'average conversion rate' figures with suspicion. Most come from online shops, where the sale happens on the page, and they say little about a tradie whose customer rings, waits for a quote and decides a week later. Your own numbers, measured the same way each time, are the only benchmark that tells you whether a change helped.

So record a baseline before you touch the site: at least four ordinary weeks, avoiding public holidays and one-off promotions. Keep it in a spreadsheet with these rows:

Check the first screen and mobile layout on a real phone

Start with the phone, because that's the version Google reads. Google's mobile-first indexing documentation says it uses the mobile version of a site's content for indexing and ranking, and won't load content that needs a swipe, click or typing to appear. So primary content that only loads after a tap is weaker for search and for the reader.

The first screen has one job: tell a stranger what you do, where you do it and how to reach you, before they scroll. Google has retired its old Mobile-Friendly Test, so use your own thumb. Take an ordinary phone on mobile data, open each service page and work through these checks:

• The headline names the service and the area in plain words, such as 'Blocked drains cleared across the Sunshine Coast', in place of a slogan that could belong to anyone.
• A Call button and a quote button are both visible on arrival, and stay within reach as the page scrolls, with no trip back to the top.
• The page fits the screen. web.dev's responsive design guide says pages need a viewport meta tag, and that making people scroll sideways or zoom out causes a poor experience.
• Every phone number is a tap-to-call link. web.dev's click-to-call guidance recommends coding each number as a tel: link in international format, which for Australia means +61 followed by the number without its first 0.
• Buttons suit a thumb. web.dev recommends touch targets of about 48 pixels, spaced about 8 pixels apart, and WCAG 2.2 sets 24 by 24 CSS pixels as its AA minimum.
• Nothing covers the content on arrival. Google's guidance on intrusive interstitials recommends small banners over full-page overlays, so drop the pop-up that hides your phone number.
• Each link from outside the site lands on the matching page. Your Business Profile website link, ad links and social posts about a service should open that service or location page, so a visitor doesn't land on the homepage and start hunting. Tag those links so you can see which ones convert. GA4 Help says that when you add parameters to a URL you should always use utm_source, utm_medium and utm_campaign, warns that missing parameters show up as (not set) in reports, and recommends consistent lowercase names so one campaign doesn't split across several rows. The Google Business Profile audit checklist covers the rest of the profile.

Measure speed and layout shift with field data

A page that stalls tests a visitor's patience, and a layout that jumps as images load is how people end up tapping the wrong button. Keep speed in proportion, though. Google's page experience documentation says good Core Web Vitals scores don't guarantee top rankings, and that Google still shows the most relevant content even when page experience is weaker. Speed is on this list for the person waiting on the page.

Read the numbers from field data where you can. PageSpeed Insights shows field data from real Chrome users over the previous 28 days beside lab data from one simulated load. Search Console's Core Web Vitals report uses the same real-user data and rates groups of similar pages Good, Need improvement or Poor.

A small local site may not have enough visitors for field data. PageSpeed Insights says it then falls back to data for the whole site, or shows none at all. If so, treat the lab result as a rough guide and fix the obvious causes: oversized hero images, sliders, video backgrounds and third-party widgets that load ahead of the content.

Google's web.dev team defines three Core Web Vitals. Its good thresholds are measured at the 75th percentile of page loads, split by mobile and desktop:

Put proof and answers where the decision happens

People letting a stranger into their home, or trusting someone with their health or money, want proof they can check. Google's helpful content guidance says trust is the most important part of E-E-A-T and asks whether it's self-evident who is behind a site's content. Put the proof, and the answers to the questions that stop people calling, close to the button. Check that each service page carries:

• Your licence or registration number, if your trade or profession has one, linked to the regulator's public search where one exists.
• Your ABN and the business name that appears on your quotes and invoices.
• Photos of your own work and people, in place of stock shots of strangers.
• The suburbs or region you cover, stated plainly on every service page.
• Genuine reviews, with the star rating shown next to the number of reviews behind it.
• How quoting works: on site or from photos, whether there's a call-out or inspection fee, and what the first visit involves. You can answer 'what happens if I call?' without publishing a price list.
• When you'll reply: a realistic response time beside the form, and what happens to an enquiry sent at night or on a weekend.
• A link to an About or team page that names the owner, shows real people and says where the business is based, since a cautious customer often opens it before calling.

Cut the friction between wanting help and asking for it

Friction is every extra step or doubt between 'I need this fixed' and a message in your inbox. Count the taps from landing on a service page to a sent enquiry, then remove what you can.

Friction also hides outside the form: a phone number saved as an image, a quote button that opens a generic contact page, a chat widget sitting over the Call button, a cookie banner covering the form. And if the Privacy Act covers your business, the form should tell people why you're collecting their details, for example through a short notice or link beside it (see the FAQ).

Reviews belong in this check too, because a review block that looks curated creates doubt. The ACCC's online reviews guidance applies to reviews on a business's own website as well as on review platforms. It says a business may mislead people if it suppresses or edits negative reviews, or removes genuine negative ones, and that a star rating should show how many reviews it's based on, for example 3 stars from 24 reviews. For collecting reviews in the first place, see the guide to reviews and honest customer feedback.

The same caution applies to urgency. The ACCC says any claim a business makes about its products or services must be accurate, truthful and based on reasonable grounds, and that the overall impression has to be accurate as well. A countdown timer that quietly resets, or a 'two spots left this week' banner that never changes, adds pressure and risk while giving the customer nothing to check.

For the form itself, web.dev's form guidance gives a clear set of rules:

• Ask only for what you need to reply. web.dev puts it plainly: don't ask for data you don't need. Name, phone, suburb and a short note about the job covers most service enquiries; access details, photos and measurements can wait for the callback.
• Use one name field, and one phone field that isn't split into parts.
• Give every field a visible label. Placeholder text used as a label disappears the moment someone starts typing.
• Set the input type and autocomplete (type tel and autocomplete tel for phone numbers) so phones show a number pad and fill in saved details.
• Show errors as people type, with the fix, instead of a list of problems after they press the button.
• Label the button with the outcome, such as Request a callback, rather than Submit.
• Replace the form with a thank-you message that says who will reply and when, so nobody sends it twice wondering if it worked.

Verify every lead event before you trust the numbers

Tracking nobody has tested is a guess. Say a carpet cleaning business in Sippy Downs sees more form leads in GA4 than enquiries in its inbox. One test lead shows why: the event fires on the button press, so every attempt that failed validation counted as a lead. Moving the trigger to the success message fixes the count from that day on.

Know what each event proves, too. Google Ads Help says its click-to-call conversion for mobile sites counts taps on your number, not actual calls. A tap is intent. Matching taps to answered calls is a separate job, covered in call tracking and conversion reconciliation.

Each way a customer can reach you should fire one event, once, at the moment it succeeds. Run this routine after any change to the site, the form or your tags:

1. List every contact action on the site: phone taps, form sends, email taps, booking widgets and map direction clicks.
2. Open the site through Google Tag Assistant. GA4 Help says this switches on debug mode, so your test events appear in GA4's DebugView as you go. If you use Google Tag Manager, its preview mode shows which tags fired and what triggered them.
3. Do each action for real, on a phone. Send the form with TEST in the name so the team ignores it, and check the email or CRM record arrives.
4. Confirm each event fires once, and only on success. Send the form with a required field empty: no lead event should fire. Refresh the thank-you page: no second event should fire.
5. Mark the genuine lead events as key events in GA4. GA4's recommended event for a submitted form is generate_lead. GA4 Help says a new key event can take up to 24 hours to reach standard reports and doesn't change historic data.
6. If you run Google Ads, check each conversion action's status too. Google Ads Help recommends Tag Assistant for troubleshooting website actions marked Unverified, Tag inactive or Needs attention; a verified action shows No recent conversions, then Recording once conversions arrive. For a new phone-click action, Google says Unverified or Inactive is normal until the first real conversion or a successful test. Google Ads also describes counting One conversion per ad click as the fit for leads.

Find out why visitors leave, then fix one thing at a time

Correct counts tell you how many people convert, but not why the rest left. Session recordings and heatmaps fill that gap. Microsoft Clarity, for example, flags dead clicks, where someone clicks an element and gets no feedback, and rage clicks, where someone clicks the same spot several times in quick succession. A dead click on a photo of your van, or a rage click on a phone number saved as an image, points straight at a fix. Clarity's masking documentation says input boxes and drop-downs are masked in every mode, and its default Balanced mode also masks numbers and email addresses; check that setting, and mention recordings in your privacy policy, before you switch it on.

Then ask people. One question on the callback, 'what nearly stopped you getting in touch?', turns up objections no heatmap shows: no idea of the call-out fee, an unclear service area, a doubt about weekend work. Note the answers beside your baseline.

A small local site rarely has enough traffic for a split test, so work in sequence: change one thing, leave it for a fixed period, compare against the baseline and log the date. The local visibility experiment guide covers test design in more depth.

Here's the method end to end. Say a pool maintenance business in Pelican Waters gets most of its visits to its pool cleaning page. All figures in this example are illustrative, not benchmarks. The owner records four ordinary weeks: 400 visits to the page, 6 form leads and 10 answered calls, so 16 leads, or 4 per 100 visits.

The test lead finds the first problem. The form event fires on the button press, and 2 of the 6 form leads were failed attempts that never reached the inbox. The real baseline is 14 leads from 400 visits, 3.5 per 100. The owner fixes the trigger and logs it as a tracking correction, because it changes the report and not the business.

Recordings show the second. On phones the quote button sits below a large photo carousel, and dead clicks gather on the carousel arrows. The owner swaps the carousel for one photo of a real job, moves Call and Request a quote into the first screen, logs the date and changes nothing else for four weeks.

The next four weeks bring 380 visits and 19 leads, 5 per 100. That's encouraging, not proof: a few hundred visits move around with the weather and school holidays, so the owner compares the same weeks last year where data exists, keeps the change, and moves on to the form fields. One change, one log entry, one comparison each round.

Conversion work and local SEO depend on each other: extra visibility is wasted on a page that can't turn a visit into a call. If you send paid traffic to specific pages, the Google Ads landing page checklist adds the ad-specific checks.

To choose what to fix first, work in this order, since each step depends on the one before. Start with tracking you can trust, because without it you can't tell whether any other fix worked. Then fix the first screen and the pages your outside links land on, since every visitor starts there. Next, cut friction in the form and around it, including overlays, widgets and the thank-you message. Then add proof and answers: licence, ABN, real photos, reviews shown honestly, how quoting works and when you'll reply. Leave speed and layout shift until last, unless a page is plainly slow, in which case move it up the list.

To keep the gains, copy the template below into a spreadsheet. Re-run the Events rows after every change to the site, a plugin, the form or your tags, and the full list at least once a quarter: