How the 2026 South Coast business website reports are made

Each year I run the same set of checks across business websites on the NSW South Coast and publish what I find, one report per area. This page is the method: where the businesses came from, how each site was loaded, what each check actually tests, and what was left out. If you're deciding whether to quote a number from one of the reports, this is the page to read first.

Where the businesses came from

Every business in the sample has a Google Business Profile listing with a website address on it. I pulled the listings through Google's Places API using a set list of search terms across a set list of industries: trades and home services, building and construction, hospitality and food, accommodation, retail, professional services, health and medical, beauty and wellness, automotive and marine, tourism and activities, fitness and sport, real estate and property, and weddings, events and photography.

For Batemans Bay, Moruya and Narooma, the list was collected in July 2026. For Bega and Merimbula it was collected in October 2026 in the same way. All four areas were checked in the same week in October 2026.

Then I took things out:

• Franchises and chains, including locally owned branches of national brands.
• Listings marked closed.
• Businesses whose only web presence is a Facebook page, an Instagram page or a directory listing. They count as having no website of their own and weren't checked.
• Listings that aren't businesses: parks, beaches, public facilities and bare street addresses that the search terms turned up.
• Listings whose website address belongs to someone else, such as a booking aggregator or a supplier.
• Two listings that share one website are checked once.

Each business is assigned to an area by the locality in its Google address. Batemans Bay takes in the town plus Batehaven, Catalina, Surfside, Long Beach, Surf Beach, Malua Bay, Mogo, Nelligen and the villages between. Moruya takes in Moruya Heads, Broulee, Mossy Point, Tomakin and Tuross Head. Narooma takes in Dalmeny, Kianga, Bodalla, Central Tilba, Tilba Tilba and Mystery Bay. Bega and Merimbula takes in Tathra, Kalaru, Candelo, Wolumla, Tura Beach, Mirador, Pambula and Pambula Beach. Businesses in other localities aren't in the sample.

The industry each business sits in comes from the search terms that found it, checked against Google's own category for the listing. A handful will still be in the wrong group.

This is a large sample of each area, not every business in it. The reports say "of the sites I checked", and they mean it.

How each website was loaded

Every site was loaded in a real browser (Chromium, run without a screen) set to a phone viewport 390 pixels wide and 844 tall, with a phone user agent. The page was given time to finish loading, then examined. Sites were loaded at their https address; a site that only works over http was loaded as listed.

The audit ran from a cloud server, not from a phone in the area. That matters for one reason: some websites refuse or challenge automated visits from server addresses. Those sites weren't scored. They're counted as "could not be checked" and left out of every percentage.

A site was also left out if it refused to load after three attempts, loaded without its stylesheets, showed almost no text, or turned out to be a parked, suspended or holding page.

Where a site's homepage linked to a contact, quote or booking page, that page was loaded too, and used for the checks on enquiry routes and phone numbers.

How speed was measured

Speed comes from Google's PageSpeed Insights API, which runs Google's Lighthouse test on the page. Each site was tested three times on separate passes, minutes apart, and the median was kept. The figure reported is Largest Contentful Paint: the time until the main content of the page is visible.

The test simulates a mid-range phone on a slow mobile connection. It is deliberately harsh and it isn't what most visitors on home wifi experience. It is the same test for every site, so sites and areas can be compared with each other, and it's a fair stand-in for a phone on patchy reception. Speed is reported on its own in every report and isn't part of any score.

The ten checks

Each site gets a pass or a fail on each of these. The reports mostly quote individual checks rather than a total.

1. Homepage title names a town or the region. The page title contains any locality in the four areas, or a regional name such as South Coast, Eurobodalla, Sapphire Coast or Bega Valley.
2. Content is readable without JavaScript. The plain HTML the server sends holds at least half the words a browser ends up showing, or at least 200 words. This is what AI assistants and some crawlers see.
3. Local business structured data is present. The page carries schema.org LocalBusiness markup, or one of its subtypes such as Plumber, Restaurant or LodgingBusiness.
4. Main content shows within 2.5 seconds. The median Largest Contentful Paint from the three PageSpeed runs is 2.5 seconds or under, which is Google's own "good" threshold. Reported separately, not scored.
5. The page and everything it loads use HTTPS. Lighthouse's "Uses HTTPS" test passes in every run.
6. Laid out for a phone screen. The page declares a mobile viewport (device width, or a set width no wider than the screen) and is no more than 8 pixels wider than the screen.
7. Phone number can be tapped to call. The homepage has at least one tap-to-call link.
8. Phone number is on the first screen. Before any scrolling, a tap-to-call link or a written phone number is visible and not covered by a popup. A number inside an image doesn't count, because a visitor can't tap it either.
9. There is a way to enquire or book online. A real enquiry form (a message box, or three or more fields) on the homepage or a linked contact, quote or booking page, or a link to a known booking or ordering system. Search boxes, newsletter sign-ups, login forms and shop forms don't count. An email link on its own doesn't count, and is reported separately.
10. Phone number matches the Google listing. The number on the Google listing appears on the homepage or contact page, with or without the area code. Not applied where the listing shows no number.

Alongside the checks, each site is tagged with its industry, the platform it was built on, and the Google rating and review count of the business.

What the 2026 run produced

"Address led nowhere" means the website link on the Google listing goes to a page that no longer exists, a domain that has lapsed, or a parked, suspended or holding page. These are reported as their own finding. "Could not be checked" is the sites that refused an automated visit or failed to load properly.

Rules I hold to

• No business is named with its results. Not in the reports, not in the charts, not when asked.
• No group under 8 sites is reported. Smaller industries are left off the industry tables rather than shown, so no figure can point at one or two identifiable businesses.
• Numbers are stated as what was measured. "41% of sites show a phone number on the first screen" is a measurement. "Batemans Bay websites are bad" is not, and the reports avoid it.
• Towns aren't ranked. On most measures the four areas are within a few points of each other, and most of those gaps are small enough to be chance. Each report says where its area does stand apart and where it doesn't.

Checking the checks

The checks are automated, which means they can be wrong in consistent ways. Before publishing, a sample of sites was given a visual check against what the script recorded, including a random set where the phone-number result matched what was on screen every time, and a larger set chosen because they were the most likely to be wrong. That second set turned up a handful of dead or broken pages that had been scored as live sites, and they were moved out of the results.

Known limits, so you don't have to find them yourself: the first-screen phone check can't see a number that's part of an image; the enquiry check matches booking systems by name and will miss ones it doesn't know; sites behind bot protection are under-represented because they couldn't be checked; and the industry labels for Bega and Merimbula were assigned by search term and corrected where Google's category was clear, so a few will be wrong.

Next year

The same checks, the same rules and the same areas run again in 2027, so the reports can be compared year to year. If a rule changes, this page will say so and the reports will say which rule.

If you'd like to quote a figure, link to the report it came from. If you'd like to argue with one, the contact page is the place.