From No Website to Google’s Map Box: Entire Excavations, Two Months In

Part 2: before and after, same scans four weeks apart

Every row compares a scan or figure from Part 1 with the same measure on 26 October 2026. Entire Excavations went from unnamed to named in Google’s three AI products, and nothing moved in ChatGPT or the broad Google Maps search.

Part 2: the repeat check, two days later

On Wednesday 28 October we ran the five AI scans again, once each, with the same settings. Google’s AI Overview kept naming Entire Excavations for the Wandin North question at all 25 points, but lower in the list, and the other answers named it less often.

Part 2: what changed in the work, and what did not

Three of the five jobs from Part 1 are done or partly done. The two we expected to matter most, directory listings and reviews, have not started to land.

Seven more services were also added to the Google profile on 13 October, pending Google’s review at the time. Google changes its results all the time, so none of these can be credited with a particular movement.

Part 2: what the AI answers cited

Google’s AI Overview now cites Entire Excavations’ Google listing, and Gemini and Google AI Mode also cite its website, including the new Yarra Ranges page. ChatGPT’s answers still cite directory listings and none of its pages.

Part 2: what comes next

We will keep repeating the scans on more than one day, because a single scan is a snapshot. The two jobs not yet done stay at the top of the list: matching directory profiles on the free directories ChatGPT reads most (Yellow Pages and Local Business Guide lead the 26 October list), and a steady trickle of reviews. The services page and the Coldstream job page still need to get into Google’s index.

Why we measure it this way

I was a tradie before I built websites, and I don’t trust case studies where every line goes up. So this one shows what has moved and what hasn’t.

One moment shaped how we measure it. Our search checks said Entire Excavations was first in Google’s map box for a Wandin North search. I got a mate to try it on his phone in the city, in a private window. It didn’t show at all. We still can’t say exactly why: his wording, where his phone was, and Google showing different people different results can each change it. One search from one phone tells you almost nothing, so everything here comes from Local Falcon grids, repeat checks on different days, and searches run from several places.

Tom Syddall, founder, 5 Star Tradie. Chris, who runs Entire Excavations, agreed to be named.

Part 1: before and after, the first two months

In late July Entire Excavations had no website of its own, and its Google profile’s website link pointed at Instagram. Here is how things stand now.

The reviews haven’t moved, and they are the biggest thing holding him back. Both say the same thing in different words: one customer wrote that Chris and his team “went above and beyond” on a very large pond build, and the other that they “went above and beyond to ensure that the job was completed to the highest standard.” Five clicks is early, and many of the times shown were for broad Melbourne searches where the site sits well down the page. For a local trade, the map box is where many customers ring from, so it is the result we watch most closely.

How to kickstart a new trade business on Google

This is the order we would use for any new trade business. With Entire Excavations we have done steps 2 to 4 so far. Directory listings and reviews are next (see What we are doing next). Google says local results are “mainly based on relevance, distance, and popularity”, so every step either tells Google more clearly what you do and where, or builds the signals that you’re a real, well regarded business.

1. Get the Google Business Profile complete and verified. Google’s own advice is to verify the business, keep the address, hours and category accurate and complete, and add photos. Pick the category that matches the work customers search for, not a vague umbrella.
2. Point the profile at a real website. A social page gives Google very little to read about your services and areas. Entire Excavations’ profile pointed at Instagram until July.
3. Give every service and every area its own page. Real job photos from real suburbs, a clear phone number and tap to call on every screen. The site went live one week after the go-ahead, with 9 pages at launch and 12 now.
4. Make sure Google has indexed every page. Check the Pages report in Google Search Console, strengthen any page listed as “discovered, currently not indexed”, then request indexing. Entire Excavations’ indexed pages went from 2 of 10 on 28 August to 11 of 12 by 25 September.
5. Get listed where AI answers look. Our scans showed AI answers leaning on directory listings such as hipages (a paid service), Yellow Pages, Local Business Guide, localsearch and True Local. Use exactly the same business name, phone number and service area on every one.
6. Ask for reviews steadily. Google says “more reviews and positive ratings can help your business’s local ranking”, and recommends replying to reviews. Ask each customer as the job finishes, never in a batch, and never offer anything in return.
7. Measure it properly. Check from several places, on more than one day, and include a weekday. The four lessons below show why.

Lesson 1: Scan more than once: Google’s AI answer changed completely in two days

We asked the same questions from the same 25 points on the map on three days. For the Wandin North question, Google’s AI Overview named a completely different set of businesses on the third day than on the first.

Here is what one of those scans looks like in Local Falcon. Each pin is one of the 25 places the question was asked from.

Lesson 2: Read the sources: they are the to do list

Local Falcon records the links each AI answer relied on. For a new business this is the most useful part of the scan, because it shows exactly where the AI looks.

Lesson 3: Check on a weekday: the weekend may flatter you

On Saturday afternoon Entire Excavations was first in Google’s map box for “excavation contractor wandin north” from all four places we searched. On Monday morning it was still in the top three, but no longer first.

Lesson 4: Win the suburb searches first, the broad ones may follow

For two short searches that name Wandin North, Entire Excavations is now in Google’s map box (on Saturday it was not in the map box for “excavators wandin north” or “earthmoving wandin north”). For the broad search “excavation contractor” across the same area, and in Google Maps itself, it isn’t.

What we are doing next

The scans point to five jobs, in the order we expect them to matter:

• The directories the AI actually reads. Complete, matching profiles on the free directories that came up again and again as sources, starting with Local Business Guide and Yellow Pages, all with the same name, number and service area.
• Reviews. Two reviews is fewer than most businesses shown on the map. We are setting up a simple process that asks past customers one at a time, run for Chris so it doesn’t depend on him finding the time.
• An area page the AI can quote. A competitor’s Yarra Ranges excavation page was a source in both 25 September Google scans. Entire Excavations will get a Yarra Ranges page that ties together the towns it works in, with real job photos from those towns.
• Every page indexed. The services page gets strengthened and resubmitted.
• The map point. Checking that the listing’s map point matches where the business is based.

In Part 2, about four weeks after these changes, we will re-run exactly the same Local Falcon scans on the same grid and show what moved and what didn’t.

Methodology

Local Falcon scans. Run through the Local Falcon API, centred on Entire Excavations’ Google listing (−37.7636, 145.5214), 5 by 5 grid, 8 km from the centre to the outer points. On 25 September: “who are the best excavation contractors in wandin north” on Google AI Overviews (1:06pm), ChatGPT (1:07pm), Gemini (2:59pm), Google AI Mode (3:00pm), “excavation contractor yarra valley” on Google AI Overviews, and “excavation contractor” on Google Maps (1:05pm). The two Google AI Overview scans and the ChatGPT scan were repeated on 26 September (about 11:30am) and 27 September (about 1pm). Counts of points, names and sources come from Local Falcon’s report data.

Map box and suburb searches. Through the DataForSEO search results service: the “excavation contractor” map box on the same centre and a 5 by 5 grid 4 km apart (25, 26 and 27 September); “who are the best excavation contractors in [suburb] Victoria” for 10 suburbs on desktop and mobile (23, 25, 26 and 27 September); and “excavation contractor wandin north” and “excavation wandin north” from Wandin North, Seville, Lilydale and Melbourne’s CBD on desktop and mobile (Saturday 26 September, late afternoon, and Monday 28 September, 10:30am), with the Google Maps results for both from Wandin North and the CBD on 28 September. Failed searches are left out of every count.

Website. Google Search Console for entireexcavations.com.au, 26 August to 23 September 2026 (re-pulled for 28 August to 25 September), and URL inspection for index status. Earlier figures (28 August, and 12 August to 9 September) come from our case study records saved in September 2026. The 16 September map box figure comes from our September snapshot. Reviews read from the public Google profile on 28 September 2026.

Part 2. On Monday 26 October 2026, 11:10 to 11:12am, we re-ran the six 25 September scans through the Local Falcon API with identical settings: the same listing, centre, 5 by 5 grid, 8 km radius, questions and platforms. Each scan was run once. On Wednesday 28 October at 3:43pm the five AI scans were run once more with the same settings; the Google Maps scan was not repeated. Before comparing, we listed the account’s reports from 24 to 29 September: all ten Part 1 scans are still there, including the Google Maps scan (see Limitations). Sources were de-duplicated with the same method as Part 1, which reproduces Part 1’s 339 unique pages from the 25 September file. Search Console figures for 26 September to 23 October (the latest day with data on 26 October) and URL inspection of all 15 pages in the sitemap were pulled on 26 October through the Search Console API. The review count is Local Falcon’s record of the Google listing at scan time, the same 2 as the public profile showed in Part 1.

Questions about the data? Email thomas@5startradie.com.au.

Limitations

• One business, one area, four days of checks. It shows what happened here, not what happens everywhere.
• AI answers change with wording and timing. Every figure is a snapshot.
• On 26 September Local Falcon captured Google’s AI answer text but no business names, so names that day can’t be compared.
• We ran the Wandin North Google scan a second time at 3pm on 25 September, while Local Falcon showed a high volume notice. It returned answers only at the two points that had none at 1:06pm. We can’t explain that, so we left it out.
• Gemini and Google AI Mode ran once, at about 3pm on 25 September, in the same busy window, but returned full answers at all 25 points, so we kept them in the 25 September count. They were not repeated.
• Map box order can change with the time of day and with which businesses are open, so the Saturday and Monday results are not a like for like comparison.
• Search Console data lags by about two days, and part of the impressions come from searches outside the service area.
• Part 2 is two scans per AI question, on 26 and 28 October, four weeks after Part 1. AI answers change between days, so every figure is a snapshot, not a position held.
• Local Falcon told users on 6 October that changes on Google’s side disrupted Google Maps scan data from 24 to 30 September, and that it reviewed every scan from that window and removed the inaccurate ones. Our 25 September Google Maps scan falls in that window and was not removed. The AI scans are a different product and were not part of that notice.
• The 26 October Google Maps scan and the 25 September one both found Entire Excavations outside the top 20 at every point, so the disruption doesn’t change that comparison.
• Search Console figures are revised after the fact, and the Part 1 and Part 2 windows differ by a few days in places. Small changes should not be read as trends.
• The 25 September scans ran between 1:05 and 3:00pm on a Friday; the 26 October scans ran between 11:10 and 11:12am on a Monday. Time of day may affect AI answers, so this is not a perfectly like for like comparison.
• Part 2 is not a final result. The directory listings and the review drive have not landed yet.
• We don’t name the other businesses. Most are Chris’s local competitors, and this is about what the AI does, not about them.