Anonymised. Composite drawn from three regional trade-services programmes.
The situation
A family-run electrical contracting business in Geelong (two vans, four staff, established 2011) had solid Google reviews, a functioning website, and a steady flow of word-of-mouth work. When the owner started noticing that new enquiries were slowing, he ran a quick test: he asked ChatGPT and Perplexity who to call for a licensed electrician in his service area. Neither named his business. One named a Melbourne-based company with a Geelong landing page. The other named a business that had closed two years earlier.
Starting Visibility Index: 11 out of 100.
What we found in the audit
The business had three separate ABN-registered trading names appearing across different directories; a legacy of a rebrand in 2019 that was never fully cleaned up. Yelp had the old name. True Local had a different phone number. The Google Business Profile had the current name but a PO Box address while the website had the street address. From a retrieval standpoint, the models were dealing with at least three partially-overlapping entities that they could not confidently merge into a single recommendation.
Schema: none beyond the basic WordPress defaults. No LocalBusiness markup, no areaServed, no licenceNumber field (which, for licensed trades in Victoria, is a trust signal the models weight meaningfully). Third-party corroboration outside Google reviews: two citations, both from 2018.
The work; eight weeks
Weeks 1 to 2: Entity consolidation. Reconciled all directory listings to a single consistent NAP. Claimed and standardised 17 directory profiles. Added VBA licence number to the GBP, website footer, and schema. Consolidated the two old trading-name profiles with a merge request. This single step moved the Visibility Index from 11 to 29.
Weeks 2 to 4: Schema and content structure. Added LocalBusiness and Electrician schema with areaServed polygon covering all six postcodes serviced, explicit priceRange, openingHours, and paymentAccepted. Rewrote homepage opening paragraph to lead with five disambiguation signals in the first 60 words: suburb, founding year, licence number, scope of work, service area.
Weeks 3 to 6: Authority content. Published four service pages (switchboard upgrades, EV charger installation, safety switch testing, solar integration) each with 1,400 to 1,800 words, FAQPage schema, and a realistic price-range table. Price tables are consistently the most-retrieved content type for trade services; they answer the question buyers actually ask first.
Weeks 5 to 8: Citation engineering. Secured placements in the Geelong Advertiser small-business section, the local HIA chapter newsletter, and two local suburb Facebook groups (managed mentions, not ads). Added the business to the Master Electricians Australia member directory. Each placement was unlinked; the mention itself is the signal.
Results at week 10
- Visibility Index: 11 to 68
- ChatGPT: named in 8 of 12 priority queries (from 0)
- Perplexity: named in 10 of 12 priority queries (from 1)
- Phone enquiries from new customers: up 34% in weeks 8 to 10 versus same period prior year
- Of 34 new enquiries in weeks 9 to 10, 11 cited asking an AI when asked how they found the business
The learning
For established local trade businesses, the most impactful work is almost never new content. It is entity cleanup: the boring, invisible reconciliation of NAP data, schema, and directory consistency that makes the model say “I know exactly who this is and where they work.” The content work accelerates compounding. But without the entity foundation, additional content just adds more signals to an already-confused entity picture.