Ihor Odariuk, SEO and GEO consultant.
I help home service businesses get found on Google and named inside AI answers, and I set up the tracking that shows which of it actually books jobs. The tracking comes first: half the jobs in a typical service CRM say nothing about where they came from, so without it nobody can tell whether an ad, a page or a ChatGPT answer paid for itself.
Three areas, run as one system.
Search
Technical and local SEO, URL and internal-link architecture across hundreds of pages, cannibalization found in live Search Console data, schema and indexation.
AI visibility (GEO / AEO)
Pages rebuilt so ChatGPT, Gemini, AI Overviews, AI Mode, Perplexity and Copilot can quote an answer straight from them, then measured on a schedule with the caveats attached.
Revenue tooling
First-touch tracking, server-side lead logging, CRM joins by phone key, call transcription and classification, ad-platform and CRM API pipelines, dashboards.
Home services (appliance repair, HVAC, plumbing, cleaning) is where this was stress-tested: jobs typed in by hand, sources missing, money agreed over the phone. That makes it the hardest case for attribution, not a limit on who I work with.
What the main engagement produced.
One client, twelve months: an appliance repair company. Every figure links to the full write-up, including what did not work.
Real leads from the website, counted at the form with spam removed, on a site the CRM had credited with three jobs a month.
How it was counted →Traced back to what produced them after tracking was joined to CRM jobs and invoice amounts, in a CRM where 45% of jobs had no source at all.
Method and numbers →AI answers citing the client site across six AI platforms in 15 weeks, from July 4 to October 21, 2026, read from Ahrefs Brand Radar, which moved to a new index in between. The first six weeks are published snapshot by snapshot.
How it was built →Pages showing up in Google after the site was rebuilt from 64 pages so city and service pages stopped competing. Monthly impressions went from 6,145 in October 2025 to 432,266 in October 2026, and monthly clicks from 441 to 2,068.
Read the case study →Of paid-lead spend identified as leads that were never customers, after 97,800 call records were transcribed, classified and matched to CRM jobs.
Paid-lead layer →Outside home services, the same method was tested on three more businesses. Figures are from the Search Console API, pulled on 5 November 2026.
Event company, Spain. A live music and event company with no ad budget went from 10 search impressions in late June 2026 to 830 in October, and the owner confirmed two jobs booked through the site, one from Google search and one from the new Spanish pages. The event company case →
Cleaning company, Texas. A site that had brought 11 calls and one real form request in six months went from 1,705 to 33,295 monthly impressions between July and October 2026, with clicks from 7 to 60 a month, its first paid job through the website and its first lead from ChatGPT. The cleaning company case →
Music education site. With no link building, it went from 50 clicks in its first 28 days to 1,105 clicks in October 2026 alone. The cross-niche write-up →
The rules I hold myself to.
Measured before it is claimed
Every number comes out of an API, not a screenshot: Search Console, the CRM, call logs and ad platforms. Write-ups include the findings that went against me, such as a dispute automation of mine that stopped paying off.
No guarantees on rankings or citations
Nobody controls what Google or an AI assistant shows. What I can promise is the measurement: every job carrying its source, so the effect of the work is visible in numbers.
Flat fees, published prices
Prices are on the pricing page, not behind a call. No revenue share and no per-lead commission: I build the numbers the work is judged by, and nobody should be paid by their own scoreboard.
Remote, written-first
Based in Europe, working with US clients across US hours. Findings arrive as documents you keep; calls happen when there is something to decide.
Business first, then search.
Ten years in business and marketing before search: brand management, B2B and B2C sales, e-commerce SEO for a major instrument distributor, paid social budgets above $10K a month. That is why the work reasons backwards from the invoice rather than forwards from a keyword list.
Published along the way: an open dataset of AI crawler requests, the full program in plain English, and working notes on the blog.
Send me the site.
Within 48 hours you get the 3 to 5 problems costing you the most, in writing. No call needed. Hiring for an in-house role instead? That is on the hire me page.