Serving Houston since 1998, invisible to every AI assistant

AI SEO playbook

Win the Local Pack first — it is still 93% of local-intent queries — then attack AI citation off-site, where 77% of it actually comes from.

• Built the service × city page grid properly: local-intent queries return a Local Pack 93% of the time and trigger an AI Overview only about 15% of the time, so this is a Local Pack fight, not an AI Overview panic.
• Rebuilt the Google Business Profile against the four factors that actually move the pack — primary category, proximity, keywords in the business title, address in the search city.
• Ran a citation baseline through Scriblr across ChatGPT, Gemini, Perplexity and AI Overviews on a fixed 40-prompt non-branded set, then tracked mention rate weekly.
• Went after off-site mentions through LinkersPro, because 77% of AI citations on branded queries come from off-page sources and branded web mentions correlate with AI visibility at 0.664 against 0.218 for backlinks.
• Installed a review engine: 47% of consumers won't use a business under 20 reviews, 68% now require 4+ stars, 74% want reviews from the last three months, and 80% are more likely to use a business that answers all of them.
• Skipped the schema-and-llms.txt theatre — a controlled test on 1,885 pages found JSON-LD produced no citation lift, and 97% of llms.txt files are never requested.

PPC playbook

Stop reporting one blended cost per lead. Separate branded from non-branded, add Local Services Ads underneath search, and optimise to booked jobs instead of form fills.

• Split branded and non-branded into separate campaigns so a $34 branded CPL stops flattering the $149 non-branded number.
• Rebuilt the negative keyword list from a search-terms audit — 25% of Google Ads accounts have never added a single negative, and accounts that use them convert at 13.0% against 4.6%.
• Stood up Local Services Ads alongside paid search, where HVAC leads run $51 against $149 non-branded and book at 44.0% on a $2,110 average ticket.
• Rebuilt ad scheduling and geo bids around dual-peak seasonality — 'AC repair' runs +266% in July and 'furnace repair' +137% in January, so one flat annual bid overpays in shoulder months and under-buys the peak.
• Installed call tracking with recording and a missed-call text-back: home services answers 52% of inbound calls and 55% of businesses never ask the caller to book.
• Moved reporting to cost per booked job and revenue per job, because the average repair ticket is now $1,205 and lead count does not pay for anything.

AI SEO timeline

1. Days 0–30

   Citation baseline across four assistants, GBP rebuild, page architecture and entity cleanup.

2. Days 31–90

   Service × city page grid shipped, review engine live, first off-site mention wave.

3. Days 91–180

   Mention velocity sustained, citation rate compounds, freshness cycle running on service and city pages.

PPC timeline

1. Days 0–14

   Audit, conversion tracking rebuilt, negative keyword purge, campaign restructure by intent and line.

2. Days 15–45

   Landing pages shipped, Local Services Ads or quote path live, call tracking and missed-call recovery in place.

3. Days 46–90

   Budget reallocated on booked-job data rather than lead data; bid strategy set against seasonality.

Targets this programme was set and measured against

These are not reported results. Northwind Air Conditioning's own reporting has not yet been reconciled against these rows — see how we define a target versus a result.

Where these targets come from