AI Pilot Plan for a Small Local Service Business: A 30/​60/​90-Day Guide

Pick the leak that costs you the most and that you can measure. For most local service businesses the candidates are the phone, lead response, reviews and your Google Business Profile, and the paperwork you already do on a computer (quoting, invoicing, writing). The data below helps you rank them.

Phones and lead response. Jobber found 60% of home-service pros reply to leads the same day and 20% within the hour, while over 55% of customers expect a reply within the hour and 28% expect an immediate one. Missed-call figures are weaker. CallRail reported in July 2026 that 28% of calls go unanswered across 225,000+ businesses, and that businesses using its AI answering saw 44% more calls answered. A third-party compilation attributes to Invoca (2024) that 27% of inbound calls go unanswered across 60M+ calls and that the average missed call costs $1,200 in revenue. CallRail and Invoca both sell call tools, and the original reports were not retrieved. Other vendor blogs cite much higher rates (45%, 62–70%, "$126K a year") with no methodology. No independent, non-vendor study of how much AI call answering improves booking rates was found. Count your own missed calls for two weeks before deciding.

Reviews and visibility. BrightLocal's 2026 survey (1,002 US adults, published February 11 2026) found 97% read reviews for local businesses, and 41% always do, up from 29% in 2025. 68% require at least 4 stars (up from 55%), and 31% require 4.5 or higher (up from 17%). 47% won't use a business with fewer than 20 reviews, and 74% only consider reviews from the last three months. 80% are likely to use businesses that respond to all reviews, and 19% expect a same-day response. 78% were asked to write a review in the past year, and 83% of those complied. If your review count or recency is thin, review requests and replies are a strong pilot candidate.

AI search. BrightLocal's survey also found 45% of consumers use ChatGPT or other AI tools for local recommendations, up from 6% in 2025. BrightLocal's stats page cites SOCi's 2026 Local Visibility Index: only 68% of business contact details on ChatGPT and Perplexity match the Google Business Profile, and visibility in ChatGPT local recommendations is "30x harder" than in Google local results. The practical takeaway is consistency: the same name, phone number and hours everywhere.

Paperwork. Quoting (54%), invoicing (52%) and business writing (51%) are the most common AI uses among Jobber's home-service owners. This is the lowest-risk place to start because nothing customer-facing goes out without you reading it. A general AI assistant is enough for it, and costs are low (see the cost section).

If you buy leads through Google Local Services Ads, start with the phone. The charging change that began October 1 2026 makes unanswered calls an immediate cost.

Shadow mode means the AI works but a person checks before anything is final. For paperwork, the AI drafts quotes, invoices or emails and you review them. For reviews, the AI drafts replies and a person posts them. For phones, the AI answers or takes messages for a limited window (for example, after hours) while you listen to recordings or read transcripts, and every booking it makes is checked by a person.

Build a test set of 20–50 real examples, as Wavect recommends: actual calls, inquiries or reviews from your last few months. Score what the AI did against what you would have done. Note every miss: wrong price quoted, wrong service area, a job it should have escalated, a tone that doesn't sound like you.

Fix and re-test. Most failures at this stage come from missing information, such as the services you don't offer, the areas you don't cover, your emergency rules and your prices, or from unclear instructions on when to hand off to a person. Update the tool's instructions and run the same test set again. Track the metric you set in the first 30 days alongside the baseline.

If you use Google's Gemini connection for Business Profile, note what Google announced on June 10 2026: you can connect a Business Profile to Gemini to analyze performance, draft review replies in your brand voice, and update hours or seasonal posts. Google also announced "Business notebooks." It rolled out globally in June 2026, excluding the EEA and UK, and the post does not specify pricing. Review replies are checked against Google's content policy and appear as the business, not a named person. Replies usually take up to 10 minutes to appear but can sometimes take up to 30 days.

Keep reviewing for tone and accuracy. Whatever the AI drafts, you are responsible for what it says in your business's name.

Prices below are from vendor pages when checked on October 1 2026 unless marked. Prices change, so confirm on the live page before you buy.

An IT vendor, tekRESCUE (published August 28 2026), says tools run from free to $20–$30 per user per month, with total monthly spend for small teams between $50 and $300. That claim is unsourced. A UK consultancy (NCS London) gives chatbot solutions at £300–£500 a month and AI training from £200, in pounds and for the UK market.

For general AI assistants, Claude Team's standard seat is $20 per seat per month billed annually or $25 monthly, with a premium seat at $100 annual or $125 monthly, a 2-seat minimum and a 150-seat maximum. Claude Pro for an individual is $20 a month or $17 a month billed annually ($200 upfront). ChatGPT Business is listed at a standard seat of $25 per user per month monthly or $20 annual, and a premium seat of $125 monthly or $100 annual, with at least 2 paid seats; those ChatGPT figures come from search results because OpenAI's pages could not be fetched directly. Google Workspace's page shows Starter $7.00, Standard $14.00 and Plus $22.00 per user per month at regular price, with 30–50% introductory discounts for three months and Gemini included; the page did not say whether those are annual-commitment prices, and a secondary source lists $8.40, $16.80 and $26.40 for flexible billing.

For phone handling, you have three routes: an AI add-on inside field-service software, a live human answering service, or a standalone AI receptionist. The table in the decision section compares them. On Jobber's pricing page, plans are Core $29 annual / $39 on a one-year monthly commitment / $49 with no commitment; Connect $99 / $119 / $139; Grow $149 / $169 / $199; and Plus $399 / $439 / $499, with extra users at $29 a month. The Jobber AI Receptionist add-on was listed at $29 a month, described as 24/7 customer service over call or text that books jobs into the calendar. Secondary sources say the add-on is $99 a month and included in a $599 Plus tier; those do not match the vendor's page, so ask Jobber for the current price. Housecall Pro lists Basic at $59 annual or $79 monthly for one user, Essentials at $149 or $189 for five users, and Max at $299 or $329 for eight users with extra users at $35. It mentions CSR AI but publishes no price.

For human answering, Ruby charges $250 a month for 50 minutes, $395 for 100, $720 for 200 and $1,725 for 500. Smith.ai charges $300 a month for 30 calls (overage $11.50), $810 for 90 calls (overage $10.50) and $2,100 for 300 calls (overage $8.50), with booking at $1.50 per call and call recording at $0.25 per call. Generic AI receptionists are listed in vendor-blog roundups at $25–$899 a month overall, $25–$160 for most small businesses and $50–$300 for home services.

A note on budget: a pilot that only uses a general AI assistant for quoting, invoicing and writing can run on a couple of seats at $20–$25 each. Adding an AI phone line typically adds tens to a few hundred dollars a month by the figures above, while human answering services start at $250–$300 a month.

Google told Local Services Ads advertisers that from October 1 2026, missed calls during business hours count as charged leads if the caller stays on the line for more than 20 seconds. If a phone menu requires a key press, the 20-second timer starts at the key press. Later calls between the same parties can be charged if the first call didn't qualify. The notice reportedly describes no dispute process.

The source is an emailed Google notice described by PPC Land and Search Engine Watch; no Google help-center page for it was found. If you advertise through LSA, check your account and email, confirm your business hours in the account are right, and decide who or what answers during those hours. An AI receptionist, a human answering service or a staffed phone could all serve here, and the pilot's phone metric (share of calls answered inside business hours) maps directly to this charge.

Several things you may want to know have no solid answer yet. There is no independent, non-vendor study of how much AI call answering improves booking rates; the 44% figure is CallRail's own. The missed-call rates and the $1,200 per call figure trace to vendors that sell call tools. Census does not publish adoption for construction, home services or other local-service sectors. Several price points are single-source or conflicting (ChatGPT Business, Jobber's AI Receptionist, Google Workspace billing). The ChatGPT Business prices came from search results because OpenAI's pages could not be fetched directly. The recording-consent state lists come from secondary sources.

This is why the pilot design matters. With no reliable outside benchmark, your own baseline and your own 90-day comparison are the evidence.