People ask me this constantly: "I can't afford an agency right now. What can I do myself?"
This is the answer. It's the same checklist I work off of. One line item on this page costs money and I flag it where it comes up. The rest you can do from your phone in a parking lot between jobs.
Fair warning: it's a lot of work. That's the honest reason agencies exist. But none of it is hard, and doing half of it puts you ahead of most of the people you're bidding against.
The doors
Your customers aren't coming from one place. One Googles "sewer line replacement near me." One asks her phone out loud. One sees a video of you cutting into a slab. One reads his neighbor's post in a Facebook group. And one never types anything at all, because an assistant compared six plumbers on her behalf and handed her a name.
You don't get to pick which door they open. You only get to decide whether you're standing behind it.
Everything below is the list of doors and what it costs, in time, to be waiting behind each one.
What changed
Skip this if you want. It explains why the checklist looks different than it did three years ago.
The customer isn't always the one searching
When a water heater dumps onto a garage floor, the homeowner used to type "plumber near me," scan ten blue links and call two. More and more, she asks an assistant instead, and the assistant does the comparing.
At I/O in May 2026 Google announced agentic booking in Search for local services. For select categories, home repair among them, a user can ask Google to call businesses on her behalf. It's rolling out across the U.S. inside AI Mode in stages. Go read Google's post rather than my summary of it, but the roadmap is published, not predicted.
BrightLocal's 2026 consumer survey put the share of people using AI to find local businesses at 45%, up from about 6% a year earlier. One survey, self-reported, and a jump that big usually means the question changed too. Hold the number loosely. The direction isn't in dispute.
You don't research drivers. You open an app and it finds someone nearby, verified, available now. An assistant picking a plumber runs the same test. The only way to look verified is to be verified, in enough places, consistently. That's why this works: it's hard to fake.
The machine vets you like a suspicious customer would
Say your name is spelled two ways across four sites, your Yelp number doesn't match your truck, and your site claims the whole metro while every review comes from one suburb. Nothing penalizes you. It's simpler than that. The machine can't corroborate you, so it recommends the contractor whose story holds together. You never find out you were skipped.
You used to buy attention. Now you earn a recommendation by being legible. Keep the storefront clean and make every fact match every other fact.
One more thing
In July 2026 OpenAI licensed Yelp's review data for ChatGPT. Whatever you think of Yelp as a lead source, and I have opinions, your free Yelp profile just became infrastructure. Door 3.
The calendar
Not a door. It's what all the doors open onto, and almost nobody has it.
Here's the sequence that loses you jobs all day long. Somebody finds you, likes what she sees, and hits a wall. A contact form routed to an email you check at 9pm. A phone that rings while you're under a house. "Call for availability," which means call and hope.
A person who hits that wall is a coin flip. An agent that hits it is gone. It dials the next contractor and logs nothing. You don't get a missed call. You get a silence you can't measure.
No page, post, video or profile ends without a booking path, and every booking path is the same path. One booking page. One phone number. On the truck, on the invoice, on the profile, in the caption.
Make it honest
A booking calendar is worth something only if it tells the truth. Slammed until Thursday, it says Thursday. Fridays off, Friday isn't bookable. Don't do water heaters anymore, don't list water heaters.
Contractors treat this as a downside. "I don't want to show I'm booked out." It's the opposite. Booked until Tuesday is the most credible thing a contractor can say, and most marketing has to fake it. If you have a backlog, showing it is free proof. It also protects you: a calendar that reflects reality means nobody books a slot you can't keep, which is how you lose a review before you've done the work.
Doing it for free
You don't need to buy scheduling software to be bookable. Two moves.
A number that can text. Customers send photos of the leak, which is your diagnosis delivered free before you drive anywhere. Google Voice's free tier is designated for personal use, and business texting on it runs into the carrier registration rules. Budget $10–$30 a month for a business line that can text. I'd rather tell you that than pretend the free version works. A plumber who can't receive a photo is losing money in a way that's hard to see.
Make yourself bookable and make the calendar honest. The rest of this list gets people to the door. This is the door opening.
The big one. On Google you pay for a direct contact from someone searching for your service right now, not for a lead record resold to a list of pros. A homeowner can still call two of you. But nobody sold your name four times over. Do this door first.
Step zero
The profile
Two more free things
What gets you suspended
Keyword-stuffing your business name. "ABC Plumbing | Emergency Plumber Houston 24/7" is the number-one killer in home services. Your name must match your real-world branding. Google says padding it can get the profile suspended. Contractors do it because it used to work. Now it's a coin flip on your whole lead flow.
A virtual office or mailbox address. Rent a mailing address you don't operate out of and the location isn't eligible. Coworking requires signage, your own staff, and receiving customers during business hours.
Duplicate listings, one per city. Common in home services, against the guidelines, and it splits your standing instead of multiplying it.
Call-tracking numbers as your primary number. The guidelines prohibit numbers or URLs that redirect somewhere other than the business. It's also the classic own-goal on name, address and phone consistency. Track with destination URLs and campaign tags.
Fake or incentivized reviews. They don't just get deleted. Google can put a public banner on your profile telling customers fake reviews were removed. Worse than having none.
What's oversold
"Post every 7 days or visibility drops after 30." I've repeated a version of this myself. Google has never published it. Practitioner folklore that hardened into a rule. Post because a live profile beats a dead one and posts carry your booking link, not because there's a cliff on day 31.
"Turn on Business Profile messaging, seed your own Q&A." Both gone. Profile chat shut down in July 2024. The Q&A API was discontinued in November 2025 and the consumer feature went with it; by third-party accounts it's been replaced by an AI-generated answer surface that pulls from your profile, your reviews and your website. So your FAQ content has to live on your site now, because that's what's being read. Any 2026 guide telling you to turn on GBP messaging was written from a 2022 template.
"68% of local searches get an AI Overview." That's from a 2025 study, averaged across three query types. For local-intent queries like "plumbers in Phoenix" it was closer to 15%. The 92% figure was for informational questions. Useful distinction: informational content is where AI answers show up, which is a reason to write it. It was never a number about the searches you compete for.
Your website
The information desk. For local service work your website usually isn't where you get discovered. Discovery happens on the map, in the assistant, in the feed. The site is where you get corroborated and where you get booked. That changes what belongs on it.
The bones
The content that earns you a mention
The listicle is dead for this. "7 Ways to Summer-Proof Your Home" has been written ten thousand times by people who've never opened a plenum, and there's nothing in it a machine can only get from you. Write the shape a real customer call has.
What's oversold, and one line not to cross
Judging this content by traffic. A lot of these pages get read without anyone clicking through. Measure them by sessions and you'll conclude they failed and kill a program that's working. Measure by whether you come up when you ask an assistant your own top ten questions.
Thirty near-identical pages with the city swapped. Six symptoms times five ZIPs is the classic local-spam pattern, it's well detected, and it makes you harder to pin down. One page per symptom with genuine local specifics. A neighborhood earns its own page only if you can write 400 words true there and false one suburb over.
Inventing a customer. "Julie from Westbury asks…" attached to a person who doesn't exist is fabricated testimonial content. Writing the question is fine; every FAQ page does it. Attaching a fake person is the one thing that can undo a business whose differentiator is telling people the truth. The upside is zero. An unattributed question performs identically. Want a named story, go get a real customer's permission.
Maps & directories
The unglamorous door, and in 2026 one of the most important. When an assistant answers "who's a good plumber near me," a large share of what it cites is directories and review platforms. Not social. Claiming these is free and takes an afternoon.
Claim these, in this order
The audit that ties it together
What's oversold
Paying a service to manage your listings. Those tools are priced for multi-location chains. One location, you can do it by hand in an afternoon.
The old free data-aggregator submission forms. Several have been retired or moved behind paid plans; the one formerly called Localeze runs about $99 a year under a new name. Don't burn a Saturday chasing a form that no longer exists. The list above is the list.
BBB accreditation "for SEO." The free listing carries the same citation. Buy accreditation for the badge if you want it. Not for the algorithm.
Meta — Facebook & Instagram
Be honest about what this door is. Organic reach for a local service page has been poor for years. Your Facebook page isn't a discovery engine. It's a credibility asset that converts referral traffic: somebody hears your name at a barbecue, looks you up, and decides in eleven seconds whether you're real. The traffic comes from groups.
Setup
Automation Meta hands you
Groups, where the traffic is
What's oversold
"List your services on Facebook Marketplace." Traces to a 2018 launch that still ranks well and gets recycled into "2026 guides." Marketplace has no real services category today; providers list under product categories as a workaround and risk removal. Use it to sell physical things: surplus material, a retiring truck. Not your labor.
"Facebook reviews get you cited by AI." No good evidence on local queries. The analyses showing Facebook high in AI citations measure all content types, not local business recommendations. For "who's a good plumber near me," citations concentrate in Google's own data, review platforms and directories. Collect Recommendations because humans read them.
YouTube
The slow door with the longest tail. A channel for a trade is a 6-to-18-month asset, not a lead spigot. But the videos keep working long after you've forgotten you made them, and YouTube is one of the most-cited sources across AI answer engines. Set expectations there and it's one of the best uses of an hour a week you'll find.
Worth knowing, and what's oversold
Long-form is what gets cited. A 2026 study by OtterlyAI, a company that sells AI-search monitoring, found roughly 94% of AI citations of YouTube go to long-form video, with Shorts under 6%. Use Shorts for reach and personality. Use the eight-minute "here's what's wrong with your water heater" video for the long game.
Read that finding carefully. The researchers note their dataset only contains videos already cited, so it explains repeat citation, not how you get cited the first time. Anyone converting it into "do X and AI will cite you" is over-reading it. None of these studies segment local queries either, so there's no evidence a local plumber's channel gets cited when someone asks who to call. It gets cited for the how-to question. Still worth having.
Links in Shorts descriptions and comments aren't clickable. Don't build a funnel on it.
Tags barely matter. Titles, descriptions and transcripts carry the weight.
TikTok
Two years ago I'd have told you to be careful. The U.S. ownership question closed in January 2026; the platform runs under a U.S. joint venture with a retrained algorithm, and the ban risk that made unpaid effort a bad bet is gone. What's left is ordinary platform risk. In February 2026 TikTok launched a Local Feed in the U.S., its first real local-discovery surface here.
What's oversold
"40% of Gen Z uses TikTok instead of Google." That line came from a Google executive in 2022 describing where young people look for lunch. It was never about hiring a plumber. The closest current read: an Adobe Express–commissioned survey in early 2026 found the share of Gen Z more likely to rely on TikTok than Google fell from 8% to 4%, and that asked which platforms are most helpful for search, people said Google 85%, Reddit 29%, ChatGPT 26%, YouTube 24%, TikTok 16%. Small Gen Z subsample, vendor-funded, so hold it as loosely as the BrightLocal number. But nobody has data showing homeowners hire trades off TikTok search.
The Local Feed's real reach is unknown. It needs users to turn on precise location, which is off by default, on a platform that just changed owners and then asked for that permission. No published adoption numbers. A reason to claim the surface early. Not a reason to bet your month on it.
"Here's what content works for trades on TikTok." I looked for a credible study and there isn't one. It's all agency blog content with no methodology. The only verified guidance is TikTok's own: location, topic, recency. Everything else, including what I'd tell you from experience, is craft, not data.
No lead capture on organic TikTok. Bio link only, so make it your booking page and nothing else.
TV & radio
I spent twenty years in a television newsroom, so take this from someone who loves the medium. For a one-to-five-truck operation, buying TV or radio is almost always the wrong move.
The arithmetic is the argument. Say you clear $5,000 on a good job. A real local TV schedule, call it $20,000 a month, and radio can be had for a good deal less, means four jobs a month to break even, before profit, with no way to know which four came from the spot.
There's a second problem. Broadcast is a branding medium and branding is what big budgets do well. You can't buy I'll tell you the truth and I won't upsell you in a thirty-second spot. It doesn't survive the format. You prove it, in the places where proof accumulates.
And when do you get out of it? With digital, today. Try that with a radio contract.
The free version, which is good
When it does make sense
When brand recall is your constraint rather than lead volume: multiple crews, a real service footprint, already winning the searches. If you're reading a free checklist about claiming your Google profile, you're not there yet. Almost nobody is.
Angi, Thumbtack, HomeAdvisor
I put these in a trash can in my drawing, and I want to be careful, because the reflexive contractor rant about lead marketplaces is usually lazier than the truth. Facts first, then the counter-argument, then what I'd do.
The facts, from primary sources
The FTC brought an action against HomeAdvisor, an Angi company, and finalized the order in April 2023. The allegations covered conduct going back to 2014: that HomeAdvisor falsely claimed the leads it sold matched the pro's stated service type and geography, made unsubstantiated claims about how often leads turn into work, and misrepresented a software subscription's first month as free. The settlement was up to $7.2 million.
The redress math is the part worth sitting with. Contractors were paying a roughly $288 annual membership plus per-lead fees. Refunds ran up to $30 per affected pro for the lead misrepresentation and up to $59.99 for the software charge. About $3 million went out the door.
The economics are visibly strained. From Angi's Q1 2026 release: revenue down 3% year over year, monthly active pros down 22% to about 105,000, pro churn up from 4.5% to 5.0% a month, consumer marketing spend at 43% of revenue against 29% a year earlier. Their third-party network channel fell hardest, requests down 55% and revenue down 56%, while their proprietary channel grew 17%.
You should have that from me rather than from someone correcting me: the company isn't collapsing. But it's spending nearly half of every dollar it takes in to keep the top of the funnel full. That cost has to land somewhere, and there aren't many places it can land.
Their BBB file shows roughly 1,870 complaints closed over three years. They hold an A+ rating and aren't accredited; accreditation is a paid membership, so read nothing into that either way. Read the pattern in the complaints instead: contractors alleging charges for leads outside their category with no practical way to dispute, and charges continuing after cancellation. Allegations, not findings. But there are a lot of them, and they rhyme with what the FTC alleged.
"Your lead goes to four other guys" is no longer universally true, and if you say it flatly somebody will correct you. Angi changed its matching model in early 2025 so homeowners choose which pros to contact rather than having a request blasted out. Thumbtack and others still commonly send the same request to several pros. The platforms don't publish the number. Assume you're competing and paying for the privilege, but don't overstate the mechanics.
When these do make sense
Real cases exist, and pretending otherwise would make the rest of this less credible.
What makes it work is discipline: a hard monthly cap, cost tracked per booked job rather than per lead, a kill date committed to before you start, and a plan to turn every job into a review on a platform you own. The failure isn't using Angi. It's using Angi as your permanent lead source, renting your customer base forever at rent that goes up.
Claim the free profile. Never enter a card. A free listing is still a name-address-phone citation and a place reviews accumulate, and it costs fifteen minutes. But claiming triggers persistent sales outreach, and the documented complaint pattern is charges that start and don't stop. If any step of claiming asks for payment details "just to verify," stop there. And don't let anyone sell you the free profile as an AI-visibility play; in the citation analyses I've seen, Angi, Thumbtack and HomeAdvisor sit well behind Yelp and BBB for the queries you care about.
Reviews
Everything above assumes them. They drive map placement, they lower what you pay for leads, and they're what an assistant corroborates you against. Nothing here substitutes.
Recency beats volume. A competitor with 200 reviews and nothing new in eighteen months is beatable by 20 reviews in six months. You're not as far behind as the number on his profile makes you feel.
What to count
Most of the numbers people track for a service business are now misleading. These aren't.
| Measure | Why this one |
|---|---|
| Answer rate on inbound calls, business hours | The new fundamental. A missed call used to be a lost lead. Now it's an invisible loss: the assistant dials the next contractor and logs nothing. |
| Time to first contact on web leads | Minutes, not hours. It also lowers what you pay per lead on Google, since responsiveness feeds ad rank. |
| Bookings completed vs. leads needing a callback | Tells you whether the calendar works or is decoration. |
| Review velocity, last 90 days | Weighted heaviest. |
| Name/address/phone audit, pass or fail, quarterly | Fifteen minutes, four times a year. |
| Citation spot-check, monthly | Ask an assistant your top ten symptom and cost questions and see whether you come up. Tedious, manual, and the only honest measure there is. |
| Gross profit by channel, not lead count | The one that saves you from your own reporting. Twelve drain calls will out-report one repipe and be worth a fraction as much. Count dollars. |
Don't judge your symptom and cost content by website traffic. Those questions get answered without anyone clicking. Sessions are structurally misleading here, and measuring that way is the most reliable method I know for killing a program that's working.
The order to do it in
There are ninety-odd checkboxes on this page. Don't try to do them in a week; you'll quit.
The first hour
Find your Google Business Profile. Claim it, or create it if it truly doesn't exist, and start verification, because that queue runs in the background while you do everything else. Then fix your primary category. That hour outperforms the next twenty.
The first weekend
Finish the profile: services in customer language, real photos, hours, service area. Set up a free booking tool and paste the link into the appointment field. Write your name, address and phone in a note and make Yelp, Apple, Bing and Facebook match it. Then pull your customer list and ask for reviews, all of them, one time, in your own words.
The first quarter
One symptom page a week, harvested from calls you took. All your emergency pages written before the season that needs them. Claim BBB, Nextdoor, Foursquare, MapQuest. Automate the review ask so it stops depending on your memory. Pick one social platform, the one you'll keep doing, and post to it consistently instead of posting to four badly.
Be king of one neighborhood before you touch the next. Four finished jobs on one block, with four reviews and four sets of photos, beats forty scattered across a metro. It's defensible to a customer, to Google and to an assistant, and unlike most of marketing it's just true.
The prompts
These are the ones I actually use. Paste them into ChatGPT, Claude, Gemini — whatever you've got open. Fill in the brackets. Nothing here needs a subscription; the free tiers handle all of it.
One note before you use them: an assistant will happily invent a price range, a neighborhood detail, or a customer. You are the fact-checker. Every one of these prompts is written to pull the answer out of your head and put it in publishable shape — not to make things up on your behalf. If it tells you something about your own trade that you know is wrong, it's wrong. Say so and make it fix it.
You are an experienced local marketing strategist who works specifically with home-service contractors. I run a [TRADE — e.g. plumbing] business in [CITY/AREA]. I have [NUMBER] trucks/crews and my average job is around $[AMOUNT]. I want you to audit how findable and bookable my business is right now, and give me a prioritized list of free things I can fix myself. Do this in three steps: STEP 1 — Interview me. Ask me up to 12 short questions, a few at a time, to find out: - whether I have a claimed and verified Google Business Profile, and what my primary category is - whether someone can book me online right now, and whether that calendar shows my real availability - how many reviews I have, on which platforms, and when the most recent one was - which of these I've claimed: Yelp, Apple Business, Bing Places, BBB, Nextdoor, Facebook - whether my business name, address and phone are identical everywhere - what my highest-margin job is, and which ZIP codes or neighborhoods I actually want to work in - what I'm currently spending on leads or advertising, if anything - how quickly inbound calls and web leads actually get answered Wait for my answers before continuing. STEP 2 — Score me. Give me a simple A–F grade on each of these six areas, with one sentence of justification each: 1. Google Business Profile 2. Bookability (can a stranger, or an AI assistant, book me in under 90 seconds?) 3. Reviews (volume, recency, and whether the asking is automated) 4. Consistency of my name/address/phone across the web 5. Website as a place to corroborate and convert, not to be discovered 6. Whether every channel leads to the same single booking path STEP 3 — Give me a punch list. Exactly 10 actions, ranked by (impact ÷ effort). For each one: what to do, roughly how long it takes, and what specifically changes if I do it. Mark anything that costs money — I want the free list first. Rules: Be blunt. Don't flatter me. If something I'm doing is actively hurting me, say so plainly. Don't recommend buying software unless there is genuinely no free way to do it.
I run a [TRADE] business in [CITY, STATE]. My most profitable work is [DESCRIBE YOUR MONEY JOB]. Work I'll take but don't chase: [LIST]. Work I don't want at all: [LIST]. Help me get my Google Business Profile categories and services right. 1. Recommend ONE primary category and up to 4 secondary categories, using real Google Business Profile category names. For the primary, explain in one sentence why it beats the obvious alternative. 2. Draft my Services list — 10 to 20 entries, written in the exact words a homeowner would use on the phone, not trade language. Put my money job at the top. For each, add a one-sentence description a real customer would understand. 3. Flag anything I listed that I should REMOVE because it invites the low-margin calls I don't want. 4. Give me the same list formatted as the service pages my website should have, so the two match word for word. Then ask me one question: is there anything on the list I can't actually do well, today, in my service area? I'd rather cut it than rank for it.
Write a web page for my [TRADE] business in [CITY/AREA] answering this exact customer question: "[PASTE THE QUESTION IN THE CUSTOMER'S OWN WORDS — how they actually said it on the phone]" Follow this structure exactly: 1. HEADLINE — the symptom, in the customer's words. Not a topic. Not "Common X Problems Explained." 2. THE ANSWER — first two sentences. Name the 2 or 3 real causes, most likely first. No introduction, no throat-clearing. 3. HOW TO TELL WHICH ONE — what a homeowner can check themselves in about 60 seconds, safely, before calling anyone. 4. WHAT IT COSTS — a real range, with what moves it up or down. Leave the numbers as [BLANK] for me to fill in; do NOT invent prices. 5. LOCAL DETAIL — one short paragraph. I'll give you the facts; use only what I give you. Here is what's true about my area: [BUILD ERA, CONSTRUCTION TYPE, ORIGINAL MATERIALS, SOIL, TREES, ANYTHING ELSE]. 6. WHAT WE'D DO — plain language, then a single clear line pointing to booking. Rules: - One question per page. Don't broaden it. - Do not invent a customer, a name, or a story. Write the question; never invent the questioner. - Do not write any geographic sentence that would also be true two neighborhoods over. Test each one: would a competitor there be lying if they published it? If not, cut it. - Do not write anything I wouldn't say out loud standing in someone's yard. - 500-800 words. Plain English. Short paragraphs. After the draft, list every factual claim you made that I need to verify before I publish.
Write an emergency instruction page for my [TRADE] website. The emergency is: [E.G. WATER ACTIVELY LEAKING / SEWAGE BACKING UP / NO HEAT IN A FREEZE / BURNING SMELL FROM THE FURNACE]. Assume the reader is standing in the mess, on a phone, scared, and will read about four lines before acting. Structure: 1. FIRST LINE = the single physical action to take right now. Nothing before it. No preamble. 2. Then 3-5 numbered steps, shortest possible sentences. 3. A "STOP — call someone now, don't DIY" section: exactly which conditions mean they should not touch it. 4. What to do while waiting to protect the house from further damage. 5. What this typically costs to fix — leave as [BLANK] for me to fill in. 6. One line at the bottom pointing to booking and a phone number. Also tell me what PHOTOS I need to shoot and add to this page to make it genuinely useful — for example, both types of shut-off valve, where they sit in a house like the ones in my area, and which way to turn them. Be specific about the shots. Write it so it's useful to someone who will never hire me. That's the point.
I'm a [TRADE] working in these specific areas: [ZIP CODES / NEIGHBORHOOD NAMES]. Interview me to build a "neighborhood fact bank" — the verifiably true, locally specific facts that make my content impossible for a competitor in the next town to copy. Ask me, one area at a time: - What era were the houses built, and what's the construction type (slab, pier-and-beam, crawlspace, basement)? - What were the original materials I keep finding? (cast iron, Orangeburg, galvanized, polybutylene, original ductwork, aluminum wiring, whatever applies to my trade) - What's the tree cover, and are there species whose roots I deal with specifically? - How old is the typical system relative to when it should have been replaced? - What's the one failure I see over and over in this area that I don't see elsewhere? - Anything about permits, the local municipality, or code that's specific to this area? Then, for each area, produce: (a) A short fact sheet I can keep and reuse. (b) 5 content headlines that are only writable by someone who actually works there. (c) A blunt verdict: does this area deserve its own page on my website, or just a mention? Rule: it gets a page ONLY if I can write 400 words that are true here and false in the next neighborhood over. Tell me honestly which ones fail that test. Use only what I tell you. Do not fill gaps with general knowledge about the region — if I don't know something, mark it "verify" rather than guessing.
I'm going to paste in reviews of my [TRADE] business. Write a reply to each one. Rules for every reply: - Sound like a working tradesperson wrote it, not a marketing department. Short. Warm. No corporate phrases, no "we strive to," no exclamation-point spam. - 2-3 sentences maximum. - Where I've told you the job and the neighborhood, name them naturally in MY words — e.g. "glad we got that cast iron replaced for you over in [NEIGHBORHOOD]." This is my own honest description of the work, which is allowed and useful. Never ask the customer to say anything. - Never offer a discount, a gift, or anything of value in a reply. - Vary the structure. Ten replies that read from the same template look automated, because they are. For negative reviews: acknowledge the specific thing, don't argue the facts in public, state plainly what I'd like to do about it, and move it offline with a real phone number. Never blame the customer. Never mention what they paid. Here are the reviews, with the job and neighborhood where I know them: [PASTE]
Help me decide whether a paid lead source is worth it. Be a skeptic, not a cheerleader. My numbers: - Average job value: $[AMOUNT] - Roughly what I keep after materials and labor: $[AMOUNT or %] - Cost per lead they're quoting: $[AMOUNT] - Any membership or subscription fee: $[AMOUNT] per [MONTH/YEAR] - Contract length and cancellation terms, if I know them: [DETAILS] - My realistic close rate on a COLD lead from a stranger: [%] — and note this is usually far lower than my close rate on referrals, so if I've given you my referral number, tell me. Calculate: 1. My true cost per BOOKED JOB, including the membership fee amortized over realistic lead volume. 2. What percentage of my profit on each job that eats. 3. The break-even close rate — below which I'm losing money on every job. 4. How many leads a month I'd need for this to move my business at all. Then tell me: - Whether this passes or fails, in one sentence. - What the same money would buy on Google Local Services Ads instead — where I'm paying for a direct contact rather than a resold lead record, though a homeowner can still contact more than one business — and what I'd need to check to compare the two fairly. - If it passes: the exact guardrails I should set before I start — monthly cap, kill date, and what I need to be tracking from lead one. Finally, flag any assumption above where I'm probably kidding myself. Especially the close rate.
Here is my canonical business information, exactly as it should appear everywhere: Business name: [EXACT NAME] Address: [EXACT ADDRESS, or "service area business — no public address"] Phone: [EXACT NUMBER, formatted the way I want it] Website: [URL] Hours: [HOURS] Now here is what each of my listings currently says. I'll paste them in: [PASTE WHAT GOOGLE SAYS] [PASTE WHAT YELP SAYS] [PASTE WHAT FACEBOOK SAYS] [PASTE WHAT MY WEBSITE FOOTER SAYS] [PASTE ANY OTHERS] Produce a table: platform, field, what it says now, what it should say, and whether it matches — flagging even tiny differences. "St" vs "Street," "&" vs "and," a suite number present in one place and missing in another, a different phone format. Those small ones are the ones that actually break things. Then give me a fix list in priority order, and tell me which of these differences are cosmetic and which ones genuinely matter.
I'm a [TRADE] in [CITY/AREA]. My business is [BUSINESS NAME].
First: give me a list of 10 questions a real homeowner in my area would ask an AI assistant that should surface a business like mine. Mix them across three types:
- Direct hiring intent ("who should I call to...")
- Symptom questions ("why is my ... doing ...")
- Cost and decision questions ("how much does ... cost" / "should I repair or replace ...")
I'm going to ask each of those in a fresh chat, in this assistant and in one or two others, and record what comes back.
Then build me a simple tracking table I can fill in and reuse every month, with columns for: the question, whether my business was named, which competitors were named, what sources were cited, and the date.
Finally, tell me: for each of the three question types, what specifically I would need to change to have a realistic shot at being named — and be honest about which ones I probably can't win and shouldn't chase.
Do them in the order they're numbered the first time through, then keep three of them in rotation forever: #3 once a week off a real phone call, #6 whenever reviews come in, and #9 on the first of the month. That's a marketing department that costs nothing and takes about an hour a week.