American Heating & Cooling · Columbus, GA & Smiths Station, AL
Christina, thank you again for adding me to the Local Services account. As promised, I went through it end to end, every lead, every call, and every message since the program restarted in February, and this report is the full picture of what I found. I want to say up front that the news is mostly good: this is a healthy, growing account that buys real customers at a fair price, and your team handles what reaches them well. What I found beyond that are four specific places where paid leads slip through without anyone ever seeing them, and all four are the kind of gap a system closes, not the kind that asks more of your people.
In plain terms: at roughly $188 in advertising per booked job, against average tickets that run well past $2,000 in this trade, Local Services is one of the best purchases this business makes. The four findings in Section 4 are about protecting that purchase. The largest of them is quantified precisely: since February, 11 of your 19 message leads, all of them paid for, never received a response of any kind. I verified each one individually before writing that sentence. My recommendations for closing each gap follow in Section 5, and every recommendation keeps your team first in the flow.
All account data in this report was pulled directly from your Local Services account on July 20, 2026, through Google's advertising reporting interface. That includes all 81 leads received between February 1 and July 20, and all conversation activity connected to them: phone calls with their durations, messages, texts, and emails, with timestamps and the direction of each. For the unanswered-message finding, I examined each of the 19 message leads individually and checked for a reply by text, a reply by email, and any subsequent phone contact with that same lead before counting it as unanswered. Spend figures come from the account's own reporting; cost-per-lead figures marked as such come from your Local Services dashboard; review figures come from your public Google profile. Two limits worth naming honestly: connected-call durations are available on a subset of calls, and the reporting does not expose how many inbound calls reached voicemail, so I make no numeric claim there.
The restart has worked. February produced 3 leads on $61.88 of spend; July has produced 30 leads on $1,084.46 with ten days still to go. June and July both ran essentially at your full $288-per-week budget, which means the account is no longer testing, it is deployed, and Google is feeding it more volume each month. The complete monthly ledger is in Appendix A1.
Cost efficiency is where I would want it. Dividing the year's $3,298.84 across all 81 leads gives $40.73 per lead received; across only the 53 leads Google actually charged, $62.24. Your dashboard's blended figure of $58.50 per lead ($66.85 for phone leads, $47.50 for message leads) sits right in line with that, and all of it compares well against the published industry benchmark for HVAC Local Services of roughly $51 per lead (Searchlight Digital, February 2026, from $14.9M in tracked contractor spend). You are buying leads at market price or better, and about a third of your leads (28 of 81) arrived free because Google credited them automatically.
Quality is equally sound. Of the connected calls I could measure, 87% ran longer than 30 seconds, the median conversation lasted about a minute and a half, and the longest ran nearly nine minutes. These are homeowners with real problems, not spam. Your office converts roughly one charged lead in three into a booked job, around 17 to 18 booked jobs so far this year at roughly $188 in advertising each. For reference, the trade's benchmark booking rate on this channel runs near 44% of leads (same Searchlight dataset), which with nights and weekends currently uncovered reads to me as headroom rather than criticism. And the reputation those jobs have earned, a 4.8-star rating across 94 reviews, is an asset most competitors in the market do not have.
One structural note before the findings: nearly one in four of your leads now arrives as a message rather than a call, 19 of 81 this year, and messages are your cheapest lead at $47.50. Google has been steadily pushing customers toward this channel. That trend is what gives Finding 1 its weight.
Four findings, in order of cost. I want to repeat the framing from the summary because I mean it: none of these are effort problems. Melanie and the front desk are quick and good when a lead reaches them, the data shows exactly that. These are the places where no system exists to catch what falls outside those moments.
Since February, 19 customers contacted the business through Local Services messages. Eight received a reply, and the account data speaks well of the team here: those replies went out in a median of 14 minutes, six of the eight within the hour. The other eleven never received anything: no text reply, no email reply, and, having checked each lead's full history, no phone call afterward either. Because Google charges for message leads regardless of whether they are answered, all eleven were paid for. At your $47.50 message-lead cost, that is $522.50 spent this year on customers who reached out and heard nothing back, and at the current pace and mix it runs to roughly $2,000 over a full year.
The pattern is binary, and that is what tells me it is a coverage gap rather than a people gap. When someone happens to be at the desk and sees the message, the response is fast and the lead gets a real chance. When nobody sees it, there is no alert, no queue, and no backstop, so it dies silently. With messages now nearly a quarter of your lead flow and climbing, this is the account's most expensive leak, and it compounds, because of how Google treats responsiveness.
Today, a Local Services call that Melanie or Christina cannot pick up goes to voicemail with a callback later. The reporting does not tell me how many calls that affects, and I will not invent a number where I do not have one. What I can tell you is what the research says about the ones it does affect: lead-response studies consistently find that contact rates collapse within minutes of an inquiry, with the odds of reaching a lead several times higher in the first five minutes than at thirty. A homeowner in an Alabama July with a dead air conditioner rarely waits for a callback; they call the next name on the list. Each of those calls is a $66.85 lead, and your own call data shows what they are worth when answered: 87% become conversations of substance.
Your 4.8-star rating across 94 reviews is genuinely excellent, and it was earned the slow way: customers delighted enough to find the review page on their own, occasionally helped by a manual ask when someone thinks of it. That is essentially the only mechanism at work. Over the past year the business has added roughly three reviews a month, while completing around six booked Local Services jobs a month at the June and July pace plus everything that arrives through other channels, a small fraction of the completed work ever becomes a review. Home-services businesses that ask systematically after every job typically capture 20 to 40% of jobs as reviews (industry reputation-platform reporting), which at your volume supports six to ten a month. About a third of the reviews you do receive get a response; the rest sit unanswered, and responsiveness to reviews is itself one of Google's stated Local Services inputs.
I want to be precise about why this is a Local Services finding and not just a reputation nicety: review count, rating, and responsiveness are ranking inputs in Local Services, per Google's own documentation. Ranking decides how many leads your $288 a week buys. A steady six to ten reviews a month is the cheapest lever available for pushing lead volume up and effective cost per lead down without spending another advertising dollar.
I verified this one directly against the account's lead records rather than assuming it. Of the 81 leads this year, zero carry Google's booked status, every one still sits as new or active, and across the entire year there is exactly one credit request on file, still pending, with none completed. The 28 leads you were not charged for carry no credit activity at all: Google simply chose not to bill them at its own discretion, not because anyone on your side disputed them. Both halves of that cost something. Junk leads, the wrong-service or out-of-area calls every account gets, are being paid for when a thirty-second dispute would credit them back. And with no leads ever marked booked, Google's matching system receives no feedback about which leads suit this business, which dulls its targeting over time. A two-minute weekly pass over the lead inbox closes both halves of this gap at zero cost.
Four recommendations, one principle shared by all of them: your people stay first, and the system exists to catch what would otherwise fall. Nothing below changes how the office works on a good day, and nothing below asks Melanie to do more. Each recommendation maps to the finding it closes.
An assistant that watches the Local Services message inbox and your regular texts around the clock, and guarantees that the first reply goes out in under a minute, at 2 PM on a Tuesday or 9 PM on a Saturday alike. It asks the two or three questions a good dispatcher would ask, one at a time: what is the system doing, how urgent, what part of town. Then it books the appointment inside the same conversation, offering real arrival windows and confirming the slot the moment the customer picks one, because handing a warm text lead a phone number to call is where these leads go to die.
Anything that should reach a person does: emergencies, upset customers, large install inquiries, or anyone who simply asks. In those cases the office receives a clean summary of the conversation so far, not a transcript to wade through. The eleven silent leads in Finding 4.1 stop existing the week this turns on.
This is the update I described in my last email, and it is aimed squarely at Finding 4.2. Every ads call rings the office first, exactly as it does today, and on a normal day nothing changes: Melanie answers, and the system stays out of the way. The change is what happens when the call is not picked up quickly, or comes in after hours. Instead of voicemail, the assistant answers, greets the caller properly, takes the details, and books the appointment on the spot. Both sides are recorded with full notes, so you hold a complete record of every lead regardless of who answered it.
The practical effect is that the $66.85 lead calling at 7 PM gets a competent answer and a booked slot instead of a beep, and every one of those conversations earns the same responsiveness credit with Google that Finding 4.1 describes.
A few hours after a job is marked complete, the assistant places a short courtesy call: how did the visit go, is everything running well. When the answer is positive, the customer receives a text with a direct link to your Google review page along with a short suggested draft based on what they just said, which they can edit and post in under a minute. Removing the what-do-I-write friction is what moves capture rates from under 5% toward the 20 to 40% the industry sees with systematic asks. When the answer is negative, the call comes straight to the office while the problem is still fresh, so you make it right before it becomes a public one-star surprise. Every customer receives the same review invitation regardless of sentiment, which keeps the process fully inside Google's review policies.
One ask per job, always within polite hours. At your current volume this supports six to ten new reviews a month instead of three, and each of them feeds the ranking flywheel described in Finding 4.3.
A standing weekly pass over the lead inbox that does the two things the front desk never has time for: every lead that became a booking gets marked as booked, and every lead that did not gets Google's short lead-quality survey completed, with the ones that never belonged, the wrong-service calls, the out-of-area requests, the solicitors, disputed for credit.
The payoff runs in both directions. The disputes come back as credits, found money at $47.50 to $66.85 per bad lead. And the marking is how a business tells Google's matching algorithm exactly what a quality lead looks like: every booking flagged and every survey filed is a training signal, and accounts that feed the system this feedback get matched to better leads over time. Right now, with zero leads ever marked booked and one credit request all year, the algorithm is working blind. This routine runs on the account access I already hold, so it asks nothing of the office and can begin immediately.
On sequencing: the call coverage is already built and in final testing, so it lands first. The message system connects to the same account access I already hold, and the review process can begin with a simple job-done mark the office already makes, then plug into your field software later. The one confirmation I need from you is which field software the team runs day to day, I believe it is Housecall Pro, and that answer shapes only the review process, nothing else waits on it.
Every figure used in this report, in full. All account data as of July 20, 2026.
| Month | Spend | Impressions | Clicks | Leads |
|---|---|---|---|---|
| February | $61.88 | 412 | 31 | 3 |
| March | $420.95 | 1,603 | 133 | 11 |
| April | $78.26 | 1,022 | 75 | 2 |
| May | $423.38 | 1,306 | 97 | 7 |
| June | $1,229.91 | 2,319 | 196 | 28 |
| July (to the 20th) | $1,084.46 | 1,460 | 147 | 30 |
| Total | $3,298.84 | 8,122 | 679 | 81 |
| Measure | Value | Note |
|---|---|---|
| Leads received | 81 | 62 phone (77%) · 19 message (23%) |
| Leads charged | 53 | 65% of received |
| Leads not charged by Google | 28 | 35%; Google's own billing decisions, no disputes involved |
| Booked jobs (est.) | ~17–18 | at the office's ~33% booking rate on charged leads |
| Leads marked booked in the account | 0 | verified in lead records; Finding 4.4 |
| Credit requests filed, 2026 | 1 | still pending; none completed |
| Measure | Value | Source |
|---|---|---|
| Cost per lead received | $40.73 | $3,298.84 / 81 leads, account data |
| Cost per charged lead | $62.24 | $3,298.84 / 53 charged, account data |
| Blended CPL (dashboard) | $58.50 | your LSA dashboard |
| Phone lead CPL (dashboard) | $66.85 | your LSA dashboard |
| Message lead CPL (dashboard) | $47.50 | your LSA dashboard |
| Cost per booked job (est.) | ~$188 | $3,298.84 / ~17.5 booked |
| Industry HVAC LSA benchmark | ~$51 | Searchlight Digital, Feb 2026 |
| Weekly budget | $288 | fully deployed in Jun–Jul |
| Measure | Value | Note |
|---|---|---|
| Message leads received | 19 | 23% of all leads |
| Answered | 8 | median first reply 14 min; 6 of 8 within the hour |
| Never answered | 11 | 58%; no text, email, or later phone contact on any |
| Of the 11, charged | 11 | every silent lead was paid for |
| Cost of unanswered leads | $522.50 | 11 × $47.50 |
| Run-rate if unchanged | ~$2,000/yr | at the current pace and message share |
| Measure | Value | Note |
|---|---|---|
| Calls measured | 74 | duration available on a subset of all calls |
| Under 30 seconds | 10 | 13.5%, the likely junk/misdials |
| 30–90 seconds | 29 | 39% |
| Over 90 seconds | 35 | 47% |
| Median duration | 86 sec | longest ran nearly 9 minutes |
| Measure | Value | Note |
|---|---|---|
| Rating | 4.8 ★ | public Google profile |
| Reviews, lifetime | 94 | |
| New reviews per month | ~3 | past-12-month average; a small fraction of completed jobs |
| Reviews receiving a response | ~1 in 3 | responsiveness is an LSA ranking input |
| Supportable at current volume | 6–10/mo | at industry 20–40% capture with systematic asks |
The heart of this report is simple: you have built something that works. The account buys a booked job for about $122, your team treats the leads that reach them well, and the market clearly trusts you, 94 reviews at 4.8 stars says so. My job is to make sure nothing you pay for falls in the seams, and right now the seams are messages after hours, calls past the second ring, and finished jobs that never turn into reviews. Each has a clean fix, each fix keeps your people in front, and the first one is already being tested. I would be glad to walk you and Greg through any part of this, line by line if you like, whenever suits you both. Thank you, as always, for the trust you place in me with this account.