Marketing Engineering for Local and Service Businesses: The Small Version That Pays

Forty percent of law firms picked up the phone

In 2024 the legal software company Clio hired a research firm to contact 500 law firms as a new client would. Only 40 percent picked up when called, and 48 percent could not be reached by phone at all, even after a message. One in three answered the email.

Those firms were not short of marketing. The prospect had already found them, chosen them and dialed. Five years earlier, in Clio's 2019 study, 56 percent of firms had picked up.

The work died in the gap between one step and the next.

That gap is where marketing engineering pays in a local business. A marketing engineer builds the systems behind the marketing, so the next step happens because a system does it, not because someone remembered.

Ask the owner of a 10 to 50 person shop what is missing and the answer is usually more leads. Often the real answer is fewer dropped handoffs.

This is the small version, sized for an HVAC company, a med spa, a law firm or a dental office. Four systems, what they cost, what to skip, and a one-page map you can fill in this week.

The small version of a big-company job

The marketing engineers in the news work at software companies. Figma's Marketing Engineer, AI Deployment posting lists $127,000 to $296,000 and asks for agents, data pipelines and code. A 20-person HVAC company is not hiring that person, and should not.

What carries over is the habit, not the headcount. Pick one task your team repeats every week and drops some of the time. Build the thing that does it, with a person approving anything a customer will see.

In a local business those tasks are short and plain: call back the lead, send the quote, ask for the review, remind the patient. Nobody skips them on purpose. They get skipped on the busy days, which are also the days with the most leads.

A system counts when it still does the task next Tuesday with nobody thinking about it.

Who builds it is a separate decision. You can train someone you already employ, rent an outside team, or buy one tool per task.

The HVAC example, moved from the software company to the shop

Greg Isenberg used an HVAC example in his 35-minute episode on the role, and it is the best one going. His imaginary company sold software to commercial HVAC contractors. Its sharpest marketing angle was not a promise to run the business better.

It was one moment on a service call. The technician finishes the repair, sees the unit is near the end of its life, and the replacement quote never goes out. His point was that a pain that specific beats a general promise in every ad, email and page.

Now move the example from the software company to the contractor. That unsent quote is a handoff, and it is often the most valuable one in the shop. The repair paid for the truck roll; the replacement is the job.

A marketing engineer's first build for that contractor would not be a content calendar. It would be a rule: every completed call with a replacement flag creates a quote within a day, a text two days later, and a call from a person after a week. On Monday the owner sees how many flags became quotes and how many quotes became jobs.

Every trade has its version: a dental office has treatment plans presented and never scheduled, and a med spa has clients due back at the treatment interval. A law firm has consults that never signed.

Those are the money handoffs. More traffic fixes none of them.

The Handoff Map: find the gap before you buy anything

Before any tool, write down the path a customer takes from finding you to coming back. In a service business it has seven steps, and each one is a handoff from one person or channel to the next. We call this the Handoff Map.

Fill it from records you already keep. Your phone provider can show missed calls by hour of day. Your calendar shows bookings and no-shows, and your field service or practice software shows the quotes and treatment plans that were written.

For each step, write two numbers for last month: how many arrived and how many moved on. The step with the biggest drop is where your first system goes.

Fill it from records, not memory. Memory keeps the ad you paid for and forgets the quote nobody sent.

One map, one afternoon. It sets the order for everything below, and the template is in the figure.

Systems one and two: answer fast, and make every job create the next

System one answers every inquiry in minutes. The best-known study of web leads found that firms that tried to reach a lead within an hour were nearly seven times as likely to qualify it as firms that waited one hour longer. The three-layer setup, an instant text, a human reply and an alert to the owner, is in speed to lead.

Phones are the harder half. One full-time receptionist covers 40 of the 168 hours in a week, at a US median of $18.27 an hour in May 2025, and an AI receptionist for the other hours can take the message, book the slot and pass urgent calls to a person. It should say plainly that it is automated.

Law firms carry one more rule. The ABA's first formal opinion on generative AI, issued in July 2024, says lawyers using these tools still owe competence, confidentiality and clear communication with clients. So intake automation answers, books and routes; it never gives advice.

System two makes every finished job create the next one. In med spas this is most of the business: the American Med Spa Association's 2024 survey put the average share of repeat patients at 73 percent, up from 65 percent in 2022. Those repeat visits happen only if someone asks at the right time.

The build is a set of plain rules tied to the software you already run. A completed job with a flag creates a quote, a treatment-plan reminder or a rebooking message at the right interval. A person approves the first message and the system follows up on a schedule.

The measure is one ratio: flags raised against next jobs booked. If it does not rise in four weeks, the rule is wrong, and you change the rule before you add a second system.

Systems three and four: reviews in, and one number on Monday

System three asks every finished job for a review and answers every review. Google says local results rank mainly on relevance, distance and prominence, and that more reviews and positive ratings can help a business rank. The same page says you cannot pay for a better spot.

Buyers read reviews closely and want them recent. BrightLocal's 2026 Local Consumer Review Survey found 47 percent will not use a business with fewer than 20 reviews, 74 percent care only about reviews from the last three months, and 89 percent expect the owner to reply.

So the system is short. A text with a direct review link goes out after every finished job, which is what Google's own advice on getting reviews comes down to: ask. A draft reply waits for every new review, a person sends it, and once a week the reviews are read for the words customers use, the small version of a customer truth system.

Two rules keep it legal. The FTC's 2024 rule on reviews bans fake reviews and bans paying for reviews on the condition that they say something positive. In healthcare a reply must never confirm that the reviewer is a patient: a California dental practice paid $23,000 to settle with federal regulators after replying to reviews with patient names, treatment and insurance details.

System four puts one number on your desk every Monday. Inquiries, booked, showed and sold, by source, for last week, with the change from the week before. It replaces the conversation about whether the ads are working, and the full version is the growth cockpit.

The same records go back to the ad platforms. Google's offline conversion import lets you tell Google which ad clicks became booked jobs, so its bidding learns from buyers instead of form fills. On Meta, spend by distance the way Radius Rings lays out, and judge each ring by booked jobs, not leads.

AI assistants are now part of how people find a local business. In BrightLocal's follow-up report, 45 percent of consumers used AI tools for local recommendations in the past year, up from 6 percent. You do not need a separate system for that yet; a complete profile, recent reviews and pages that answer buyer questions plainly serve both.

What it costs, and what to skip

The honest cost is mostly time, not software. Google Business Profile is free, offline conversion import is part of Google Ads, and the first build is usually rules inside tools you already pay for.

Money goes to the build and to the tools that answer phones and send texts. Vendor prices change too often to quote, so price the gap instead of the tool.

Here is the sum with illustrative numbers. Say your map shows 30 missed or slow inquiries a month, one in four would have booked, and a first job is worth $400. That is $3,000 a month of work going to someone else.

Put your own three numbers in. If catching those inquiries costs less than they are worth, build it; if not, the map is pointing you at a different step.

Then the owner's time: about 30 minutes on Monday to read the number and approve messages, plus an hour a week in the first month to correct drafts. A system you will not read is a system you should not buy.

Skip AI that dials strangers. In February 2024 the FCC ruled that calls using AI-generated voices count as artificial under the Telephone Consumer Protection Act, which requires prior consent. Answering your own phone is one thing; outbound calls need permission first.

Skip tracking pixels on patient pages. The federal guidance on online tracking by HIPAA-covered entities says tracking tools must not hand protected health information to vendors without the safeguards the rule requires. If your practice is a covered entity, check every pixel and chat widget with your compliance adviser before an ad goes live.

Skip volume before the handoffs hold. More ads into a phone nobody answers make the problem more expensive, not smaller. And skip building all four systems at once; four half-built systems are four new ways to drop a customer.

"We are a 15-person shop. This is tech-company stuff."

Half of that is true. The postings, the salaries and the tools in the news belong to software companies with growth teams, and a local business does not need a data warehouse or a fleet of agents.

The other half is not. The jobs these systems do are older than the title: call back fast, send the quote, ask for the review, count what came in. What changed is the cost of building them, because rules that once needed a developer can now be written and checked by one careful person with an AI assistant.

There are also good reasons not to start. If you cannot take more work this quarter, fix capacity first. If your phones are answered in minutes and reviews arrive every week, your next dollar may belong in ads, not systems.

We would rather say that here than build you a system you do not need.

How we would start in your shop

Anar is a marketing engineering team in Phoenix, and we work with companies from 10 to 500 people. An engagement starts with an audit in weeks one and two, builds one system in weeks three to eight, and then hands it to your team or keeps it running. We do not start the second system until the first has moved its number.

The same rules run our own agency. Our article library cannot publish until a script confirms every external link loads, and every ad render is measured by scripts and checked by a separate reviewing agent before a person sees it.

Each trade has its own constraints. Contractors can see what clicks cost in the trade on our HVAC page, and there are matching pages for med spas, law firms and dental practices.

You are not committing to anything by asking. Bring your filled-in Handoff Map, and we will tell you which row we would fix first and how our marketing engineering engagements would measure it.

What to read next, and why

The Marketing Engineering Audit [tool]. Twenty questions that score your growth systems. The Handoff Map tells you where the gap is; the audit tells you how far behind you are.

The First 90 Days With a Marketing Engineer [plan]. One system, not five, with the red flags to watch for by day 30. The right pace for a shop with no spare people.

AI for Law Firms and CPA Firms [sector]. Intake and scheduling first, professional judgment last. The order we would follow in any firm that sells expertise.

So the simple takeaway is this: before you buy more leads, count the ones you already dropped, and build the one system that catches the biggest drop.