Is a managed AI receptionist better than a DIY AI phone tool?

You've priced both versions of the same idea. One is a login: you build the call flow, point it at your calendar, pick a voice, and your phone is handled by Sunday night. The other is done for you, costs more, and from the outside it looks like the same robot answering the same phone.

So the question is fair, and every owner asks it. What's the extra money buying? Somebody clicking buttons you could click yourself?

Here's the short version. Both are a monthly subscription. The difference is who does the work on it. With a DIY tool you build the call flow, keep it current, and notice when it starts getting calls wrong. With a managed receptionist somebody else does all three and answers for it when it slips. So the question isn't which software is better, it's who's going to run it in month seven. Go DIY if that's honestly you. Go managed if it honestly isn't.

Worth saying up front: Willison is a managed AI receptionist, so we sit on one side of this comparison. Weigh what follows with that in mind.

Why this feels like a solved problem

Because every other tool you've bought did work this way, and the demo backs it up. You set up the field software, the scheduling app, the invoicing: rough for a week, then it runs. Then you talk to a phone agent, it sounds like a person, it books a fake job on a fake calendar without stumbling once, and it's hard to sit through that and believe there's much left to do.

But a phone line isn't an invoicing app. It talks to strangers while you're up on a roof, and its mistakes happen where you can't see them. That's what makes this a different kind of purchase. Four things about it are easy to assume, and each one costs money if it turns out to be wrong.

Myth: if it's answering, it's working

Answering is the easy half. The half that pays is the next ninety seconds: asking the questions that sort a real job from a tire-kicker, and then asking for the work.

Invoca's 2026 home services report found that on 55% of the home services calls it scored, nobody asked the caller to buy or book the job. Those are calls real people answered, perfectly pleasantly. On better than half of them, the ask never came.

A phone tool does whatever you configured, and what most owners configure on the first pass is helpful. Helpful takes a name and a number and says somebody will be in touch. It sounds great in a test call. It's the same failure that report measured, except now it runs at three in the morning and never takes a lunch break.

Myth: you set it up once

Your configuration is a snapshot of the business on the day you wrote it. The business doesn't hold still.

You raise prices in February. You drop the one job type that was never worth the truck roll. You add two towns and pull out of a third. You stop taking Saturdays. Your part-timer leaves and the after-hours number changes.

Every one of those is an edit somebody has to make in the tool, and nothing prompts you to make it. The bot doesn't know your price moved, so it keeps quoting last spring's with total confidence. You find out when a homeowner repeats a number back to you that you haven't charged since March.

Myth: you'd know if it broke

This is the expensive one. Software you use every day tells you when it's broken, because you're standing right in front of it. A phone tool fails quietly, on a call you'll never hear, to a person you'll never meet.

It doesn't crash. It books an inspection in a town you pulled out of last year. It hands an active leak the same unhurried script it hands a question about gutter cleaning. It takes a polite message from the one call that should have put your phone in your hand.

There's an old measurement of how wide that blind spot can get. An audit of 2,241 U.S. companies, reported by Oldroyd and colleagues in Harvard Business Review in 2011, found 23% never responded to a web lead at all, and that among the companies responding within 30 days the average took 42 hours.

Those were web forms in 2011, not calls to a phone tool today, so take the direction and not the number. The direction is what matters here: a lead path can be badly broken for a long time without anybody inside the business noticing, because nothing about it makes a noise.

Myth: one setup covers a trade like yours

Trades that look identical from the road do not put the same phone in front of you. That same Invoca report measures how often a caller reaches a person across nine home-services sub-industries and finds a spread running from 32% to 74%. Invoca puts that gap down to routing, process and staffing, and it measures who picks up rather than how a call needs handling, so read it for what it is.

It still makes the point. Phones that look alike from outside behave nothing alike, and a setup is only ever tuned to the one whoever wrote it had in mind.

Roofing is the case worth walking through, and it isn't among the nine that report measures. In March the phone is manageable and any setup looks fine. Then a hail line goes through on a Thursday afternoon and the calls arrive in a few days instead of across a season.

They aren't all customers, either. Canvassers, adjusters, a neighbor who wants your guy's number, and three homeowners with water coming through a ceiling who'll hire whoever gets on a roof first. A setup that never had to sort those apart won't start learning on the day it matters. More on that week in how roofers stop missing calls during storm season.

What the DIY case actually gets right

Quite a lot. And since this is published by a company that sells the managed version, this is the section to read hardest.

It's genuinely cheap, and it's genuinely yours. Nobody stands between you and the settings. If a customer says something that bugs you and you want the greeting changed at ten at night, you change it at ten at night, and you don't ask anyone.

There's an honest cost on the managed side that's the mirror image of that. You don't hold the keys. A change goes through somebody else, on their clock. If you're the owner who built your own dashboard and enjoyed the afternoon, DIY is a legitimate answer and this post isn't going to talk you out of it.

What DIY costs is mostly not the invoice. It's the evenings, and nobody can tell you how many of those it takes, because it depends entirely on how much your business changes. The FAQ below has a way to size that for yourself before you commit to it.

For a sense of what an hour of skilled trade time is worth where you are, the Bureau of Labor Statistics puts median roofer pay at $26.65 an hour and HVAC mechanics at $29.33, both 2025 figures. Those are wages for the people doing the work, not for whoever runs the place, so treat them as a floor under the hour you'd be spending rather than a price on it.

So how should you actually choose?

Two questions settle it, and both are about you rather than about the software.

• Will you maintain it? Not could you. Will you, in month seven, in a busy week, when it's working well enough that nothing is screaming at you. Plenty of capable owners answer no to that, and it isn't a character flaw, it's a schedule. But an unmaintained tool is worse than the voicemail it replaced, because it sounds competent while it's wrong.
• Who's accountable when a call goes sideways? On DIY it's you, every time, including the Saturday ones. On a managed account it should be a name you can reach. If a provider calls itself managed and gets vague when you ask exactly that, you're looking at a DIY tool with a nicer invoice.

One thing this framing leaves out, and most shops your size are living in it. If your spouse, your dispatcher or a part-timer already covers the phone from eight to five, you aren't choosing a replacement for them. You're choosing what covers nights, weekends, the lunch hour and the calls that land while they're already on one. That narrows the question a lot, and it's the shape most of these end up in.

Ranked: how a trades phone gets answered, best first

Here's how the options stack up for a shop that lives by its phone, on the one measure this post is about: who carries the work of keeping it right. A list scored on price or setup speed would come out in a different order. These are approaches, not brands, and a shop can run more than one at once.

1. 1

   Willison: the best all-around, a managed AI receptionist that books the job and runs the follow-up

   Willison ranks first on this measure because none of the upkeep comes back to you. It's built for the trades, it answers every call around the clock in seconds, and ten calls landing at once are answered at once, so the last caller waits no longer than the first. It qualifies the job and books it to your calendar rather than leaving a message for you to work, and it keeps going after the pickup, chasing the quote that's gone quiet. Willison runs the software, reads the calls, and the founder reviews the account before your line goes live.

2. 2

   A live answering service

   Second on this measure, and the reason is straightforward: somebody else does the staffing, the training and the covering of the shift, which is real work you never touch. A trained person on your line at two in the morning beats a ring-out every time. What it holds about your pricing and your service area is whatever the account was built to hold, so that part still has to be specified and kept current, same as any of these.

3. 3

   A DIY AI phone tool you run yourself

   It drops a place here only because of the measure, and on other measures it would rank higher: it's the cheapest way in, it's immediate, and nobody stands between you and the settings. The whole of the upkeep is yours though, and it's the kind of upkeep nothing reminds you about. Right for the owner who'll genuinely do it, and it's worth being honest with yourself about that before you buy rather than in month seven.

4. 4

   Voicemail with a callback habit you actually keep

   It ranks above nothing, and only for the owner who really does clear the box every couple of hours. Nobody standing under a leak waits for that, so treat it as a stopgap rather than a plan.

5. 5

   Letting it ring

   The job goes to whoever picked up. Every missed call is a competitor's job.

The thing to settle before you buy either one

Buy the version that matches the honest answer, and either way put a recurring reminder in your calendar to call your own line like a stranger every month or two. That habit is worth keeping whoever runs the thing, and it's how you'd catch a managed provider drifting as fast as your own setup.

Willison sits on the managed side on purpose, so it's only fair to hold it to the questions above. Changes go through Willison and you never touch the software. Willison reads every call, so nobody is relying on you to audit them. Setup is done for you, with your hours, service area and pricing built in before the first live call and the founder reviewing the account before your line goes live.

What it won't do is worth saying just as plainly. It doesn't dispatch a crew, send a truck, or promise anyone an ETA, and it can't help a calendar that's already full. If holding the keys yourself matters, if you want to change a greeting at ten at night without asking anybody, managed is the wrong shape and a DIY tool is the honest answer. More on the category in what an AI receptionist is and how it works.

And a demo settles less than it seems to, so use one for the narrow thing it's good for. Hear a real call on willisonhq.com, from first ring to booked visit, then call the live demo lines right below it and give the HVAC line the awkward version of your own call, the one with the address it has to get right.

Frequently asked questions

Nobody can hand you an hours figure that's true for your shop, and be wary of anyone who tries. Size it yourself instead: list everything about your business a caller could ask about, then count how many of those changed in the last twelve months. Prices, service area, job types, hours, who gets paged after five. That count is your annual edit list, and it's the workload nobody mentions in a demo.

Three questions, and the answers should come quickly. Who makes a change when I call, and how fast does it land? What happens on your end when a call goes wrong, and how would I hear about it? Does a human read the calls, or does the system only report on itself? Vague answers to those tell you the word is doing marketing work rather than describing anything.

Yes, and it's a reasonable way to learn what you need. What carries over isn't the configuration, it's the education: after a few months you'll know when your calls really land, which questions callers ask that you never anticipated, and where your own setup kept coming up short. Bring that list to whoever you move to. It's worth more than the settings file.

The stakes change more than the reasoning does. A water or fire call is triage, not lead capture, and the caller is standing in the middle of the emergency at two in the morning with no patience for a second-best script. An out-of-date setup that mishandles that call costs the job that night, not next quarter. Willison answers, calms the caller, triages the details and hands them to your team, and it never dispatches or promises an ETA.