Review Automation
Your Project Manager Is Not a Review System: The Walkthrough Ask vs. Automated Review Requests in Restoration
Most restoration companies rely on the PM asking for a review at the final walkthrough, and most of those asks never convert. Here is how to measure your own ask-to-review ratio, an honest look at when the manual ask is enough, and what an automated request has to do to beat it.
by Jerrod Anthraper
Day 19 on a finished-basement water loss. The adjuster signed off last week, the drywall texture matches, the carpet pad is new, and the homeowner is walking the space with your project manager. She says what most of them say at this point: "You guys were great through all of this." Your PM says what most PMs say: "Glad to hear it — would you mind leaving us a Google review?" She says of course. That exchange is the entire review system at most restoration companies. And most of those "of course"s never turn into a review.
The leak: forty jobs a month, twenty reviews total
Pull up your Google Business Profile and count. A restoration company running 30 to 60 jobs a month, year after year, routinely sits at 20 or 30 reviews total. The work was good. The homeowners were grateful. The reviews just didn't happen.
That gap has a price. Each additional star in average rating correlates with roughly 12% higher conversion. In restoration, that conversion happens at 11pm, with two inches of water in the basement, on a phone screen, comparing four companies the homeowner has never heard of. The one with 140 reviews at 4.8 gets the call. The one with 22 reviews at 4.6 — with the better crews and the faster dry-out — gets skipped. Nobody is calling around to compare drying logs. They're comparing the number next to the stars.
Why the walkthrough ask fails
It isn't laziness, and it isn't that your PMs don't care. Look at what's on the PM's list at the final walkthrough: punch-list items, the certificate of satisfaction, the final invoice, the supplement that still needs to be reconciled with the carrier, the equipment count, and getting to the next job. The review ask is item nine on a list where items one through eight have money attached to them.
The ask itself is also structurally weak. It's verbal, so there's no link. It happens at the walkthrough, so the homeowner has to remember it later, on her own, after a three-week ordeal she'd like to stop thinking about. And it depends on which PM ran the job. Your best PM asks every time. Your newest PM forgets half the time. Nobody knows which is which because nothing is measured.
Contrast that with what happens when a job goes badly. Nobody has to ask that homeowner for a review. She finds the button herself.
The comparison, run on your own numbers
Before you buy anything or change anything, run this. It takes about an hour and it settles the question of whether the PM-owned ask is working for you.
1. Compute your ask-to-review ratio
Pull the last 90 days of closed residential jobs. Count them. Now count Google reviews received in that same window. Divide. A company with 120 closed jobs and 9 new reviews is converting 7.5% of completed jobs into reviews. Write that number down — it's the baseline every option gets measured against. If you're above 25%, the PM-owned ask is working and you can stop reading. Most companies are under 10%.
2. Map who owns the ask on each job
For each of those 90 days of jobs, mark who asked and when: PM verbally at walkthrough, office by email with the final invoice, nobody, or unknown. If "unknown" is your biggest category, that's the finding. A system nobody can describe isn't a system. You'll usually see one PM responsible for most of the reviews you did get.
3. Run a two-week text test
Split the next two weeks of closing jobs in half. For half, keep doing what you do. For the other half, the PM texts the homeowner a direct review link from his own phone within ten minutes of leaving the walkthrough, with one sentence that names the job: "Thanks again for trusting us with the basement — if you have 60 seconds, here's the link." Compare the two halves. Companies that run this test usually see the texted half convert three to five times better, because the link removes the memory step. That tells you the ask is the problem, not the customers.
4. Decide honestly which model fits your volume
Here's where the comparison should be fair to the manual approach. A PM-owned ask, tightened up with the text-a-link habit from step three, is a reasonable system if you're an owner-operator running under 15 jobs a month with one or two PMs you see every day. You can enforce it at the Friday meeting. The cost is zero and you'll get to 15-20% conversion.
It stops being reasonable when you cross roughly 20 jobs a month, run three or more PMs, or carry a heavy insurance mix where the PM's walkthrough attention belongs on the supplement, not the review. At that point the ask depends on memory across too many people and too many job types, and the variance between your best PM and your newest one becomes the whole story.
5. If you automate, hold it to three requirements
Any automated review request for restoration has to do three specific things, or it's just a scheduled email. It has to trigger off job status in your job management system, not off a calendar, so the request lands the day the job closes rather than on some batch date. It has to name the specific job and the specific PM, because "How did we do?" from a company name gets deleted and "How did the basement turn out, Karen? — Mike" gets answered. And it has to handle the reply: a homeowner who answers "honestly, the baseboards still look off" needs a human within the hour, not a second reminder to leave five stars.
The proof
Going from a 3.5-star to a 4.5-star average rating can lift conversion by up to 270%. Roughly 92% of consumers read reviews before buying, and about 88% trust them as much as a personal recommendation. For a restoration company, that "personal recommendation" is being made at midnight to someone who has never heard your name and is deciding in under a minute which of four companies to call. Review count and rating are the only referral that homeowner has access to at that hour. Run the 90-day count from step one and you'll know exactly how many of those referrals you've been leaving on the table.
Where Tykon fits
James is the AI sales agent Tykon builds for service businesses, and the review request is one of the things it runs on its own once a job is marked complete. It sends the request by text, names the job and the PM, follows up once if there's no response, and routes any unhappy reply to a person on your team before it becomes a public review. It isn't a build from scratch — it's a tested framework, already run through weeks of rigorous testing, so you skip the tuning phase. The same system answers your after-hours calls and asks for referrals after the job, which is why we describe it as a flywheel rather than a funnel: the review feeds the next 11pm search, which feeds the next job.
The low-pressure next step: run the 90-day ratio from step one. If it's under one review per five closed jobs, that's worth a 20-minute conversation. If it's over 25%, keep doing what you're doing and text us the number — we'd like to know what's working.