There's a salesperson working for your business right now. They're on the clock at 11pm on a Sunday. They talk to every single prospect before you ever get the chance to. They decide who calls you and who calls the guy across town.
That salesperson is your reviews. And almost nobody trains them.
I've watched owners drop five figures a month on ads to send traffic to a profile with 14 reviews, a 4.2 average, and the most recent one from eleven months ago — while their competitor sits at 380 reviews with a fresh one from yesterday. Then they call me and tell me their ads don't work. Their ads work fine. Their reputation is losing the sale before the ad ever gets credit for it.
Nobody chooses you. They eliminate everyone else.
Here's what actually happens when someone needs what you sell. They don't research you. They open a map, see five options, and start deleting. Low count — gone. Old reviews — gone. One angry review with no reply — gone. Whatever survives that thirty-second cut gets the phone call.
Buying is elimination, not selection. Your job isn't to be perfect. Your job is to be the last one standing after the cut.
And this got bigger this year, not smaller. AI assistants are now the layer between your customer and the phone call, and they read the same signals — volume, recency, rating, whether the owner responds. When an AI gets asked "who's the best in St. George for this," it's summarizing your reputation and handing back one name. I wrote about that shift in depth over at Rooster AI Agents, but the short version: your reviews are no longer just social proof. They're training data.
I learned this the hard way
I've lost everything twice. Not "had a rough quarter" — lost it. Both times the rebuild came down to one question: who trusts me enough to say yes today? The first time around I had nothing to point to. No proof. Just me talking about how good I was, which is the least persuasive thing on earth.
The second rebuild I did it differently. I stopped talking and started collecting evidence. Every result, every happy customer, every "you saved us" text — captured, public, dated. Within a few months I wasn't selling anymore. Other people were selling for me, in writing, on a page my prospects were already reading.
You can't argue someone into trusting you. You can only show them a pile of other people who already did.
That's the whole game. And most owners are sitting on that pile and never picked it up.
The reputation system I run
This isn't a "remember to ask for reviews" pep talk. That's what everyone's already failing at. It's a system, and it has four parts.
1. Ask at the peak, not at the end
There is a moment — the job's done, the result landed, they're genuinely grateful — and it lasts about an hour. That's your window. Ask inside it, by text, with a one-tap link. No "check your email for a survey." No QR code on an invoice they'll never look at. Text, link, tap, done.
The businesses I work with go from four reviews a month to thirty-plus on this one change alone. Same customers. Same service. Different timing.
2. Make it automatic, not heroic
If asking depends on someone remembering, it stops the first busy week. Every job that closes should trigger the ask on its own, without a human deciding. When it runs automatically, review count stops being a mood and becomes a metric that only goes up. That's the same principle behind everything I build — the win isn't effort, it's the machine that doesn't need effort.
3. Reply to every single one, especially the bad ones
A negative review with no response reads as guilt. The same review with a calm, human, next-day reply reads as a business run by an adult. I'll take a 4.7 with thoughtful owner replies over a silent 5.0 every day of the week, and so will your buyer.
Rule I don't break: respond within 24 hours, never defensive, never a copy-paste template. Name the issue, own your part, state what you did. You're not writing to the reviewer. You're writing to the next 200 people who read it.
4. Put the proof where the decision happens
Reviews stuck on one profile are worth a fraction of what they could be. Pull them onto your homepage, into your quotes, into your follow-up texts, into your ads. When we rebuilt the presence for The Bridge Health Recovery Center, surfacing real outcomes and real voices where families were actually making the decision changed the conversation completely — because families choosing care aren't comparing features. They're looking for someone they can trust with a person they love.
The 90-day version
If you want the compressed plan, here it is.
- Week 1: Count reviews for your top three competitors. That number is your target, not some arbitrary goal.
- Week 1: Reply to every unanswered review you have, oldest to newest. Yes, all of them.
- Week 2: Build the automatic ask. Job closes, text fires, one tap.
- Weeks 3-12: Never let a week pass with zero new reviews. Recency is a ranking signal and a trust signal at the same time.
- Week 12: Put your best proof on the pages and messages where money changes hands.
Ninety days of that and you are no longer one of five options getting eliminated. You're the one that survives the cut — in the map pack, in the AI answer, and in the head of the person holding the phone.
The uncomfortable part
Here's what nobody wants to hear: if your reviews are thin, it's not because your customers were unhappy. It's because you never asked. You did great work and let it evaporate. That's not a marketing problem, that's leaving money on the table on purpose.
Fix the asking. The rest takes care of itself.
If you want this built for you instead of by you — the automatic ask, the reply system, the proof surfaced everywhere it matters — that's exactly what we do at Rooster AI Agents. Either way, go reply to your reviews today. Your best salesperson has been talking without a script long enough.
Frequently asked questions
How many Google reviews does a local business actually need?
Enough to beat the top three competitors in your city, not a round number. In most local markets that means passing the leader's review count within 90 days and staying above a 4.7 average. Volume and recency matter more than a perfect 5.0 — buyers trust a 4.8 with 300 recent reviews over a 5.0 with 11 from two years ago.
What is the best time to ask a customer for a review?
Inside the first hour after the moment they were happiest — job completed, results delivered, problem solved. Ask by text, with a direct link, one tap. Wait until tomorrow and your response rate falls off a cliff.
Should I respond to negative reviews?
Always, within 24 hours, calmly and publicly. The negative review isn't written for the person who left it — it's read by the next 200 buyers deciding whether you're safe to hire. A steady, human reply turns your worst review into proof.
Andrew Smart ($MART) is the founder of Rooster AI Agents and CMO of The Bridge Health Recovery Center. Built it, lost it twice, rebuilt it with systems.