agentleFS
SkillMedevSkillMedev/local-service-business-engine/skills/review-engine/SKILL.md

review-engine

Skill0 starsChanged 4 months agoMITView on GitHub

Builds a systematic review generation and response engine for a local service business - the ask-at-peak-satisfaction timing rule, SMS ask scripts, a review-velocity target benchmarked to the top competitor, and word-for-word response procedures for 1-3 star reviews. Use when an owner asks "how do I get more Google reviews", "a customer left a bad review, what do I say", "should I text customers for reviews", or "why does my competitor have way more reviews". Do NOT use for the rest of the Google Business Profile setup - use google-business-profile-optimizer instead. Do NOT use for turning existing reviews into website FAQ content - use review-to-faq-builder instead.

4 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

11 rules. 0 are common, 11 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 11

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Ask at the moment satisfaction peaks: the tech has just fixed the problem, shown the customer the result, and the relief is fresh.

    line 32
  2. Distinctive:

    Use the direct review link from the profile (the short share URL).

    line 62
  3. Distinctive:

    Never argue, never share job details publicly, never blame the customer - even when the review is unfair.

    line 82
  4. Distinctive:

    If resolved, ask once, politely, whether they would consider updating the review.

    line 85
  5. Distinctive:

    Ask timing: within 1 hour of completion.

    line 136
  6. Distinctive:

    Do not buy or fabricate reviews, or review-gate (asking only pre-screened happy customers via a filter survey) - both violate policy and …

    line 153
  7. Distinctive:

    Do not offer discounts or payment for reviews.

    line 156
  8. Distinctive:

    Do not argue publicly with a reviewer, even a lying one.

    line 159
  9. Distinctive:

    Do not send more than one nudge.

    line 161
  10. Distinctive:

    Do not batch asks weekly - the peak-satisfaction window is measured in hours, not days.

    line 163
  11. Distinctive:

    Every completed job triggers exactly one ask and at most one nudge, with a named owner or automation for the send.

    line 168

“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Review Engine

Reviews are the prominence engine of local ranking and the trust engine of conversion: they feed the map pack (see local-seo-playbook) and they close the customer who is comparing three plumbers at 9 pm. The costly mistake this skill prevents is leaving reviews to chance - happy customers stay silent by default, angry ones self-select, so an unmanaged profile drifts toward a rating the business does not deserve. Run this after google-business-profile-optimizer has the profile itself in order.

Worked business throughout: Summit Plumbing, 4 techs, $450 average ticket, about 167 jobs a month at $75,000 monthly revenue.

Operating procedure

Step 1: Set the velocity target against the top competitor

Count the top map-pack competitor's total reviews and how many they gained in the last 90 days (check their newest review dates). Target: match or exceed their monthly velocity. Summit's top competitor adds about 10 reviews a month; Summit gets 3. Target: 12 a month. At 167 jobs a month, 12 reviews is a 7 percent review rate - very achievable, since a well-timed SMS ask converts around 10-20 percent of happy customers.

Step 2: Install the ask-at-peak-satisfaction rule

Ask at the moment satisfaction peaks: the tech has just fixed the problem, shown the customer the result, and the relief is fresh. For trades this is on-site at job completion or within the hour; for clinics and salons, at checkout or same evening. Every hour after peak, response rates fall. An ask three days later via email is why the old system produced 3 a month.

Step 3: Send the ask by SMS, primed by the tech

Text beats email roughly 5x for local review asks - it is opened in minutes and the review is one tap away. The tech primes it verbally at handoff, then the office (or automation) sends the SMS within the hour:

Tech, at handoff:
"Glad we could get that sorted. We're a small local shop, so Google
reviews genuinely decide whether people find us. You'll get a text from
us with a link - if you have 60 seconds, it'd mean a lot."

SMS, within 1 hour of job completion:
"Hi [FILL: first name], it's [FILL: tech name] from Summit Plumbing.
Thanks for having us out today! If you were happy with the work, would
you leave us a quick Google review? It takes about a minute and helps
our small team a lot: [FILL: review link]"

If no review after 3 days, one follow-up only:
"Hi [FILL: first name] - [FILL: tech name] from Summit Plumbing again.
No pressure at all, just a friendly nudge in case the review link got
buried: [FILL: review link]. Thanks either way!"

Use the direct review link from the profile (the short share URL). One nudge maximum; a third message is pestering and earns the opposite of a five-star.

Step 4: Respond to every review, positive ones included

Respond to five-star reviews within a week, two to three sentences, naming the service and city where natural ("Thanks, Maria - glad the water heater swap in Oak Hill went smoothly"). This is conversion copy for the next prospect reading the profile and reinforces relevance.

Step 5: Run the 1-3 star response procedure

The audience for a bad-review response is not the reviewer - it is the hundred prospects who will read the exchange later. The procedure:

  1. Respond within 24 hours. Speed signals a business that cares; a week of silence reads as guilt.
  2. Take it offline. Apologize for the experience (not necessarily fault), give a direct name and phone number, invite the conversation off the public page.
  3. Never argue, never share job details publicly, never blame the customer - even when the review is unfair. Health businesses: never confirm the person was a patient (privacy law).
  4. If resolved, ask once, politely, whether they would consider updating the review. Never make a refund conditional on editing it.

Word-for-word script:

"Hi [FILL: name], I'm [FILL: owner], the owner of Summit Plumbing. I'm
sorry your experience didn't meet the standard we aim for - that's on
us to fix. I'd like to make it right. Please call me directly at
[FILL: phone] or email [FILL: email] and I'll take care of it
personally."

Good/bad contrast on a real one-star ("Tech showed up 2 hours late and tracked mud through my hallway"):

BAD:  "We were late because of an emergency call, which we texted you
      about. Also our techs always wear boot covers, so the mud claim
      doesn't add up."
GOOD: "Hi Dan, I'm Alex, the owner. A 2-hour delay and a mess left
      behind is not how we operate, and I apologize. I'd like to make
      it right - please call me at (555) 014-2200 and I'll sort this
      out personally, including having the hallway cleaned."

The bad reply wins the argument and loses every reader. The good reply loses the argument and wins the next twenty customers.

Step 6: Track velocity monthly and feed the pack-mates

Log reviews gained per month against the target on one line of the KPI sheet. Review velocity is a ranking input local-seo-playbook depends on - if velocity is at target for 90 days and rankings still lag, the problem is relevance or citations, not prominence. And every five-star moment this engine detects is the referral trigger: hand those customers to local-referral-engine for the referral ask.

Inputs to elicit

Collect: current review count and rating, top competitor's count and 90-day velocity, jobs per month, how job completion is recorded (this triggers the SMS), who sends the ask (tech, office, or software), and the direct review link. Defaults: ask 100 percent of completed jobs, SMS within 1 hour, one nudge at day 3. If the competitor's velocity is estimated from review dates, label it an estimate.

Concrete thresholds

  • Velocity target: at or above the top competitor's monthly rate; Summit's is 12 a month.
  • Ask timing: within 1 hour of completion; response rate decays after peak.
  • Expected conversion: 10-20 percent of asked customers; below 5 percent means the ask is mistimed or the link is broken.
  • Negative reviews: respond within 24 hours, always.
  • Rating floor: a profile below 4.0 loses the comparison by default - pause paid spend into a sub-4.0 profile and fix service delivery first (check the economics with local-unit-economics).

Deliverable

A one-page review operating system: the velocity target with the competitor benchmark, the tech handoff line, the SMS ask and nudge templates with fields filled for the business, the 24-hour negative-review procedure with the owner's script, and a monthly velocity log line.

Do NOT

  • Do not buy or fabricate reviews, or review-gate (asking only pre-screened happy customers via a filter survey) - both violate policy and platforms remove the reviews, sometimes the profile.
  • Do not offer discounts or payment for reviews; incentives for reviews are prohibited, and one screenshot can undo years of reputation. Incentivize referrals instead - that is local-referral-engine's lane.
  • Do not argue publicly with a reviewer, even a lying one. The reply is for the readers.
  • Do not send more than one nudge; pestering converts five-star intent into silence or worse.
  • Do not batch asks weekly - the peak-satisfaction window is measured in hours, not days.

Quality bar

  • Every completed job triggers exactly one ask and at most one nudge, with a named owner or automation for the send.
  • The velocity target cites the actual competitor number, not a wish.
  • Negative-review scripts contain a real name and direct contact, not "the management team."
  • The monthly log distinguishes velocity (new reviews) from totals, because velocity is what ranking rewards.

Escalation and routing

Defamatory or fake reviews: flag through the profile, and consult a lawyer before responding to anything with legal exposure - this skill is not legal advice. Clinics and health practices must keep every response free of any patient-relationship confirmation. Route profile setup to google-business-profile-optimizer, ranking strategy to local-seo-playbook, the post-five-star referral ask to local-referral-engine, and whether review volume justifies its labor cost to local-unit-economics.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.