Building a Review Generation Engine

Lesson 1: Building a Review Generation Engine

Learning objectives

By the end of this lesson you'll be able to select the right review request channels for your business, write request messages that actually get answered, time requests so they land while the experience is still fresh, and turn review generation from a campaign into a permanent operating habit.

Why systematic beats sporadic

Most Indian businesses generate reviews in bursts. A campaign runs for two or three weeks, produces twenty reviews, then stops. Six months later someone remembers and runs another burst. In between, review velocity is close to zero.

This pattern has two problems. Google's systems read a spike-then-silence pattern as slightly unusual — a lot of reviews in a short window followed by nothing looks different from organic accumulation, even when every review is genuine. And more practically, velocity is what review velocity actually measures — not the total count sitting on your profile, but whether new reviews keep arriving. A profile with 400 Google reviews and none in the last two months reads as stale. A profile with 120 reviews and fifteen new ones every month reads as active.

A review generation engine is the opposite of a campaign. It's a process wired into how the business already operates — every appointment, every meal served, every checkout — so that requests go out automatically and reviews arrive every month without anyone deciding to "do a push." This is the second lesson in Angryturtle's learning centre reviews track; if you haven't yet, start with what local SEO is and how Google ranks local businesses for the fundamentals this lesson builds on.

The mechanism: why WhatsApp outperforms everything else in India

The reason channel choice matters this much comes down to a simple funnel: open rate × completion rate = reviews per hundred requests sent.

WhatsApp clears 60–80% open rates in India and converts 20–35% of opens into a completed review — by far the best combination available. It works because it's where Indian customers already are, the message arrives with a name attached (not a faceless "no-reply@"), and the review link opens directly in the browser with no app-switching.

QR codes sit at the other end: 3–8% response, but zero marginal cost once printed. A QR code on a clinic's reception counter or a restaurant's bill folder captures customers who never gave a phone number.

Email in India runs 10–20% open, 3–8% completion — useful as a supplement for hotels and e-commerce businesses with strong email habits, weak as a primary channel.

Verbal requests cost nothing and convert almost nothing on their own, but stacked with a same-day WhatsApp follow-up they lift completion meaningfully — the customer has already said yes once, in person, before the link arrives.

A worked example: two businesses, two engines

A dermatology clinic in Koramangala, Bengaluru, sees roughly 40 patients a day. Its receptionist's end-of-day patient list feeds a WhatsApp template sent that evening: "Hope your visit went well today — a quick Google review would help other patients find us." At a 25% completion rate on 40 daily contacts, that's roughly 10 new reviews a day if every patient consented to WhatsApp contact — in practice closer to 4–6 once no-shows and opt-outs are subtracted, still a strong monthly velocity with zero campaign effort. This maps directly to the healthcare industry profile, where review velocity is one of the most competitive local signals.

A CA coaching institute in Kothrud, Pune, has a completely different rhythm: no daily transaction, but a sharp emotional peak when exam results are declared. Its engine triggers once, within 48 hours of result day, to every student in that batch — "Congratulations on your result. If you'd like to share your experience at the institute, a review would help other students deciding where to study." One trigger, one batch, often 15–25 reviews in a single week. The clinic's engine is continuous; the institute's is event-triggered, closer to what an education business needs. Both are still engines, not campaigns, because both run without anyone having to remember to start them.

Message templates and the direct review link

Every request should use the direct Google review link, not a general invitation to "search us on Google." Generate it from the GBP dashboard under "Ask for reviews," and store it somewhere every team member sending requests can copy it from. A direct link removes the step where the customer has to find the right listing among several similarly named competitors — a step where a meaningful share of intent gets lost.

Healthcare, post-appointment: "Hi [Name], hope your appointment went smoothly today. If you have 2 minutes, we'd appreciate a Google review — it helps other patients find us: [link]." Restaurant, same day: "Hi [Name], great having you with us today. If you enjoyed your meal, a quick Google review would mean a lot: [link]." Hotel, 24–48 hours post-checkout: "Hi [Name], thank you for staying with us. We'd love your feedback on Google: [link]. Hope to welcome you back soon."

Keep these under three sentences. Longer requests get skimmed and abandoned before the link is even tapped.

Timing: send while the emotion is still fresh

Sending while the customer is still on the premises reads as pushy. Waiting three weeks means the specific, positive detail that makes a review useful has faded from memory, and the request itself has to compete with everything else that's happened since.

What practitioners get wrong

The most common failure isn't picking the wrong channel — it's treating the engine as a one-time setup instead of an owned, monitored process. A WhatsApp flow gets built, runs for two months while someone's paying attention, and then quietly stops when that person moves teams or gets busy. Nobody notices for a quarter because nobody's tracking monthly review count against a target.

The second common mistake is requesting from every customer indiscriminately, including ones who had a visibly bad experience. This isn't a legal or policy problem — genuinely soliciting reviews from everyone is fine — but it does mean a frustrated customer's negative review sometimes arrives because you prompted them. That's not a reason to filter who gets asked (selective asking is closer to the review-gating line Google actually does police); it's a reason to make sure your response process for negative reviews is solid before you scale up request volume.

The third mistake is generic templates with no personalisation slot. A message that reads identically to every recipient, with only the name swapped, gets a lower completion rate than one that references what actually happened — a specific dish, a specific doctor, a specific service.

Making it permanent

An engine has three parts running without a stated "off switch": a trigger tied to an operational event (appointment completion, checkout, result declaration), a channel that fires automatically once triggered, and a monthly check — reviews received this month, against target, with a look at why if it's below.

That monthly check is the piece most businesses skip, and it's the one that catches the engine quietly dying. It takes five minutes: open GBP, count new reviews, compare to last month. For businesses running review generation alongside broader reputation work across other platforms, the same monthly rhythm covers both, and feeds directly into the local SEO measurement work covered later in this track. Angryturtle's review management product tracks this velocity automatically and drafts responses, but the discipline described above works whether or not you use a platform for it.

FAQ

How many reviews does a business need before AI systems start citing it? There's no published threshold from Google, OpenAI, or any AI provider naming an exact review count. Review signal behaves as a gradient — more reviews and stronger recent velocity generally correlate with stronger citation likelihood — not as a gate with a fixed number on the other side. Treat velocity as the thing to keep improving rather than chasing a specific count. See reviews and AI search visibility for the full picture.

Is it against Google's policy to ask every customer for a review? No. Asking every customer, regardless of how their experience went, is standard practice and within policy. What's against policy is filtering by rating before it posts — asking happy customers to post publicly while diverting unhappy ones to a private form. That's review gating, covered in the reputation management glossary entry.

Should review requests go out from a personal number or a business number? A dedicated business WhatsApp number is worth setting up once volume passes a handful of requests a day — it keeps the request log separate from personal messages and lets more than one staff member manage the flow. Below that volume, a personal number tied to the front-desk role works fine.

What if a customer doesn't respond to the WhatsApp request at all? One follow-up after 24–48 hours is reasonable. Beyond that, move on — repeated requests read as nagging and can produce the opposite of the intended review. Move to the next trigger rather than re-prompting the same customer.

Next lesson: Responding to reviews at scale →