Who is this assistant for?

Your competitor has four times your reviews and you cannot work out where they came from. Someone told you to give away a dessert for every review and it sounded reasonable. Your servers ask for reviews when they remember, and they hardly ever remember. You started asking everyone and your average score went down.

Most restaurants ask for reviews in the worst possible way: whenever a server remembers, to anyone walking past, with a QR code that lands anywhere. This assistant turns that good intention into a system of five parts that fit together —moment, selection by facts, who asks, physical bridge and post-visit message— and writes down the hard rule that saves the listing: reviews are not bought, nothing is given in exchange, and the link is hidden from nobody over what they might write. It raises the average score at a pace the platform will not flag, and tells you every Monday whether it is working.

AI assistant · N4 Executable plan

◆ Included in the annual plan⭐ Reviews, Reputation and Automated Messagingv1.2.0 · 2026-08-27

Review Request System Builder for Restaurants

Choose at the top which moment of the cycle you are in: asking for reviews, reading what they already say, or winning back whoever left upset before they write. In this one —asking— you build the full system: the exact moment it is asked, who is asked and who is not, the twelve-word line and who says it.

Depth

10/23

Your data

Compatible AIs

Steps to use it

Depth of the resulttap a level to see what you get

◀ Put out today's fireRun the year ▶

🗺️

N4Executable plan

A plan with owners, deadlines and the metric you review. Not advice: an agenda someone can execute on Monday.

◆ This assistant 23 of the 164 published assistants are at this level

Is this for you?

What it does and what you get

Most restaurants ask for reviews in the worst possible way: whenever a server remembers, to anyone walking past, with a QR code that lands anywhere.
This assistant turns that good intention into a system of five parts that fit together —moment, selection by facts, who asks, physical bridge and post-visit message— and writes down the hard rule that saves the listing: reviews are not bought, nothing is given in exchange, and the link is hidden from nobody over what they might write.
It raises the average score at a pace the platform will not flag, and tells you every Monday whether it is working.

What you end up with

Five chained deliverables: the system table moment by moment, the two selection lists with their observable signal, the twelve-word sentence with variants and a script for the no, the physical bridge map with owner and date, and the post-visit message with its weekly count sheet.

EXAMPLE This is what you getAn example built with Sazón de Origen, the house's fictional restaurant.Not your data: your result is built with YOUR restaurant's.See the full example →

Sample result

Sazón de OrigenFictional contemporary Colombian restaurant in Medellín, 2 locations, 6 years in business. Every sample result in the library comes from this same case. Fictional restaurant

The owner asked this: "I have 3.9 on Google with 61 reviews and the place across the street has 4.6 with over 400. My servers ask for reviews when they remember. I want a system for the Laureles location."

Moment	Signal that triggers it	Who asks	Destination listing	Physical bridge
Weekday lunch	The plate came back empty and the table said so	Whoever served	Google	Card on the coaster
Friday dinner	They asked for a second dessert	Whoever served	Google	Card in the check folder
Delivery order	The order arrived hot and complete	Nobody at that moment	Channel listing	Sticker on the bag

The window: 15 min from the signal, a cap of 2 tables per shift. Past 24 hours it waits for the next visit — asking late sounds like a charge.

SUPUESTO: no «industry normal» rate of reviews per 100 covers is hard-coded. The target starts against Sazón de Origen's own count over the first 4 weeks; if the platform changes pace, that is said and the target recalculated.

The list does not filter opinions, it filters FACTS. That difference decides whether the listing survives: choosing by what happened at the table is judgement; asking "how was everything?" first and showing the link only to whoever answers well is filtering the opinion, and platforms hunt that down.

Observable signal	Ask or not	What happens to that table
Plate finished and something else ordered after the main course	Yes	Ask within the next 15 minutes, folder already on the table
Third visit or more, recognised by the server	Yes	Ask, and log the server's name in the shift notebook, not the guest's
Asked about the recipe, about the chef or about the Trucha al ajillo	Yes	Ask using the chef variant
The plate came out after 25 minutes	No ask	The floor manager apologises at the table and the dessert is on the house; the QR stays in the folder, as at every table
Something sent back to the kitchen or swapped	No ask	Ask what went wrong and log the dish, which is what the kitchen can fix
First visit at the 1:00 p.m. peak with rushed service	No ask	Invite them back on a Tuesday, when the house can serve them properly
Already left a review in the last 90 days	No ask	Thank them by name and nothing more

The hard rule, written so nobody interprets it. Reviews are not bought. Nothing is given in exchange for a review: no dessert, no discount, no monthly prize draw, no "coffee is on me if you leave it right now". And the link is hidden from nobody over what you suspect they will write. What you may do is a gesture for whoever COMES BACK: the third-visit guest gets the house coffee, with no condition, with no mention of reviews and with nobody checking whether they left one. The difference matters because an incentive per review buys the opinion and makes it useless as a signal —for the guest reading it and for you, who no longer know whether the 4.8 is true— while a gesture for coming back rewards behaviour counted in visits and touches not one word of what the guest writes.

Illustrative example generated with a fictional restaurant. Not a promise of results, not a client case.

This is 3 of 5 parts. Behind the wall sit the two pieces that get executed. The physical bridge map, with columns Piece, Where it sits, Who replaces it, How often and What gets checked: the QR sticker across the twelve check folders with its owner and its day, the own-delivery card in batches of 200, the register sign, the written decision to insert nothing on the delivery platform, and the Monday test that confirms the code lands on the write-a-review box and not on the listing. And the post-visit message with its weekly sheet —Week, Asked, Left, Rate, Average score—, the 45-minute send window cut off at 9:00 p.m., the cap of one per guest every 90 days and the rule that sends you to check the bridge before the sentence whenever the rate stays under 12%. The sheet closes Monday at 10:00 a.m.

See the full example — free account

Works with these AIs

ChatGPT — the best for this assistant The best fit for this assistant: it does the asks-per-week arithmetic and writes the short sentence, without switching register in between.

How to use it

Choose the moment of the cycle at the top. Three chained acts, none replacing another: the chip takes you to the one you need today.

Open the listing where you want the reviews and write down three numbers: how many it holds today, what score it shows and how many arrived last month.

Copy the assistant into your preferred AI along with your business details and answer its questions about slots, channels and who works each shift.

Print the physical bridge, write down the name of whoever replaces it, and train the twelve-word sentence in five minutes of pre-shift.

Tape the count sheet next to the register and log asked and left per shift; close the count every Monday at the same hour.

Which of your data it uses

of 23

This assistant works with 10 of the 23 fields in «My Restaurant», among them Restaurant name, Business type and City and country. It does not ask out of curiosity: these are what make the answer speak about YOUR scale instead of an industry average — the same calculation on a business of another size returns a number you cannot decide with. You fill them once and they apply across the library, so the second assistant you open already starts with them in place. Whatever you leave blank, the assistant asks for one thing at a time instead of inventing it, and says so before answering, so you know what it is working from.

Restaurant nameBusiness typeCity and countryYour main guestStrong and slow daypartsActive sales channelsTeam sizeCommunication toneBest-selling dishesActive social channels (with the @ or the link)

Always up to date

Every assistant carries a visible version and date. When an AI changes how it works, the assistant's version goes up and you see what changed.

v1.2.02026-08-27It becomes the head of its marketing family: the chip gathers into one card what used to be two or three, without retiring any. It was done because the sisters read the same from the shelf even though they differ inside, and the owner was choosing by the title instead of by what they had in hand.

v1.1.12026-08-26The body stops contradicting the answer ladder. It used to say «do not move on until it arrives», the very order the ladder was born to replace: the assistant asked the owner for what the owner does not have and stalled. It now says what the house doctrine says — ask one at a time, and if they do not know, continue with a marked assumption. FORMAT OF THE RESULT was not touched, so the example still corresponds and its seal moves up with the card.

v1.1.02026-08-25Body rewritten to pay down duelo debt: it scored 78.8 out of 90.0, the highest in the batch and still below. `lexico` read ZERO: the system said when to ask and through which mouth without ever naming the trade where it happens —the delivery order that arrived hot, the occupancy peak where you do not ask, the closed order ticket next to which asking sounds like a toll, and the rate counted per 100 covers served instead of per server, which is the very trap this assistant already fought without using the trade's word for it—. `cifras` sat at 1 because the windows were spelled out in words: «fifteen minutes», «ninety days», «four weeks»; they are now 15 min, 90 days and 4 weeks, which is the only form an owner can meet or miss. `cuidados` and `verificacion` rose with two checks that were missing: the star-dish bias —asking only where the star dish went out leaves the listing describing a menu that does not exist, and the sales mix the reviews push shows on next month's order tickets— and the sentence surviving team turnover. A SUPUESTO line is added. The FORMAT is deliberately kept as lettered blocks so the card keeps measuring the N4 it declares. Rises to 96.7.

v1.0.02026-08-25The assistant is born.

Same method, another variant

Inside the app this is a single control: you press it and the assistant reframes itself, including the variants that don't appear here because they don't have their own page yet.

Which moment

ask for reviewsread what they already say win back whoever left upset