Everything else about review requests is about people. This one is about plumbing, and it is worth separating, because a venue can fix its scripts, designate its askers and read reviews at pre-shift, and still collect almost nothing because the link lands in the wrong place.
The failure is silent. A guest who scans, waits, gets a map of your neighbourhood, and gives up does not tell you. There is no metric for it.
1. The link goes to your profile instead of the review box
The most common and most costly fault. A link to your Google listing puts the guest on a page with photos, opening hours and directions, where writing a review requires scrolling and finding the right button. Most people do not.
Google generates a direct link for exactly this purpose — in your Business Profile, under the option to ask for reviews. That link opens the review dialog with the stars ready. Use that one and nothing else.
Test it on a phone that has never been to your restaurant. Not yours; yours has history that makes everything work.
2. It requires signing in, and the guest is not
A review needs a Google account. Roughly a fifth of guests on a given evening will hit a sign-in screen, and that is where the attempt ends.
You cannot fix this, but you can stop wasting the interaction: an ask made earlier in the visit gives the guest time to deal with it. An ask made while they are standing up to leave does not survive a sign-in prompt.
3. The code was generated by a service that has since expired
Free QR generators frequently produce redirect links rather than direct ones, and the redirect is hosted by the generator. When the free tier ends, the subscription lapses, or the company folds, every printed card in the building points at a dead page or an advert.
The tell is a QR code that encodes a short domain you do not recognise. Decode one of your existing cards with a phone and see what URL comes back. If it is not a Google address, reprint.
Use a generator that produces a static code encoding the destination URL directly. There is no ongoing service to fail.
4. It is printed too small, or with the wrong contrast
Practical minimum for a table card is about 2cm square, and 2.5cm is safer for older phones and dim rooms. Below that, low light plus a moving hand plus a scratched lens produces failures.
Two other print faults, both common: insufficient quiet zone — the code needs clear white space around it, and cards where the border crowds it scan poorly — and inverted or low-contrast colour. A brand-coloured code on a dark card looks considerably better than it scans. Dark code on light background, always.
Laminate them. A card that has been on a table for six weeks is greasy, and grease scatters light.
5. It is in a place nobody is sitting
Table talkers get moved, stacked and used as coasters. Window stickers are read by people walking past who have not eaten. The two placements that consistently work are the bill presenter and the receipt, because both arrive in the guest's hands at the one moment the ask makes sense.
If you print one thing, print it on the receipt.
6. There is no reason next to it
A bare QR code with "Review us" gets scanned by people who were going to review you anyway. A code with a sentence attached does noticeably better, and the sentence should say who benefits and be honest about it: "We're a small independent — reviews genuinely help."
Nothing about five stars, nothing conditional on having enjoyed it. That phrasing is review gating and it is the one thing on a printed card that could actually cause you trouble.
7. Multi-site venues sharing one code
Every location needs its own link, pointing at its own listing. Shared artwork across sites is a standard way to route six restaurants' reviews to one profile, where they are useless for comparing locations and actively misleading for guests.
Check by scanning a card from each site. This takes ten minutes and has caught more than one group with two sites' cards swapped after a refit.
The test worth running before the next print run
Borrow a phone that is not yours and has never used your venue's wifi. Scan every printed code in the building. For each one, note whether it reaches a review box with the stars showing, in one tap, without a redirect.
Anything that does not is discarded. Then order the reprint.
It is a twenty-minute job that most venues have never done, and the reason is that the person who ordered the cards tested them on their own phone, where being signed in and having the app installed makes every version work. That is the single most useful sentence in this piece: your phone is the worst possible device to test this on.
Once the plumbing is right, the constraint moves back to people — whether the ask keeps happening, which is a separate and more stubborn problem.