Ask right after the good moment, ask the same way every time, and ask every customer, including the ones who don't look thrilled. That last part trips up more businesses than timing or wording ever does, because Google's rules specifically ban asking only your best customers.
Most advice on this stops at "ask at the right time" and leaves out the part that actually gets businesses in trouble: who you ask, and whether you're asking everyone the same thing.
When to ask
For a service you deliver in one sitting (a haircut, a repair, a consultation), ask before the customer walks out the door. The experience is still fresh and they're already standing in front of you with a phone in hand.
For anything that ships or gets delivered, wait until the customer has actually used it. Three to five days after delivery works well for most physical products; asking on the day of checkout catches people before they've formed an opinion.
Either way, ask once at the natural moment and follow up once, a few days later, if there's been no response. A second reminder lifts response rates; a third starts to read as nagging.
Where to ask
In person beats every other channel for service businesses, because it happens at the exact second goodwill is highest. A QR code review card at the counter or on the receipt turns that moment into a scan instead of a conversation you have to remember to have.
For anything remote, email and SMS both work. Email is still the most common channel businesses use, and SMS gets read faster but needs to be shorter and more direct. Send one, not both, to the same customer for the same visit.
Whatever channel you use, the ask should take the customer to a rating screen in one or two taps. Every extra step between "here's the link" and "here's the star row" costs you a chunk of the customers who intended to leave one.
What the ask should never do
Here's the part that's easy to get wrong: Google's policy prohibits discouraging or filtering out negative reviews, and it prohibits soliciting reviews only from customers you expect to be happy. Even a phrase like "if you enjoyed your visit, leave us a review" counts as review gating, because it quietly tells everyone else they weren't invited.
The fix isn't complicated. Ask everyone the identical thing, with the identical wording, regardless of how the visit seemed to go. What can legitimately differ is where an unhappy customer's feedback ends up: a business is allowed to give every customer the same prompt and let a private feedback channel exist alongside the public one, as long as the choice belongs to the customer and not to a staff member eyeing their mood on the way out.
That's a different mechanism from gating. Gating decides in advance who gets asked. A rating-first flow asks everyone the same question and lets the customer pick the destination after they've answered, public review site or a private note, which means an unhappy customer can still tell you something, just not necessarily in public.
A script that holds up
In person: "Would you mind leaving us a quick review? Takes about ten seconds, here's the code." No mention of stars, no "if you liked it," no steering.
By text or email: "Thanks for stopping by today. If you have a minute, we'd appreciate a review: [link]." Same message to every customer on the send list, sent the same day relative to their visit.
Neither script asks for a rating, a specific word count, or a positive tone. That's deliberate: a request for a five-star review is a request for a verdict, not feedback, and it's the kind of language that shows up in FTC guidance on review manipulation as well as Google's own policy.
Get the team asking the same way
The biggest gap in most review programs isn't the wording. It's that only some staff members ask, and the ones who do ask inconsistently. Per-staff attribution fixes the second problem by making the habit visible: when a team can see whose customers actually leave reviews, asking stops being something only the outgoing staff member remembers to do.
Training takes five minutes: show the team the exact phrase, show them where the code or link lives, and tell them it goes on every customer's bill or ticket, not a curated subset. Consistency matters more than charm here. A flat, identical ask beats a warm one that only goes to the customers a staff member liked.
Measuring whether it's working
Track two numbers, not one. Response rate, the share of customers you asked who actually left a review, tells you whether the ask itself is working. Star distribution tells you whether you're hearing from everyone or just the easy cases; if a location's public reviews are suspiciously uniform, that's usually a sign staff are quietly filtering who gets the ask, not that every customer loved the visit.
If you're running this across a few counters or staff, keep the ask itself boring and identical everywhere, and let a private feedback option absorb the complaints that would otherwise never reach you and never show up publicly either. That combination, the same public ask for everyone plus a private route for what wouldn't help to post, is the idea behind how Starboard's rating flow is built. It's a lot easier to run once the code is on every counter instead of only the ones near your best-performing staff.