How to Ask for Reviews Without Being Annoying

The question of how to ask for reviews without being annoying is nearly always asked about wording, and wording is not where the problem lives. Teams rewrite the subject line, soften the opening, add a line of thanks, and the response rate does not move. It does not move because the customer is not reacting to that email. They are reacting to it being the fifth automated message from you this fortnight. Last updated: September 2026.

Omniconvert has measured how post-purchase programmes are built across the CROBenchmark dataset of 7,000+ websites in 15+ industries, against 248+ audit criteria, over 13 years in eCommerce [CROBenchmark Report 2026, Omniconvert]. In most programmes the review request is designed by one team, in isolation, against its own response rate, while nobody owns the total volume of automated contact a single customer receives.

This article covers the part the neighbouring guides leave alone. The survey of all three levers is in get more product reviews, ownership and routing sit in how to build a Review-Generation system, and the rules around rewards are in how to incentivise reviews without breaking platform rules. What follows is frequency, channel and consent.

What actually makes an ask for reviews feel annoying

Ask someone why they unsubscribed from a brand and they rarely name a single email. They describe a feeling of being contacted too much, and then they name whichever message happened to be on screen when they reached that conclusion. Review requests are disproportionately that message, for a structural reason: they usually sit at the end of the post-purchase sequence, after the confirmation, the dispatch note, the delivery note, the how-is-it-going note and the cross-sell.

The second factor is proportion. Every request has a size, measured in what it asks the customer to do, and every channel has a tolerance for that size. An email asking for a paragraph of writing is proportionate. A text message asking for a paragraph of writing is not, because the channel implies brevity and urgency and the request implies neither.

The annoyance budget, and how stores overspend it

The annoyance budget is defined as the finite tolerance each customer holds for unrequested contact. The model is simple enough to hold in your head. A customer buys, and from that moment you have a small allowance of their attention. Each message spends some. A dispatch notification spends little, because it gives information they wanted. A newsletter spends more. A request for a favour spends most, because it transfers work to them.

There is a reason to care beyond politeness. Bain and Company's work with Fred Reichheld holds that a five percent improvement in retention can raise profits by 25 to 95 percent [Harvard Business Review, 2014], and the contact programme is one of the few levers that touches retention directly in both directions. The same sequence that brings a customer back can, at a slightly higher volume, be the reason they stop opening anything.

Channel decides how much you are allowed to ask

SMS is different in kind. It arrives where personal messages arrive, it is usually read within minutes, and it cannot be ignored without a small act of attention. A text message is the right vehicle for a request that resolves in one tap, such as a single rating, and the wrong vehicle for anything that requires the customer to compose a sentence.

Then there are the channels most stores underuse, which are the ones where the customer is already engaged. A prompt on the order-status page. A printed card in the parcel. A request inside an app session the customer opened themselves. These spend almost nothing from the budget, because the customer arrived voluntarily and the request does not intrude on anything. Nielsen's long-running work on advertising trust has consistently found that recommendations from other people outrank brand-controlled messages [Nielsen, 2012], which is the argument for collecting reviews diligently and for collecting them where the asking is cheapest.

The five rules for a request that gets answered

1. Count every message, not just the review ones. Build the actual list for one order across every system that sends. Until that list exists, any decision about review frequency is being made without the relevant information.
2. Send one ask and at most one follow-up. Two is the ceiling. Make the second ask a different, smaller question rather than the first one repeated, because a customer who ignored an open invitation will often answer one specific thing.
3. Match the channel to the size of the request. Open-ended written feedback belongs in email. One-tap ratings can go to SMS where consent exists. Push notifications should not be asking for favours at all.
4. Ask for the smallest useful thing first. A single rating, answerable in one action, with the invitation to add detail appearing only after the customer has already engaged. Asking for the essay up front is what produces silence rather than short reviews.
5. Suppress anyone who has already reviewed. Recently, on any channel, for any product. This is a suppression rule rather than a campaign setting, and its absence is the most common defect in review programmes that otherwise look well built.

What each channel actually costs you

What to change this week

• Count the fortnight. One order, every automated message, every system. It takes an hour and it usually settles the argument about frequency on its own.
• Cut the sequence to two. If a third request exists, remove it and watch the unsubscribe rate rather than the response rate. The response you lose is smaller than teams expect.
• Write the suppression rule. Anyone who reviewed in the last ninety days receives no further request, on any channel. Check it works by looking for a customer who has.
• Put a rating prompt on the order-status page. The traffic is already there and voluntary, which makes it the cheapest collection you will ever build.

Where the harder problem is that review, support and order data sit in three systems that never meet, Omniconvert Commander (formerly Nexus) is an AI for eCommerce growth engine that unifies commerce data, segments customers by behaviour and value, and ranks the next actions worth taking, with a human approving what goes live. Suppression rules of the kind above are simple to state and hard to enforce precisely because that data is scattered.

FAQ: asking for reviews without being annoying

How many times should I ask a customer for a review?

Once, plus at most one follow-up, and the follow-up should ask a different and smaller question rather than repeat the first. A third request adds very little response and reliably adds unsubscribes. If two well-timed asks produce nothing, the customer has answered you by not answering, and the correct response is to stop rather than to escalate.

Which channel is best for a review request?

Email for anything open-ended, because it is the channel people expect commercial contact on and can ignore without friction. SMS only for a request that resolves in a single tap, and only where the customer has explicitly agreed to it. Push notifications are the wrong channel for asking a favour: they interrupt rather than wait, and the cost of the interruption is paid by every future notification you send.

Does asking for a review annoy customers enough to lose them?

One well-timed request rarely costs you a customer. The damage is cumulative rather than singular: a review request landing on top of a shipping update, a survey, a cross-sell and a newsletter is the message that tips a tolerable volume into an intolerable one. The review request then gets blamed for a problem the whole programme created.

Should I offer an incentive so the ask feels worth answering?

An incentive changes who answers rather than how the ask feels, and it carries platform rules that vary by marketplace. Treat it as a separate decision with its own constraints, not as a way to buy back goodwill you spent by sending too many messages. Fix the frequency first, because an incentive attached to an irritating sequence still reads as irritating.

How do I know if my review requests are annoying people?

Read the unsubscribe rate on the review email against your other post-purchase messages, and read what people write when they unsubscribe. Then count the total number of automated messages one customer receives in the fortnight after an order. Most teams have never counted, and the number is usually higher than anyone in the room would have guessed.

Teams that notice annoyance usually rewrite the review email. They soften the copy, add a thank-you line and test subject lines, while the sequence around it stays the same. The customer still sees the request as the fifth automated message of the fortnight, so the response does not move and the team concludes that its audience does not review.

AliveCor used Omniconvert to run a structured A/B testing programme and achieved +21% conversion rate, +5% revenue per visitor, and 94% statistical relevance across their experiments. [Omniconvert, AliveCor case study] Test the two-ask cap the same way: run it against your current sequence and compare unsubscribes as well as responses.