Nine one star reviews land in a fortnight after one bad shipping run, and your support queue is already behind. You cannot write a considered reply to all of them. So you write none, or you paste the same paragraph nine times, and both read badly to the shopper who finds them in March.
The useful question about how to respond to negative reviews is not what tone to strike. It is operational. Which reviews get an answer this week, who writes it, and what the words are for each complaint you keep seeing.
This is the playbook I run with the stores I manage. Triage first, ownership second, wording third.
Which negative reviews do you answer first when you cannot answer them all?
Reply capacity is finite and small. Treat it as a queue with a priority order, not an obligation to answer everything.
Rank on reader impact, not on how much the review stung.
- Traffic beats recency. A two year old complaint on the page taking most of your sessions outranks yesterday's one star on a product that ships twice a quarter.
- Anything factually wrong goes near the top. A shopper cannot tell a mistaken claim from a true one, and a wrong statement about sizing or compatibility does damage every day it sits there.
- Patterns before one offs. If four reviews name the same problem, that is one reply written four ways plus a fix.
- Two and three star reviews convert better than you think. They carry one specific objection. Answer it and the review starts working for you.
- Vague and old goes last. "Not for me" on a product nobody visits does not need your morning.
The weekly routine that stops this becoming a pile:
- Filter to reviews below four stars from the last seven days.
- Sort that list by product sessions, highest first.
- Tag each by type: factual error, our fault, carrier, old version, unclear.
- Reply top down to factual error and our fault until the budget runs out.
- Send recurring complaints to whoever owns the fix, with the review text attached.
Who decides how to respond to negative reviews inside the business?
The instinct is to hand review replies to marketing, because it is public writing. Wrong hand. Marketing does not know whether the parcel left on Tuesday.
- Draft goes to whoever holds the facts. Usually support. They can see the order, the tracking, the emails and the return.
- Approval goes to one named person. Two passes, no committee. That is what stops a defensive sentence going live at 6pm on a Friday.
- The founder signs the money and quality ones. Refunds, safety, anything that reads as a broken promise. A founder name raises the cost of writing something lazy.
- Nobody replies while annoyed. No reply written in the hour the review arrived gets posted. Draft it, read it again tomorrow.
- Never the person being criticised. If the review names an interaction, that person does not write the public answer to it.
How long should a reply be before the length itself starts hurting?
Short. Under ninety words for a routine complaint, and the good ones are shorter.
Length signals defensiveness. A reply longer than the review it sits under reads as a rebuttal, and most widgets truncate it anyway, so the best sentence hides behind a "read more" nobody taps.
Four parts, in this order:
- Acknowledge the specific thing. Not the feeling, the thing. "The lid arrived cracked" beats "sorry you had a poor experience".
- One fact. The single piece of information the reader needs.
- What changed, or what you will do. Present tense, concrete, no conditionals.
- A route off the thread. A named person and a real address, one line.
No returns policy link. No paragraph about valuing feedback. The reader is scanning.
How do you reply when the customer is factually mistaken?
Correct the record without correcting the person. Different jobs, and only one of them helps.
The trap is "as clearly stated on the product page". Usually true, always reads as contempt, and the next shopper takes the reviewer's side because nobody enjoys watching a company win.
State the fact once, then hand them a way forward:
The 40mm strap does not fit the 44mm case, which is an easy thing to get wrong on our page and something we are fixing in the size chart this week. Email me at anna@ and I will swap it for the 44mm at no cost.
It corrects the claim, takes some blame for the confusion, and answers the next reader with the same question. It does not tell the reviewer to read more carefully, and it does not ask for a rating change.
If the mistake recurs, the page is the problem. Fix the page and say so.
How do you reply when the fault is plainly yours?
Name the failure in the same words the customer used. Highest return sentence in the playbook, and almost nobody writes it.
- Say the thing. "We shipped this four days late." Not "there was a delay in fulfilment."
- Never "sorry you feel". It relocates the problem into the customer's head. Readers clock it instantly.
- Drop the conditionals. "If this did not meet expectations" hedges a fact you already know.
- Say what changed, with a date. "Orders leave the same day if they land before 2pm, since 4 January." A fix with a date is credible. A promise to do better is not.
- Do the remedy before you post. If the refund is sent, say so in the past tense. Announcing an intention and then being slow is worse than silence.
One line I avoid: "this is not typical of us". It reads as a plea, and the reader is looking at a page of reviews and can judge that for themselves.
How do you reply when the complaint is about a carrier you do not control?
Take responsibility for the outcome, not blame for the cause. The customer bought from you and the parcel is your promise. Naming the carrier as the villain tells the reader that ordering from you carries a risk you have decided you cannot manage, which is a worse story than the late parcel.
What works:
- Own the outcome in one clause. "This arrived five days after it should have."
- Skip the operational explanation. Depot, sortation, weather. None of it changes the reader's decision.
- Name the control you added. Earlier cutoffs, a faster service above a certain order value, a buffer on the checkout estimate.
- Fix the estimate before the reply. If your delivery window is optimistic, every carrier review is a review of your checkout copy.
How do you reply when the review describes a version you no longer sell?
Date it in the first sentence. The reader has no way to know the review predates the change.
This was the v1 handle, which snapped on us too. The current version has a steel core and shipped from March 2026. If you still have the old one, we will replace it.
Three rules. Do not ask for removal because the product changed, because the review was accurate when it was written. Do not bury it under new positives and call that handled. Do offer the current version in private, with nothing attached to the rating.
A dated note is also the cheapest way to stop the same complaint being written again by somebody who read the old one and went looking for the fault.
At what point should the conversation move off the public thread?
Early, and always before the second exchange. A public thread is a broadcast, not a conversation, and nothing good happens on the third reply.
Move it private the moment any of these appear:
- Money. Refund amounts, partial credits, anything negotiable. Never post a number you are willing to move on.
- Personal data. Order numbers, addresses, phone numbers, health details. Do not confirm publicly that someone bought a specific thing.
- Health or safety claims. Straight to whoever handles them internally. The public reply says only that you are looking into it.
- Escalating tone. If the reviewer answers angrier, the public exchange is already lost.
The route has to be real. A named person and a monitored address beat a generic inbox, which is what the reviewer already tried. If they never take it up, fine. The reply was for the reader.
On what grounds can a review legitimately be taken down?
Narrow ones. Removal is for content that breaks the rules, not content that hurts.
- Off topic. A gripe about a different company, with no product content in it.
- Personal data or defamation. Named staff, a home address, a claim of criminality.
- Abusive or discriminatory language. Remove it and log why.
- Wrong product. Genuinely about an item they did not buy from you, common after a merge or an import.
- Duplicates. The same review posted three times by a retrying form.
Not on the list: low rating, unfair reasoning, a customer who did not read the page. Fraudulent reviews are a separate problem with a separate process.
Keep a log. Date, review ID, ground, approver. If you ever defend your corpus to a platform, the log is the defence, and it cannot be built afterwards. The guide to Shopify product reviews covers the display side.
Where does Edge Reviews fit in handling replies at volume?
Edge Reviews is our app and it is new. It has no rating on the Shopify App Store yet and I am not going to imply otherwise. It has a free plan and imports the reviews you already own, so switching does not reset your corpus.
On this job it is built for the queue: negatives across the catalogue in one list, high traffic pages first, with a draft and approve step before anything goes public.
If your priority is the deepest free tier with the longest track record behind it, Judge.me is the honest answer and the app to beat. Its free plan is permanent and covers unlimited reviews including photo and video, and it carries 5.0 stars across 43,884 reviews. If your storefront leans on photo and video proof, Loox at 4.9 stars across 9,051 reviews is built around that gallery, though its pricing scales with order volume.
| App | Rating | Free tier | Entry paid plan |
|---|---|---|---|
| Judge.me | 5.0 (43,884) | Permanent, unlimited reviews, photo and video | $15/mo |
| Loox | 4.9 (9,051) | Up to 500 orders | $49.99/mo, 300 orders included |
| Yotpo | 4.8 (4,392) | Up to 50 monthly orders | $15/mo |
| Edge Reviews | No reviews yet | Free plan, imports existing reviews | n/a |
Pricing and features checked on 22 August 2026. App Store listings change without notice, so verify on the listing before you commit to a plan.
The app is not the lever here. Triage, a named owner and ninety words are, and they work in whatever tool you already have. Once replies are running, the next constraint is volume, covered in how to get more product reviews.
Questions people ask next
How fast does a reply need to go up to still be worth writing?
Within a week is fine for most stores. The reply is not for the reviewer, who has already moved on. It is for the shopper reading the product page in three months. What actually matters is that the reply is there when they scroll, and that it is not sitting under a complaint you have since fixed without saying so.
Should I reply to every one star review or only some of them?
Only some. Reply to anything a shopper might believe and act on: a factual error, a recurring complaint, a failure you have since fixed. Skip the vague one liners on low traffic products. A reply under every negative review and none under the positives makes the negatives look like the important ones.
Does replying to a bad review make it rank higher on the page?
Not by itself. Sort order comes from your widget configuration, not from whether a reply exists. What a reply does change is how the review reads once someone reaches it. An unanswered complaint reads as true and current. The same complaint with a dated, specific answer under it reads as handled.
Can I ask the customer to update their review after I fix the problem?
You can ask once, in private, after the fix has actually landed. Do not ask in the public reply and do not attach anything of value to the request. Tying a refund, a replacement or a discount to a rating change is manipulation, and most review platforms treat it as grounds for removing the review entirely.
Who should sign the public reply if we are a small team?
A real first name from whoever holds the facts, plus the store name. Anonymous replies signed by the brand read like a form letter and invite the reader to discount them. If the complaint is about money, quality or safety, the founder should sign it. A named person raises the cost of writing something careless.
Anurag Chandra
Founder, Edgecoms
Anurag runs Edgecoms, a studio of Shopify apps. He spends most of his week inside merchant stores working out why a number is lower than it should be.
Connect on LinkedInRead this on your assistant
Opens with a summary request for this page already written.