Customer Voice
In plain English
Your analytics can tell you people didn't come back. It can't tell you why.
That "why" is sitting in your negative reviews, written out in plain English by the people who experienced it. The problem is that reading a few hundred one- and two-star reviews is grim, time-consuming work, and it's easy to come away with an impression rather than a finding.
This tool reads them for you and sorts the complaints into six buckets:
| Theme | What people are saying |
|---|---|
| Sizing & fit | It didn't fit, it runs small, it runs large |
| Quality & durability | It broke, it's flimsy, it fell apart |
| Shipping & delivery | It was late, it never arrived, still waiting |
| Price & value | Too expensive, not worth it |
| Not as described | It looks different, misleading, not what I expected |
| Usage & setup | Confusing, hard to use, no instructions |
For each one it tells you how many people said it, what share of your complaints it accounts for, an example in a customer's own words, and — the useful part — what to do about it.
It only reads reviews rated 3 stars or below. Those are already complaints, so there's no ambiguity about whether "small" means a problem.
How to use it
- Click "Customer Voice" in the left-hand menu.
- Look at the themes, ordered by how often they come up.
- Read the "share" column. If sizing is 40% of your complaints, that's your biggest single fixable problem.
- Read the sample quote. It tells you what the theme actually means for your products.
- Do the suggested fix. Each theme has one — a size guide, clearer delivery expectations, better photos.
- Come back in a month. If the fix worked, that theme's mentions will fall.
⏱ ~10 min/month · 💳 Pro+ · 🎯 Find out why people don't come back
Why this matters for your business
There's a category of problem that costs a lot of money and is almost impossible to see in your numbers.
Say 40% of your complaints are about sizing. Every one of those customers ordered, waited, opened the box, and found it didn't fit. Some returned it — that's a cost. Some kept it, unhappy — that's a customer who won't order again. And behind each one who complained are many more who quietly had the same experience and said nothing.
None of that shows up as "sizing problem" in your reporting. It shows up as a return rate, a low repeat rate, and a conversion rate that isn't quite what you'd like. Three symptoms, one cause, and the cause is invisible unless someone reads the reviews.
The fix, in that example, is a size guide and clearer fit copy. That's an afternoon's work. The reason it doesn't get done isn't difficulty — it's that nobody knew it was the biggest problem.
The suggested fix is what makes this actionable. "Customers are unhappy about sizing" is an observation. "Add a size guide and clarify fit in the product copy" is a task someone can pick up on Monday.
What this typically unlocks
| What you get | Typical result |
|---|---|
| Reading hundreds of negative reviews | ~10 minutes instead of a lost day |
| The single biggest complaint, quantified | Named and measured |
| Returns after fixing the top theme | Often down 20–40% on affected products |
| Repeat rate | Improves once the cause is removed |
| Cost to run | None — no AI charges |
What you actually get
The six themes and their fixes
| Theme | Suggested fix |
|---|---|
| Sizing & fit | Add a size guide and clarify fit in the product copy |
| Quality & durability | Address durability in copy and review supplier quality |
| Shipping & delivery | Set clear delivery expectations on the product page |
| Price & value | Add value-justification copy or a bundle offer |
| Not as described | Align imagery and description with the real product |
| Usage & setup | Add usage guidance or an FAQ to the product page |
What each theme shows you
| Column | What it tells you |
|---|---|
| Mentions | How many negative reviews raised it |
| Share | What percentage of your complaints it accounts for |
| Sample | A real customer quote, so you can see what it means |
| Suggested action | The specific merchandising fix |
Why only negative reviews
Only reviews rated 3 stars or below are scanned.
This is what makes the results reliable without needing AI. In a one-star review, the word "small" almost certainly means a complaint about sizing. In a five-star review, "small" might be praise — "lovely small details." Scanning everything would need a model to judge intent, would cost money to run, and would be less predictable.
By only reading text that's already a complaint, simple keyword matching becomes accurate. It's cheap enough to run across thousands of reviews, and it behaves the same way every time.
Two ways to read the numbers
Mentions is the raw count. Share is the percentage of your complaints.
They answer different questions:
| Situation | Mentions | Share | What it means |
|---|---|---|---|
| Large shop, many reviews | 60 | 12% | Real, but one of several issues |
| Small shop, few reviews | 9 | 45% | Nearly half your complaints — bigger problem |
Share is usually the better guide for deciding what to fix first. A small shop with 9 sizing complaints out of 20 has a worse sizing problem than a large one with 60 out of 500.
How it reaches your to-do list
Every theme is passed into your Daily Ops worklist, where it competes with everything else for attention.
| Mentions | How it appears |
|---|---|
| 1–7 | Info — worth knowing |
| 8 or more | Warning — needs attention |
Themes carry a moderate confidence level, reflecting that keyword matching is a good indicator rather than a certainty. Read the sample quote before acting.
Themes clear themselves
If a theme stops appearing in new reviews, it clears from your worklist automatically after a few weeks.
So the list always reflects what customers are complaining about now. Fix your sizing problem and the sizing theme fades without you closing anything — and if it comes back, that's a real signal that something regressed.
One current limitation
Themes are store-wide, not per product. You'll see "sizing is 40% of complaints," not "sizing is a problem on these four products."
How to work around it: use the sample quote to identify which products are involved, then check those products' reviews directly. In practice the concentration is usually obvious — sizing complaints tend to cluster on a particular range or supplier.
How it works (without the technical bits)
Real merchant scenarios
Scenario A — 40% of complaints, one afternoon's fix
Setup. Apparel brand, return rate 24%, repeat rate 12%. Both poor and neither explained by their analytics.
Customer Voice, across 1,400 reviews (310 negative):
| Theme | Mentions | Share |
|---|---|---|
| Sizing & fit | 124 | 40% |
| Quality & durability | 61 | 20% |
| Shipping & delivery | 48 | 15% |
| Not as described | 44 | 14% |
| Price & value | 21 | 7% |
| Usage & setup | 12 | 4% |
Sample quote:
"Ordered my usual medium and it was more like a small. Beautiful
top but I had to send it back."
Action. Added a proper size guide with real measurements, a "runs small — consider sizing up" note on the affected ranges, and model height and size on every product photo.
Result after four months:
| Before | After | |
|---|---|---|
| Return rate | 24% | 14% |
| Sizing mentions | 124 | 31 |
| Repeat rate | 12% | 19% |
The return-rate drop alone was worth about £84,000 a year. The work took one person about two days.
Scenario B — Share mattering more than count
Setup. Small homeware shop, 180 reviews, 22 negative.
| Theme | Mentions | Share |
|---|---|---|
| Shipping & delivery | 10 | 45% |
| Not as described | 6 | 27% |
| Quality | 4 | 18% |
Ten mentions sounds small. But it was nearly half of all complaints.
Cause. Their product page said "3–5 working days." Their courier averaged 8.
Action. Changed the stated time to 7–10 days and added tracking emails. Didn't speed anything up — just told the truth.
Result: shipping complaints fell to 1 over the following quarter, and their average rating rose from 4.1 to 4.6.
Nothing about the delivery changed. Only the expectation.
Scenario C — "Not as described" pointing at photography
Setup. Furniture retailer, rising returns on one collection.
Not as described — 38 mentions, 31% share
"The colour is completely different in real life. Looks warm grey
online, it's much browner."
Cause. The collection had been photographed under warm studio lighting. A charcoal fabric rendered brown.
Action. Re-shot under neutral light, added a fabric swatch close-up, and offered free swatch samples.
Result: returns on the collection fell from 22% to 7%. The theme dropped out of the list within two months.
Scenario D — A theme that should have been ignored
Setup. Premium skincare brand saw "Price & value" at 18 mentions and considered a price cut.
Reading the samples changed the decision:
"Lovely product but way too expensive for what it is."
"Works well, just can't justify £60 for a small bottle."
What they noticed. Every quote praised the product. These weren't people saying it was bad value — they were people outside their target market who'd bought anyway.
Cross-checked against their positioning. Their best customers, with the highest repeat rates, never mentioned price.
Action. Didn't cut the price. Added value-justification copy — ingredient sourcing, why the bottle is small, cost per use — and tightened ad targeting.
Result: price mentions fell to 6, and average order value rose 9% because they stopped attracting the wrong buyers.
The suggested fix said "add value-justification copy or a bundle offer" — not "reduce your price." That distinction saved their margin.
Best practices
✅ Sort by share, not mentions. Share tells you what proportion of your unhappy customers each theme represents.
✅ Always read the sample quote before acting. Scenario D turned on it.
✅ Fix the top theme first, then re-check next month.
✅ Fixing expectations often beats fixing the product. Scenario B changed nothing but a delivery estimate.
✅ Use the suggested action as written. They're specific for a reason.
✅ Watch for a theme reappearing. That usually means a supplier or process regressed.
❌ Don't cut prices because "price & value" appears. It's often a targeting problem — Scenario D.
❌ Don't panic over a low-share theme. Two mentions out of 300 complaints is noise.
❌ Don't expect per-product detail. Themes are store-wide; use the quote to find the products.
❌ Don't stop collecting reviews to reduce complaints. Fewer reviews means less visibility, not fewer problems.
Plan tiers
| Capability | Free | Starter | Pro | Agency | Enterprise |
|---|---|---|---|---|---|
| Six objection themes | — | — | ✓ | ✓ | ✓ |
| Mentions and share | — | — | ✓ | ✓ | ✓ |
| Sample quotes | — | — | ✓ | ✓ | ✓ |
| Suggested merchandising fix | — | — | ✓ | ✓ | ✓ |
| Themes in the Daily Ops worklist | — | — | ✓ | ✓ | ✓ |
| Automatic theme clearing | — | — | ✓ | ✓ | ✓ |
| Multi-store theme comparison | — | — | — | ✓ | ✓ |
Frequently asked
Why only negative reviews? Because keyword matching is reliable when the text is already a complaint. "Small" in a one-star review means a sizing problem; in a five-star review it might be praise.
Which ratings count as negative? Three stars and below.
Does this cost anything to run? No. It's keyword matching, not AI, so there are no per-run charges.
Can I add my own themes? Not currently. The six cover the overwhelming majority of commerce complaints. For deeper, custom analysis see Jobs to be done.
How is this different from the Jobs tool? Opposite questions. This finds why people were unhappy. Jobs finds what people wanted. This is always-on and free to run; Jobs is a deeper AI-based analysis.
Why is a theme store-wide instead of per product? Current limitation. Use the sample quote to identify the products, then check those directly.
How many reviews do I need? It works with what you have, but share percentages get more meaningful above roughly 20–30 negative reviews.
A theme disappeared — did I fix it? Probably. Themes clear when they stop appearing in new reviews. If it returns, something regressed.
See also
- Daily Ops overview — where these themes appear in your worklist
- Jobs to be done — the positive counterpart
- Reviews & CRO audit — collecting more reviews
- CRO & storefront copy — implementing the copy fixes
- Image SEO — for "not as described" photography issues
- Customer 360 — who's complaining
- Positioning — checking whether price complaints are a targeting problem