The starting point
A direct-to-consumer furniture brand with hundreds of thousands of customers who liked what they bought. Their Google rating told a different story: 3.6 stars from about 4,000 reviews, a tiny fraction of their buyers.
Most of those reviews were about one thing, a delivery problem from 2023 that had long since been fixed. Nobody else had a reason to write anything. The happy customers sat on their sofas and said nothing, because nobody asked them.
When people asked AI models whether the brand was any good, the answers repeated the delivery complaints. The number and the sentence both came from a bad few months, years earlier.
What we did
Find the happy moment
We mapped the customer journey to find where satisfaction actually peaks. For furniture it isn't the day of purchase or the day of delivery. It's a few days after the piece is in the room, assembled and used. That became the moment to ask.
Ask everyone, once
One request, at that moment, to every customer. No incentive, no discount code, no filter that only sends the ask to people who rated a survey highly. Asking everyone is what makes the result honest, and it's what keeps it on the right side of platform rules.
Answer every one-star
Every critical review got a public reply from a real person: what went wrong, what changed, and how to get it sorted. The old delivery reviews stayed up, answered. Anyone reading them could see the problem had been fixed.
What changed
- Google rating rose from 3.6 to 4.5 in four months.
- Monthly review volume grew nine times, so the rating now reflects recent customers rather than a bad season in 2023.
The product didn't change. The sample did.
By the numbers
| Measure | Day one | Month 2 | Month 4 |
|---|---|---|---|
| Google rating | 3.6 | 4.2 | 4.5 |
| Google reviews, total | 4,180 | 7,960 | 13,370 |
| New reviews per month | 300 | 1,900 | 2,700 |
| Reviews mentioning delivery problems | 62% | 21% | 9% |
| One-star reviews with a public reply | 0% | 100% | 100% |
| Models naming them (out of 5) | 2 | 3 | 4 |
What we tested before choosing the moment
We sent the review request at three points in the journey to a small group of customers each, and measured both the response rate and the average rating that came back.
| Moment | Response rate | Average rating |
|---|---|---|
| Day of purchase | 4% | 4.1 |
| Day of delivery | 7% | 3.9 |
| Five days after delivery | 11% | 4.7 |
Five days after delivery won on both counts. On delivery day, the furniture was often still in the box.
In their words
"We'd assumed the score was a product problem. It was a sampling problem. Our customers were happy the whole time; nobody had asked them."
Head of Customer Experience, Furniture brand
What made it work
Timing the ask, and refusing every shortcut. A score built with incentives or gating can disappear in one platform sweep. This one is built from real customers asked at the right moment, so it holds.