We checked 273,000+ software reviews for a public vendor reply. Almost nobody answers, nobody prioritises the unhappy, and the few replies that exist go overwhelmingly to customers who were already satisfied.
There is a comfortable assumption in software that reviews get managed. Somebody on the customer success team watches the review sites, replies to the angry ones, defuses what can be defused, and turns a bad rating into a recovered customer. Every guide to reputation management assumes this is happening.
We checked. Across 273,000+ software reviews covering 13,000+ companies, we counted how many carry a public vendor response. The answer is 4,000+, which sounds like a lot until you divide: 1.57% of reviews get a reply. Roughly 65 out of every 66 go publicly unanswered.
The number that reframes it is at the company level. Only 424 of 13,000+ companies have ever posted a single public reply to any review. Responding is not a practice that most software companies do badly. It is a practice that 96.8% of them have never once attempted.
The obvious defence is that replying to reviews is not worth anyone's time. So we tested that against a neighbouring category. Across 136,000+ App Store reviews, developers reply at 21.47%, and unlike B2B vendors they reply most often to the angriest users. Same industry, same kind of work, fourteen times the engagement. The silence in B2B software is a choice.
That is the entire finding, and it is worth sitting with before the detail. The standard advice to founders is that incumbents are entrenched and hard to displace. This dataset says the incumbents are not even performing the cheapest possible retention activity, which is typing a reply to a person who took twenty minutes to explain what is wrong with the product.
We used our Capterra review corpus, which holds 273,000+ individual software reviews across 13,000+ companies, each with a star rating, the reviewer's industry, free-text pros and cons, and, where one exists, the vendor's public response and the date it was posted. A review counts as answered if the vendor response field contains text.
Every figure here is a simple count over that corpus, re-run live in September 2026. There is no modelling and no sampling. The full corpus sits behind our complaint analysis platform, and the wider method is described in mining Capterra reviews for SaaS ideas.
Before publishing we checked the fields for the failure mode that ruins studies like this: a column that is empty because the scraper never captured it rather than because the world is empty. Four fields on this table failed that test and were discarded entirely, including the reviewer's stated reason for choosing the product. The vendor response field passed, because its population pattern is plausible, it correlates sensibly with company identity, and the text it contains reads like genuine vendor writing.
| Limitation | Effect on the findings |
|---|---|
| Public replies only | This is the most important one. A vendor may answer the same customer by email, in-app or by phone and never post publicly. We measure public visibility, not whether anyone cared. |
| One review platform | A vendor investing in a different review site would look silent here. Treat the rate as platform-specific rather than as a universal measure of vendor behaviour. |
| Coverage is uneven across categories | Our Capterra collection is incomplete in places, so industry-level cuts depend on which companies were captured. Read the industry table as directional, not as a census. |
| Star distribution is platform-shaped | 66.9% of reviews are 5-star, which reflects how reviews get solicited as much as how users feel. We use the distribution to weight replies, not to claim satisfaction levels. |
| Point-in-time snapshot | A reply posted after we scraped a page would not be captured, which biases the measured rate slightly downward for recent reviews. |
The headline rate of 1.57% understates how concentrated the behaviour is, because those replies are not spread evenly across companies. They come from a small group that does it repeatedly.
| Measure | Value |
|---|---|
| Reviews in corpus | 273,000+ |
| Reviews with a public vendor reply | 4,000+ (1.57%) |
| Companies in corpus | 13,000+ |
| Companies that have ever replied to anything | 424 (3.2%) |
| Companies that have never replied once | 96.8% |
424 companies carry essentially all of the replies in a corpus of 13,000+. Reviewing this as a market, public review engagement is not an industry norm with laggards. It is a niche habit practised by roughly one company in thirty-one, and ignored by everyone else.
Here is the result we expected to be wrong, and re-ran twice because of it. If vendors reply selectively, the obvious selection is damage control: answer the 1-star reviews, let the 5-star ones speak for themselves. That is not what happens.
| Star rating | Reviews | Share of all reviews | Got a reply | Reply rate |
|---|---|---|---|---|
| 1 star | 5,000+ | 2.1% | 90 | 1.56% |
| 2 stars | 3,000+ | 1.4% | 80 | 2.16% |
| 3 stars | 13,000+ | 5.0% | 254 | 1.84% |
| 4 stars | 67,000+ | 24.6% | 1,000+ | 1.59% |
| 5 stars | 183,000+ | 66.9% | 2,000+ | 1.53% |
Read the reply-rate column: 1.56, 2.16, 1.84, 1.59, 1.53. There is a slight bump at 2 stars, and otherwise the line is flat. A 1-star review is almost exactly as likely to be answered as a 5-star one. Whatever process generates these replies, it is not triage, because triage would produce a visible slope.
A flat rate applied to a lopsided distribution produces a lopsided outcome. Because 5-star reviews are 66.9% of everything, the flat reply rate sends the overwhelming majority of replies to satisfied customers.
Roughly 65% of all vendor replies land on 5-star reviews. Meanwhile, 1-star and 2-star reviews combined received 170 replies in total across the entire corpus of 273,000+ reviews and 13,000+ companies. One hundred and seventy.
That is the finding in its sharpest form. The scarce resource of vendor attention flows almost entirely to people who were already going to renew, and almost none of it reaches the people actively documenting why they are leaving. It is the reputational equivalent of thanking the customers walking in and ignoring the ones walking out.
The content of the replies explains the distribution. Sampling the vendor responses, the dominant form is a short, warm acknowledgement of praise. Representative examples, anonymised:
"Thank you for your review and for your kind words for our support team. I'll make sure to pass those along."
"Thank you for your kind words and taking the time to share them. We are honored to serve you and look forward to many more years of helping your school."
These are perfectly pleasant, and they cost the vendor nothing. They are also not service recovery, not product research, and not a signal to a prospective buyer reading the page that this company engages with criticism. The replying behaviour that exists is a marketing reflex pointed at the easiest possible target.
Set that against what sits in the unanswered low-rated reviews. These are real complaints from the corpus, each from a 1 or 2-star review with no vendor response, anonymised to the platform:
"If you added a bank account, you can't remove it."
"Admin has no mobile option, many steps involved in completing transactions, especially if you need to send pictures with an invoice."
"It was buggy and constantly required tech support for workarounds. They would make quick fixes when they were important to them, but if they didn't feel an update was important, they would file the suggestion in the round filing cabinet."
Notice the difference in information content. The replies carry none. The unanswered complaints are specific, reproducible and in several cases trivially fixable. "You cannot remove a bank account once added" is a bug report with a clear acceptance criterion. It received no public response, and it is sitting on a public page where every prospective buyer of that product can read it.
This is the raw material our complaint-backed business ideas work is built on, and why we treat review text as a primary source rather than as reputation noise.
Among reviews that do get an answer, the median gap between the review and the reply is 7 days. The mean is 58.2 days.
That gap between median and mean is the interesting part. It means a core of vendors reply within a week, and a long tail replies months later. A reply arriving two months after a frustrated customer wrote a review is not recovery. The customer has either churned or adapted, and the reply exists for the benefit of whoever reads the page next, which is a different job entirely.
Cutting by the reviewer's industry produces a finding with some irony in it.
| Reviewer industry | Reviews | Reply rate | Avg rating |
|---|---|---|---|
| Accounting | 4,000+ | 5.22% | 4.50 |
| Education Management | 9,000+ | 3.10% | 4.55 |
| Higher Education | 3,000+ | 2.78% | 4.54 |
| Marketing and Advertising | 12,000+ | 2.40% | 4.59 |
| Financial Services | 7,000+ | 2.14% | 4.54 |
| Information Technology and Services | 18,000+ | 1.53% | 4.60 |
| Real Estate | 6,000+ | 1.42% | 4.51 |
| Computer Software | 12,000+ | 1.32% | 4.55 |
Accounting buyers are close to four times as likely to get a public answer as Computer Software buyers. The industry that builds software is the one least likely to engage publicly with feedback about software, despite its buyers being the most technically capable of writing a useful bug report.
Average ratings barely move across the table, running from 4.49 to 4.60, so this is not explained by some industries being angrier than others. The variation is in vendor behaviour, not customer sentiment. Vertical software serving accountants and schools engages; horizontal software serving technologists does not, which is consistent with what we see in pain-point-backed SaaS ideas and in the vertical categories covered in boring business ideas.
A reasonable objection to everything above is that we might be looking at a legacy problem. Review response could be a recent practice, with the corpus dominated by older reviews from before anyone took review sites seriously. The year-by-year data says no.
| Review year | Reviews | Got a reply | Reply rate |
|---|---|---|---|
| 2018 | 26,000+ | 312 | 1.19% |
| 2019 | 29,000+ | 436 | 1.50% |
| 2020 | 26,000+ | 443 | 1.69% |
| 2021 | 30,000+ | 393 | 1.30% |
| 2022 | 35,000+ | 578 | 1.64% |
| 2023 | 35,000+ | 585 | 1.63% |
| 2024 | 36,000+ | 848 | 2.35% |
| 2025 | 30,000+ | 512 | 1.69% |
Across eight years the rate moves from 1.19% to 1.69%. There is a single visible bump in 2024 at 2.35%, which falls back the following year. Nothing here looks like an industry adopting a practice. It looks like a stable, low-level behaviour that persisted through exactly the period in which software buyers moved almost entirely to researching purchases online.
The stability matters competitively. A gap that has stayed open for eight years is not about to close because you noticed it, which is a very different risk profile from a gap created by a recent change. That durability is the property we look for when running a market gap analysis, and it is why unaddressed complaints feed our pain-point-backed idea work rather than being treated as noise.
Even the 3.2% who reply are not really 3.2%, because the replies concentrate hard inside that small group.
Of the 4,000+ replies in the corpus, 78.4% come from companies that have posted 10 or more. A minority within the minority does almost all of the public engagement, presumably because they have an actual process and someone who owns it. The rest have posted a handful each, consistent with occasional ad-hoc responses rather than a policy.
That compresses the real picture considerably. On the order of a hundred-odd companies, out of 13,000+, are systematically engaging with their public reviews. Everyone else is silent or sporadic. If you are sizing up a category with the approach in validating a startup idea, assume silence is the default and treat an engaged incumbent as the exception worth noting.
One honest caution belongs here, because it cuts against the obvious conclusion. Companies that reply do not carry higher ratings. Their average is 4.479 against 4.528 for companies that never reply, a difference small enough to be noise and pointing the wrong way regardless. Nothing in this dataset supports the claim that replying improves how a product is rated. If you adopt the practice, adopt it for the product intelligence and the prospect-facing signal, not because it will lift your score. The same caution applies to the narrow feature bets in single-feature micro SaaS ideas.
Everything above could be read as a fact about software, or about how busy software teams are, or about the futility of replying to reviews in general. There is a clean way to test that: check a different review platform. We ran the same measurement across 136,000+ App Store reviews covering 6,000+ apps.
The result is not close.
| Star rating | B2B software reply rate | App Store reply rate |
|---|---|---|
| 1 star | 1.56% | 22.64% |
| 2 stars | 2.16% | 22.81% |
| 3 stars | 1.84% | 22.82% |
| 4 stars | 1.59% | 22.04% |
| 5 stars | 1.53% | 17.13% |
| Overall | 1.57% | 21.47% |
| Share of publishers who ever reply | 3.2% | 35.7% |
App developers reply to reviews at roughly fourteen times the rate of B2B software vendors, and they are eleven times as likely to have ever replied at all. More than a third of apps engage with their reviews. One in thirty-one B2B software companies does.
The shape is different too, and this is the part that settles the argument. On the App Store the reply rate is highest for the worst reviews: 22.64% at 1-star, 22.81% at 2-star, 22.82% at 3-star, falling to 17.13% at 5-star. That is a visible slope in the direction you would expect from a team doing triage. The B2B line, by contrast, is flat at around 1.5% regardless of how angry the customer is.
Mobile developers prioritise unhappy users. B2B software vendors do not prioritise anything, because they are almost entirely absent. Whatever is causing the B2B pattern, it cannot be that replying to reviews is impossible, unrewarding or culturally alien to software people. An entire adjacent category of software companies does it at scale, every day, and skews its effort toward the complaints.
One caveat on this comparison, and it is important. Our App Store corpus is deliberately weighted toward low-rated reviews, because we collect it to mine pain points, so 1-star reviews are a far larger share of it than they are of the Capterra corpus. That is why the comparison above is drawn per star rating rather than as a single blended figure: comparing overall rates across two differently composed samples would be misleading. Within each star band the two platforms are directly comparable, and the gap holds at every band. The App Store review data feeds our App Store review analysis work.
The practical reading for a founder: if you are building B2B software, the norm in your category is silence, and adopting the mobile norm would make you visibly unusual to every prospect reading a review page. If you are building a mobile app, replying is table stakes and gets you nothing on its own.
The data does not contain motive, so what follows is interpretation rather than measurement, and should be read that way.
The flat reply rate is the clue. A deliberate policy of any kind, whether damage control or reputation polish, would bend the line. A flat line across ratings looks like the output of a process that is not reading the reviews at all: a periodic sweep, a delegated task, a marketing routine that replies to whatever is in front of it without regard to content. Combine that with the fact that 96.8% of companies do nothing, and the most parsimonious explanation is that public review response is simply not owned by anyone at most software companies.
There is also a plausible incentive story. Replying to a 5-star review is costless and pleasant. Replying to a detailed 1-star review requires admitting a fault in public, coordinating with a product team, and committing to something. The asymmetry in effort matches the asymmetry in the data.
The App Store comparison narrows the candidate explanations considerably. Any theory that explains the B2B silence has to also explain why mobile developers, working on the same kind of product with the same constraints, behave completely differently. Theories about software teams being busy, or reviews being unimportant, or replying being futile, all fail that test.
What survives is structural. App Store reviews sit on the same page as the install button, so a visible complaint costs the developer downloads immediately and the feedback loop is tight. A B2B software purchase runs through demos, trials and procurement, and the review page is one input among many, read weeks before a decision. The consequence of an unanswered complaint is real but diffuse and delayed, and diffuse delayed consequences are exactly the kind that nobody is assigned to own. That is a gap created by organisational structure rather than by anyone deciding to ignore customers, which is also why it has persisted for eight years without closing.
For founders, this is a competitive finding dressed up as a customer-service statistic, and it cuts three ways.
The complaints are unclaimed. The specific, fixable problems in low-rated reviews are not being addressed publicly, and in many cases the flat reply rate suggests they are not being read. A gap that a competitor has publicly ignored for years is a different proposition from one they are actively closing. This is the premise behind finding SaaS ideas in negative reviews and validating before you code.
The unhappy customers are identifiable and unattended. They have described their problem in detail, in public, and received nothing back. That is a warm audience by any reasonable definition, and our Reddit market research and complaint database tooling exists to surface exactly these.
Responding is a cheap differentiator if you are small. Doing something 96.8% of your category has never attempted is unusually cheap leverage for a solo founder. It does not require engineering. It requires reading. Founders working through getting a first customer routinely overlook that the people most likely to switch have already raised their hands in public.
A caution on the third point, because it is the one most likely to be over-applied. Replying is cheap leverage, not a growth strategy. The concentration data shows the companies that do it are doing it systematically, ten or more times, which means a founder who answers three reviews and stops has not actually cleared the bar. And as noted above, the companies that reply do not carry higher ratings, so treat it as a way to learn what is broken and to be visibly present on the page a prospect is reading, rather than as something that moves a number on its own.
It would be easy to overclaim here, so here is the honest boundary of what this shows.
This measures public responses on one platform. A vendor that emails every unhappy reviewer privately, fixes the bug in the next release and never posts a public word would appear in this dataset as completely disengaged. That is a real possibility and we cannot rule it out. If you want to read the finding conservatively, read it as: software vendors do not publicly engage with reviews, and prospective buyers therefore see an unanswered complaint.
Even under that conservative reading the competitive implication survives, because the buyer reading the review page cannot see the private email either. What they see is a detailed complaint and silence.
We also cannot show that responding improves anything. There is no revenue, retention or churn data attached to these reviews, so nothing here proves that a vendor who replies keeps more customers. The claim is about what is happening, not about what it earns, the same limit we put on our analysis of what SaaS homepage headlines actually say.
Three practical moves follow, in order of effort.
Read your target category's low-rated reviews before you build anything. They are the cheapest primary research available and, on this evidence, largely unread by the incumbent. Start with how to find SaaS ideas if you need a process, or the idea validation tool to check what you find.
Check whether the incumbent replies before you assume they are responsive. A category where nobody answers is a category where switching costs are emotional as well as technical. Our review and complaint tooling roundup and G2 review analysis cover how to do this across platforms.
If you ship something, answer everything. The bar is 96.8% silence. Clearing it costs you an hour a week and it is visible to every prospect who reads your review page, which is a rare case of a marketing activity and a product research activity being the same activity. Pair it with the upstream work in idea validation and market research tools.
Read the complaints nobody answered. BigIdeasDB indexes 1M+ complaints from Capterra, G2, the App Store and Reddit, so you can find the documented, unaddressed problems in any software category before you write a line of code.
Start researching with BigIdeasDB →Almost never. Across 273,000+ software reviews, only 4,000+ carry a public vendor reply, a rate of 1.57%. Roughly 65 out of every 66 reviews go publicly unanswered.
3.2%. Of 13,000+ companies in the corpus, only 424 have ever posted a single public response to any review. The remaining 96.8% have never replied once.
No. The reply rate is essentially flat across star ratings: 1.56% at 1-star, 2.16% at 2-star, 1.84% at 3-star, 1.59% at 4-star and 1.53% at 5-star. Because 5-star reviews are 66.9% of the corpus, about 65% of all replies go to customers who were already happy, while 1 and 2-star reviews received 170 replies in total.
When they respond at all, the median gap is 7 days and the mean is 58.2 days. The difference between the two reveals a long tail of replies arriving months later, too late to matter to the reviewer.
Dramatically more. Across 136,000+ App Store reviews, developers replied to 21.47%, against 1.57% for B2B software reviews, roughly fourteen times the rate. 35.7% of apps have replied to at least one review versus 3.2% of B2B software companies. App developers also reply most often to the worst reviews, 22.64% at 1-star against 17.13% at 5-star, while the B2B reply rate stays flat regardless of how angry the customer is.
Accounting buyers get the highest public reply rate at 5.22%, then Education Management at 3.10% and Higher Education at 2.78%. Computer Software is the lowest major industry at 1.32%, so software buyers are among the least likely to get an answer.
BigIdeasDB Research. (2026). 96.8% of Software Vendors Never Answer a Review. BigIdeasDB. Retrieved from https://bigideasdb.com/software-vendors-ignore-negative-reviews