We classified 162,000+ software reviews by the reviewer's job title. The person who signs the cheque and the person who uses the product file almost opposite complaints about the same software.
Ask a software company what customers complain about and you will get one answer. Ask their customers and you will get several, depending entirely on which customer you asked. That is not a figure of speech. It is measurable, and the size of the effect is larger than most teams assume.
We took 273,000+ software reviews and classified each one by the job title the reviewer gave. 162,000+ of them resolved cleanly into five seniority bands, from owners and founders down to individual contributors. Then we measured what each band actually complains about.
The result is close to a mirror image. Owners complain about money. The people using the software complain about whether it works. Both are reviewing the same products, often the same products as each other, and their complaint profiles run in opposite directions.
The practical consequence is that "what customers complain about" is not a well-formed question. For a product sold to owner-operators, price and the vendor relationship are the churn risks. For a product rolled out to staff, friction and reliability are. Optimising for the wrong one is a common and expensive mistake, and the data says both groups are loud enough to mislead you.
Our Capterra corpus holds 273,000+ software reviews, each with a star rating, free-text pros and cons, and the reviewer's self-reported job title. Titles are raw free text and extremely varied, 80,000+ distinct strings, so they cannot be grouped as-is.
We banded them with pattern matching on seniority words: owner, founder, CEO, president and partner into one band; CTO, CFO, COO and VP into another; director and head of; manager, supervisor and lead; and finally assistant, coordinator, specialist, analyst and similar into individual contributor. Anything that did not match a pattern was left unclassified rather than forced into a bucket.
Then we counted how often each band's complaint text mentions six recurring themes: price, usability, reliability, support, reporting and integration. Every figure below is a live query against that corpus, run in September 2026. The same corpus underpins our complaint analysis platform and the method described in customer pain point analysis.
| Limitation | Effect on the findings |
|---|---|
| 40.7% of titles did not classify | Job titles are free text and many are role descriptions rather than seniority markers. The unclassified group is large, so band sizes are not a census of who reviews software. |
| Banding is pattern-matched, not semantic | A word like "lead" can be a seniority marker or part of another phrase. Expect some misclassification at the edges of each band. |
| Theme detection is lexical | We count keyword mentions, so a complaint phrased without those words is missed. Absolute percentages therefore understate every theme. The comparison between bands is the finding, not the level. |
| Titles are self-reported | Nobody verifies them, and title inflation varies by company size, which likely blurs the boundary between the owner and C-suite bands in small companies. |
| One platform, uneven category coverage | Our collection is incomplete across some software categories, so the theme mix partly reflects which products were captured rather than software as a whole. |
| The industry field carries a parse artifact | A share of rows record the literal string "Used the software for:" in the industry field, a scraper label that leaked into the data. We excluded it from the per-industry table rather than treating it as an industry. |
| No churn or revenue outcome attached | We measure what each role says, never whether they left or what it cost. Nothing here proves a complaint predicts churn. |
162,000+ reviews resolved into five bands. Managers are the largest single group, and owners are close behind, which is itself worth noting: a large share of B2B software reviews are written by the person who owns the company.
| Band | Reviews | Share of all titled reviews | Avg rating |
|---|---|---|---|
| Manager | 52,000+ | 19.3% | 4.54 |
| Owner / exec | 46,000+ | 17.2% | 4.50 |
| Director | 27,000+ | 10.1% | 4.55 |
| Individual contributor | 26,000+ | 9.7% | 4.53 |
| C-suite / VP | 8,000+ | 3.0% | 4.57 |
| Unclassified | 111,000+ | 40.7% | 4.52 |
Price complaints fall monotonically as you move away from the budget. Every step down the ladder reduces how often money comes up.
| Band | Price | Usability | Reliability | Support | Reporting | Integration |
|---|---|---|---|---|---|---|
| Owner / exec | 8.00% | 7.85% | 2.66% | 9.97% | 3.75% | 1.62% |
| C-suite / VP | 7.08% | 8.53% | 2.69% | 9.45% | 6.33% | 2.06% |
| Director | 5.98% | 8.62% | 2.66% | 8.65% | 5.68% | 1.59% |
| Manager | 5.24% | 8.49% | 3.08% | 8.06% | 5.54% | 1.40% |
| Individual contributor | 4.91% | 9.94% | 3.71% | 7.05% | 5.02% | 1.44% |
8.00%, 7.08%, 5.98%, 5.24%, 4.91%. An owner is 63% more likely to raise price in a complaint than someone at the bottom of the org chart. That is not surprising in direction, but the smoothness is: there is no cliff, just a steady decline as financial responsibility recedes.
For anyone pricing a product, this reframes the feedback problem. Price objections in your review pipeline are concentrated in one band, and if you sell to owner-operators, that band is your market. The pricing patterns we measured in SaaS pricing strategies and what micro SaaS actually charges land differently depending on which side of this gradient your buyer sits.
Now read the usability and reliability columns. They run the opposite way.
Usability: 9.94% for individual contributors, falling to 7.85% for owners. Reliability, meaning crashes, bugs, slowness and freezes: 3.71% for individual contributors against 2.66% for owners and directors, a 39% gap. The further you get from daily use of the software, the less often you mention that it is confusing or broken.
This is the finding worth internalising. The two groups are not complaining about the same product more or less intensely. They are complaining about different aspects of it entirely. An owner evaluating renewal is asking whether the line item is justified. An assistant using it eight hours a day is asking why the export button fails.
Both complaints are real and neither is a proxy for the other. A product can be simultaneously too expensive for the buyer and infuriating for the user, and fixing one does nothing for the other.
The obvious objection to everything above is composition. Owners cluster in small businesses and individual contributors cluster in large ones, and those two groups buy different software at different prices. If that were driving the result, the gradient would be an artifact of industry mix rather than a fact about roles.
So we re-ran the comparison within each industry, restricted to industries with at least 300 reviews from each band. If the effect is real, it should survive inside every one of them.
| Industry | Owner price | IC price | Owner usability | IC usability |
|---|---|---|---|---|
| Marketing and Advertising | 9.09% | 7.01% | 4.88% | 7.19% |
| Information Technology | 7.21% | 6.60% | 5.69% | 7.41% |
| Health | 6.80% | 4.33% | 6.29% | 9.15% |
| Retail | 10.18% | 6.76% | 6.08% | 7.83% |
| Computer Software | 7.61% | 5.95% | 5.31% | 8.41% |
| Construction | 8.19% | 4.75% | 6.38% | 6.84% |
| Real Estate | 7.08% | 3.73% | 7.36% | 7.78% |
It holds everywhere. In all seven industries owners mention price more often than individual contributors do, and in all seven individual contributors mention usability more often than owners do. There is not a single reversal in either direction.
The magnitude varies in an interesting way. The price gap is widest in Real Estate, where owners raise price nearly twice as often as individual contributors (7.08% against 3.73%), and in Construction (8.19% against 4.75%). Both are owner-operator industries where the person reviewing the software frequently is the business. It is narrowest in Information Technology (7.21% against 6.60%), where more reviewers are salaried staff at larger employers and the owner band is likely diluted by people with titles like partner or principal who do not personally carry the budget.
That variation is itself a useful signal when picking a market. In trades and property, your reviewer and your payer are the same person, so price objections dominate the public record. The vertical categories in boring industries begging for micro SaaS and vertical AI SaaS ideas skew heavily toward exactly those owner-operator structures, which means public complaint data about them is price-weighted by construction, not by coincidence.
Retail deserves a note as the highest owner price rate in the set at 10.18%. More than one in ten complaints from retail owners mentions cost, the highest of any industry and band combination we measured. Thin margins are the obvious explanation, and it is consistent with what we see across small business software pain points.
Reporting does not follow either gradient. It peaks in the middle of the org chart and collapses at the top.
C-suite and VPs raise reporting in 6.33% of complaints, directors 5.68%, managers 5.54%, individual contributors 5.02%, and owners just 3.75%. Owners are the least likely of all five bands to complain about reporting, by a clear margin.
The likely reason is structural rather than preferential. Reporting exists to explain work to somebody else. A middle manager or VP has to justify a team's output upward, so a tool that will not produce the right view creates a real, recurring problem. An owner-operator of a small business does not report to anyone, so a missing dashboard is an inconvenience rather than an obstacle.
If you are building for the mid-market, this is a concrete signal: reporting is a genuine buying criterion for the band that evaluates you, and a much weaker one for owner-led businesses. That distinction matters when picking a segment, as we discuss in who micro SaaS actually sells to.
Support complaints follow the price gradient almost exactly: 9.97% for owners, 9.45% C-suite, 8.65% directors, 8.06% managers, 7.05% individual contributors.
Support is, notably, the single largest theme for owners, larger than price. That pairing of price and support at the top of the ladder describes a coherent worldview: the person paying evaluates the vendor as a relationship, priced and serviced, rather than as a piece of software. The user below them evaluates the software.
It also gives extra force to a finding from our study of vendor responses to reviews, where only 3.2% of software companies have ever publicly answered one. The band most likely to complain about support is the band that controls renewal.
Beyond what they complain about, owners differ in how hard they swing.
| Band | 1-2 star reviews | 5-star reviews |
|---|---|---|
| Owner / exec | 6.50% | 72.13% |
| C-suite / VP | 3.72% | 70.87% |
| Director | 3.32% | 68.60% |
| Manager | 2.47% | 65.30% |
| Individual contributor | 1.75% | 63.16% |
Owners are 3.7 times more likely to leave a 1 or 2-star review than an individual contributor, and simultaneously the most likely to leave a 5-star one. They occupy both tails. Individual contributors cluster in the middle, rating generously even while describing friction.
That combination makes owner reviews the most informative and the most dangerous in any review corpus. They carry the strongest signal about whether a product is worth money, and they will skew any average you compute without segmenting.
Here is the null result that makes the rest matter. Average ratings by band: 4.50, 4.53, 4.54, 4.55, 4.57. The spread across the entire org chart is 0.07 of a star.
If you were monitoring customer satisfaction by role using average scores, you would conclude that every role feels identically about your product, and you would be right about the score and completely wrong about the substance. The signal lives in the text and only appears once the text is segmented.
This is the general case for reading complaints rather than metrics, and it is why our corpus is built on review text rather than scores. The same logic drives complaint databases and the approach in customer review analysis.
The quotes make the split concrete. All are real, anonymised, and attributed to the platform only.
An owner, one star:
"Can't switch credit card companies without paying a fortune. You'd trapped into a high price processing company or pay a fortune to be accessible to another company."
An owner, five stars, still leading with money:
"It's a little pricey for small businesses, but considering how much time it's saved me it's been 100% worth it. I would like to upgrade to the white label version soon, but haven't due to the price."
An individual contributor, also five stars:
"While the interface is simple, there is a learning curve and the time to implement can be lengthy."
"The learning curve is steep for users unfamiliar with SQL, but the entire platform's community is friendly and helpful!"
Read those together and the pattern is unmistakable. The owners frame everything as a transaction, even when delighted. The individual contributors describe real friction and then rate five stars anyway, often apologising for the complaint. Same corpus, same rating scale, entirely different relationship to the product.
If each role is genuinely evaluating against a different question, then the pattern should show up in what they like as well as what they dislike. It does, almost perfectly.
| Band | Praises value for money | Praises ease of use | Praises support | Praises time saved |
|---|---|---|---|---|
| Owner / exec | 8.42% | 40.84% | 23.14% | 17.08% |
| C-suite / VP | 7.43% | 39.76% | 24.96% | 16.85% |
| Director | 7.16% | 43.51% | 25.55% | 16.71% |
| Manager | 5.48% | 44.99% | 23.00% | 16.92% |
| Individual contributor | 4.08% | 45.79% | 19.40% | 17.35% |
Value for money runs 8.42%, 7.43%, 7.16%, 5.48%, 4.08%. That is the price-complaint gradient again, in the same order, with owners 2.1 times more likely than individual contributors to praise value. Ease of use runs the other way, 45.79% down to 40.84%, matching the usability-complaint inversion.
So it is not that owners are negative about money and users are negative about the interface. It is that each role evaluates the product on one axis and reports both directions along it. Money is the owner's axis, in praise and in criticism. Friction is the user's axis, in praise and in criticism. Nobody is being unreasonable; they are answering different questions.
One theme is genuinely universal. Time saved sits between 16.71% and 17.35% across every band, the flattest measure in the entire study. Whatever else separates an owner from an assistant, both notice when software gives them time back, and both say so at the same rate. If you are looking for a single benefit that lands on every seat in an organisation, the data points at that one, and it is consistent with the demand signals we see in business pain points and startup idea validation.
Support praise has its own shape, peaking with directors at 25.55% rather than with owners. Combined with owners having the highest support complaint rate at 9.97%, that suggests owners hold the strongest opinions about support in both directions while directors are the most consistently satisfied by it.
A final small signal: owners write the longest complaints, averaging 176 characters against 163 for managers. Not a dramatic gap, but it points the same way as their rating polarisation. The band with money at stake writes more, rates harder, and cares along one axis.
The data contains no motive, so this is interpretation, and should be read as such.
The cleanest explanation is that each role is evaluating against a different question. An owner is asking whether this line item earns its cost, so price and the vendor relationship dominate, and the software itself is largely a black box they do not personally operate. An individual contributor never sees the invoice and cannot act on it, so the only surface they can evaluate is the one they touch, which is the interface and its reliability. Middle management sits between, judged on output they must evidence, which is why reporting bulges there.
The polarisation of owners fits the same frame. Money creates stakes. Someone spending their own company's budget has a reason to feel strongly in both directions, while an employee handed a tool has limited incentive to write a scathing review of a decision that was not theirs.
Three consequences, in order of how much they should change your behaviour.
Your roadmap depends on who you sell to, not on what complaints are loudest. If your buyer is an owner-operator, usability complaints in your review feed come from a band that is not deciding renewal, while price and support complaints come from the band that is. If you sell seats into larger companies, the reverse applies, and a reliability problem can end a rollout regardless of how happy the owner is with the contract.
Complaint mining needs a role filter. Reading a category's reviews without segmenting gives you an average of two incompatible worldviews, which describes nobody. This is the refinement we now apply when sourcing opportunities in complaint-backed business ideas and negative review mining.
A wedge can target one band specifically. The gap between what an owner pays for and what a user endures is a real product opening: tools that make the user's day better rarely get bought, and tools that satisfy the buyer often get resented. Closing that gap deliberately is a positioning decision, and it is the kind of opening we track in pain-point-backed SaaS ideas and boring business ideas.
Treat owner reviews as your early warning system. The polarisation finding has a practical edge: because owners are 3.7 times more likely than users to leave a 1 or 2-star review, a single angry owner review is worth more attention than the same review from a seat-holder, both as a churn signal and as a public asset a prospect will read. They are also the band that controls renewal and the band most likely to complain about support, which is a bad combination to leave unanswered given that 96.8% of vendors never reply to anything. If you triage nothing else, triage the owners, and answer them in public where the next prospect evaluating you will read the exchange.
The dataset has no outcomes attached. We can show that owners complain about price more often than users do. We cannot show that a price complaint predicts churn, that fixing usability retains anyone, or that any band spends more as a result of anything they wrote. There is no revenue, renewal or churn field here.
Two further honest limits. 40.7% of titled reviews did not classify, so the bands are a large sample rather than a complete census. And our theme detection is keyword-based, so every absolute percentage understates the true rate; a complaint about cost phrased without any of our price words is invisible to it. The comparison between bands is what survives those limits, because the same undercount applies to all five.
We have applied the same caution elsewhere: measuring what a corpus says is not the same as measuring what it earns, a line we also drew in what SaaS homepage headlines actually say.
Segment before you read. Before mining a category's complaints, decide which role you sell to and filter to it. The unfiltered view blends buyers and users into a persona that does not exist.
Weight price feedback by who gave it. A price objection from an owner in an owner-led market is a buying signal. The same objection from a seat-holder in an enterprise rollout usually is not.
Do not use average ratings for this at all. The 0.07-star spread across the org chart proves scores cannot see the difference. Use the text, or use a source that has already structured it, such as the idea validation tool or how to find SaaS ideas.
If you are choosing a market rather than refining one, the upstream work matters more than any of this. Start with idea validation, then market research tools and the SaaS market research guide to decide who you are building for before deciding what to fix.
Filter complaints by the buyer you actually serve. BigIdeasDB indexes 1M+ complaints from Capterra, G2, the App Store and Reddit, so you can separate what the buyer objects to from what the user endures before you commit a roadmap to either.
Start researching with BigIdeasDB →This study is a banding rule plus six keyword filters. Nothing about it is proprietary, and the method transfers to any review corpus that carries a job title, including your own support tickets, NPS verbatims or win-loss notes.
The banding rule we used, in plain terms: match the title against seniority words, and leave anything that does not match unclassified rather than forcing it into a bucket. The five patterns are owner, founder, CEO, president, principal or partner; CTO, CFO, COO, CIO, chief or VP; director or head of; manager, supervisor or lead; and assistant, coordinator, clerk, associate, specialist, technician, analyst or representative. Order matters, because a title like "VP of Operations" would also match on a looser rule, so evaluate the most senior patterns first.
Then compute, for each band, the share of complaint texts matching each theme. The critical discipline is that the comparison between bands is the finding, never the absolute level. Keyword matching undercounts every theme, because people describe cost without using the word price. That undercount applies equally to all bands, so the ratio between them survives while the raw percentage does not.
Three checks worth running before you trust your own version of this. First, control for a confounder the way we controlled for industry: if the gradient does not survive inside a single segment, it is composition rather than role. Second, look at the praise text as well as the complaint text, because a real behavioural difference shows up in both directions and a measurement artifact usually does not. Third, check the field you are banding on for parse artifacts before you trust it, which is how we caught a scraper label sitting in the industry column masquerading as an industry.
If you would rather not build the pipeline, the same corpus is queryable directly through our MCP server for AI research, and the structured version powers our review and complaint tooling.
Yes, and the split is close to a mirror image. Across 162,000+ classified reviews, owners and executives mention price in 8.00% of complaints against 4.91% for individual contributors, a 63% gap. Individual contributors mention usability in 9.94% against 7.85% for owners, and reliability in 3.71% against 2.66%. The people who pay complain about money. The people who use it complain about whether it works.
Owners, founders and CEOs, at 8.00%. The gradient falls consistently with distance from the budget: C-suite and VPs 7.08%, directors 5.98%, managers 5.24%, individual contributors 4.91%.
Owners and executives. 6.50% of their reviews are 1 or 2 stars against 1.75% for individual contributors, a 3.7x difference, and they also leave more 5-star reviews, 72.13% against 63.16%. They occupy both tails.
Reporting, disproportionately. It peaks with C-suite and VPs at 6.33%, stays high for directors at 5.68% and managers at 5.54%, and falls to 3.75% for owners, who are the least likely of any band to raise it.
Barely. Averages sit between 4.50 and 4.57 across every band, a spread of 0.07 of a star. Job title predicts the content of the complaint, not the score, which is why rating averages hide this pattern completely.
BigIdeasDB Research. (2026). Who Complains About What in Software Reviews. BigIdeasDB. Retrieved from https://bigideasdb.com/who-complains-about-what-in-software-reviews