NetSuite reviews — the score depends entirely on which record you open

The three public records, read on the same day

Every score below was pulled from the platform that issues it on 21 August 2026, not from an aggregator and not from a search snippet. That distinction is the whole reason this section exists, and we will come back to why.

• G2 — 4.1 out of 5, from 4,919 reviews. The largest B2B software record NetSuite has anywhere.
• Capterra — 4.2 out of 5, from 2,062 reviews. An independent platform, a different reviewer population, and a score within a tenth of G2's.
• Trustpilot — TrustScore 1.5, from 50 reviews. A consumer-review platform, and a sample two orders of magnitude smaller than G2's.

A score without its count is not a fact, so we have printed all three together and will not use any of them without the number of reviews attached. G2 and Capterra agree with each other to within a rounding error across nearly seven thousand combined reviews. Trustpilot does not agree with either, and it holds fifty.

Note the method too, because it changes what is possible. G2, Capterra and Trustpilot all refuse automated reads — a direct fetch of any of them returns a 403. These figures came through an API that reaches each platform's own record. If you are evaluating any vendor and find yourself quoting a star rating from a search result, you are quoting a cached summary rather than the live page, and search summaries have contradicted live pages repeatedly.

Why the three disagree, and why that helps you rather than confusing you

The temptation is to decide one platform is lying. None of them is. They are sampling different populations, and once you see how, the gap becomes information.

Who writes on G2 and Capterra. Both are B2B software directories. Reviews arrive largely through vendor-run and platform-run collection campaigns aimed at active users — controllers, staff accountants, operations managers, administrators. The people writing are inside the system daily, and many were invited to review while things were going normally. Volume is high and the distribution is wide.

Who writes on Trustpilot. Trustpilot is built for consumers reviewing merchants. Almost nobody is invited to review their company's ERP there. The people who arrive have gone looking for somewhere to write, which is a population defined by motivation. That is a real and well-understood self-selection effect, it applies to every enterprise vendor on the platform, and it is not a criticism of Trustpilot — it is what a small unsolicited sample is.

What to do with it. Weight by count, and read the small sample for its themes rather than its arithmetic. Fifty reviews cannot outvote 6,981, and they were never trying to. What fifty motivated reviews can do is tell you which failure modes are worth asking about on a reference call — which is genuinely more useful than a number.

We will apply that to ourselves, since it is fair. This project once came close to recommending a product on the strength of "4.7 out of 5 across 89 reviews" taken from an aggregator. When somebody finally opened the two platforms that aggregate claimed to draw from, one carried a rating of 1.4 and the other carried no reviews at all. The page was deleted rather than patched, and the rule that came out of it is the one this section is built on: if a number decides what a page recommends, open the source that issues it. That is why every score above has a platform and a count welded to it.

What reviewers consistently credit it with

Read across the recent G2 reviews and the praise is remarkably consistent in what it is about. It is almost never a feature. It is nearly always consolidation.

• One place for finance, inventory, purchasing and sales. Reviewers repeatedly describe work that used to live on separate platforms — job tracking and billing, AR and AP, reconciliations — arriving in a single record. This is the thing people say first, and they say it in almost the same words.
• Drill-down from a financial report to the underlying transaction. Named specifically and often. For anyone who has tried to trace a variance across three disconnected systems, this is the whole purchase.
• Role-based access at scale. One administrator describes managing over 150 users with role-level permissions and custom dashboards. That is not a small-business capability and it is not something a lighter tool fakes.
• Reporting you can keep shaping. A reviewer with a twenty-five-year career calls it one of the most customisable systems they have used, precisely because reports can be refined indefinitely rather than chosen from a menu.
• Global search. Mentioned by daily users as the thing that removes the need to remember where anything lives.

Anyone who uses NetSuite would recognise that list, and it is worth stating plainly before any criticism: for a business that genuinely needs one ledger over several entities and channels, this is a capable system and the ratings on the two large platforms reflect that.

What the same reviewers dislike — the most credible criticism there is

The useful complaints are not in the one-star reviews. They are inside the four- and five-star ones, written by people who like the product and use it every day. Those reviewers have no reason to exaggerate, and the pattern across recent G2 entries is clear.

• Reporting is powerful and expensive to learn. The recurring phrase is that meaningful reports require advanced knowledge of saved searches, custom records and system configuration. Several reviewers who rate the product highly say building the reports they need was the hardest part of the year.
• Import errors are hard to diagnose. CSV import messages are described as not always making clear what went wrong, which turns a five-minute job into an afternoon.
• Speed on certain record and report types. Raised by administrators rather than occasional users.
• Naming conventions differ from ordinary accounting language. One reviewer notes items where the outside world says GL code and reference numbers where the outside world says invoice number — trivial to a veteran, a real onboarding tax for a new hire.
• Some everyday artefacts are buried. A vendor bill PDF filed two levels deep is the kind of thing nobody notices in a demo and everybody notices in month three.

Every one of those is a training-and-configuration cost rather than a defect, which is exactly what you would expect from a system of this class. It is also why the implementation budget matters more than the licence: most of what people dislike is decided during setup.

The Trustpilot record, stated with its count and no conclusion drawn

We are going to handle this carefully, because handling it carelessly is how a page tells a lie.

The record is a TrustScore of 1.5 from 50 reviews, read 21 August 2026. The recent entries cluster around two themes: support responsiveness after the contract is signed, and renewal terms. Those are the themes. We are not going to average fifty motivated reviews into a verdict about a product with nearly seven thousand reviews elsewhere, we are not going to repeat individual allegations we cannot verify against a primary record, and we would say exactly the same if the polarity ran the other way.

What a small, motivated sample is genuinely good for is telling you what to ask. So convert it into diligence, and ask it of every vendor on your shortlist including us:

• What is the support model, in writing? Named contact or ticket pool, target response times, escalation path, and whether premium support is a separate line on the quote.
• What is the renewal uplift, and is it capped? Get the cap in the contract rather than in an email.
• What notice is required for non-renewal, and when does the window open? Put the date in a diary the week you sign.
• What happens to your data if you leave — format, completeness, and how long you have to extract it.
• Written notice of fee changes, from any vendor including us.

Ask those five on a reference call rather than reading them off a review site, and you will learn more in twenty minutes than any rating will tell you.

Date the review, then ask which partner did the implementation

Two things distort ERP reviews specifically, and neither is visible in a star rating.

First, the product moves. NetSuite ships releases on a regular cadence, and the review corpus stretches back years. A 2023 review sits in the same average as one written last week and describes a different interface, a different reporting engine and a different set of AI features. When you read reviews, sort by date and read only the last twelve months. Then ask the vendor what changed in the areas the older reviews complained about.

Second, and this one is underrated: a large share of what people are reviewing is not the software. ERP satisfaction is heavily determined by who implemented it. Configuration decisions made in week six by an implementation partner become the daily experience of everyone in finance for the next five years. Two companies on identical licences can have opposite experiences, and both will write about "NetSuite".

That gives you a question almost nobody asks, and it is the single most predictive one available: ask every reference which partner implemented them, what the scope was, and whether they would use the same partner again. Ask your prospective partner for references where the implementation went badly and what they changed afterwards. A firm that cannot produce one has either not done enough projects or is not being straight with you.

The same logic runs through the whole category. Reviews for large platforms routinely predate the current product, and reviews of implemented systems are partly reviews of the implementer. Date them, then verify on a call.

What no NetSuite review will tell an Amazon seller

Here is the boundary, stated flatly. NetSuite is a system of record. It faithfully records what Amazon charged you. It has no view on whether Amazon should have charged you that, and on an FBA business a meaningful share of the time it should not have.

There is also a reporting reality worth knowing before you judge any Amazon finance tool, including ours, by whether its numbers tie out. Amazon settles some fees up to fifteen days after the sale, which is why no profit-and-loss tool ever matches Seller Central to the penny and why the most recent days are always an estimate. That is our own reporting practice talking: we run at roughly 95% accuracy on the newest days and reconcile afterwards, and a variance of a few hundred dollars on a healthy account is normal rather than a bug. The discipline that prevents a monthly argument is labelling every figure as estimated or reconciled — because otherwise two people are both right and neither can prove it.

The Amazon-side work that no ledger performs: reorder timing and stockout risk, storage and aged-inventory surcharges, the removal-versus-liquidation decision on dead stock, fulfilment fee errors and dimensional-weight misclassification, shipment discrepancies and reimbursement recovery. That is what Dr. Stock does, run by Fable 5 and supervised by operators from Full Circle, a full-service Amazon management company with more than $500M in managed revenue across 100+ brands and 70+ brands live across the group today. Inventory purchasing decisions always come to a human. Orbit is included at no additional cost. We publish no price and go to a demo — Cin7 and Sellerboard both publish theirs, which is a point in their favour, not ours.

Two honest redirects. If you are working out whether the ERP is affordable at all, our NetSuite pricing breakdown sets out how the quote is built and what two named partners publish as implementation ranges. And if your shortlist is really mid-market inventory software rather than full ERP, the Cin7 review covers a competitor that publishes exact tier prices.