Amazon PPC software reviews: the score tells you almost nothing, the count tells you a lot

Three kinds of review, three kinds of bias

Search this category and the results sort neatly into three buckets. Knowing which one you are reading is most of the skill.

Affiliate roundups. Paid per signup, so the ranking correlates with commission rate more reliably than with fit. The tells are consistent: every tool scores between 4.2 and 4.8, every con is mild, and every entry ends in a discount code. They remain worth reading as feature inventories, because affiliates do read the documentation. Ignore the ordering.

Competitor content. Vendors write reviews of their rivals in order to rank for their rivals' names. The analysis is sometimes genuinely sharp, because a competitor knows exactly where the weak spot is. What you never get is the paragraph where their own product loses. This page is one of these, which is why it is going to try to earn its place by showing its working rather than by asserting a verdict.

Vendor self-reviews. The strangest bucket, and more common than you would expect once you start looking. It is not unusual to find a vendor's own blog running an eighteen-tool roundup of Amazon PPC software with that vendor listed first, and that single post doing most of the search work for the entire domain.

What is missing is reviews by people who ran a tool on a real account for six months and then had to explain the numbers to a finance team. There are very few of those, which is why the same four opinions circulate through this category unchanged for years.

So we opened every primary record on one day

Rather than argue about bias in the abstract, here is the experiment. On 21 August 2026 we pulled the live rating and review count for eleven Amazon advertising tools from the platforms that issue them — G2, Capterra and Trustpilot — rather than from any site that aggregates them. Every figure below is a score with its count attached, because a score without its count is not a fact.

• Quartile — G2 4.6 across 215; Capterra 4.5 across 94; Trustpilot 4.8 across 559.
• Teikametrics — G2 4.5 across 131.
• Perpetua — G2 4.4 across 79.
• sellerboard — G2 4.7 across 57.
• SellerApp — G2 4.7 across 53.
• Pacvue — G2 4.5 across 23.
• AiHello — G2 4.6 across 14; Capterra 4.2 across 13; Trustpilot 3.3 across 9.
• Ad Badger — G2 4.5 across 12.
• BidX — G2 4.8 across 3; Capterra 5.0 across 9; Trustpilot 3.5 across 21.
• Sellozo — Trustpilot 4.7 across 73; its G2 and Capterra listings exist and carry no reviews at all.
• Scale Insights — a G2 listing exists carrying no reviews.

One process note that matters for anyone trying to repeat this. All three of those sites refuse ordinary automated requests, which is precisely why most articles quote whichever record they happened to reach and present it as the rating. We reached them through an interface built for it. If you cannot, ask the vendor for a screenshot of their own profile and check the count on it — that request is not rude, and a vendor who will not is telling you something.

What that spread actually shows, and it is not what the stars say

Look at the G2 column on its own. Eight rated products, and every single score sits between 4.4 and 4.7. A third of a point separates the most-loved product in the category from the least. The score does not discriminate. If you shortlist on star rating you have chosen at random with extra steps, and every affiliate roundup that ranks by score is exploiting exactly that.

Now look at the counts. They run from 3 to 559 — a range of nearly two hundredfold across products in the same category. That is where the information is, and it is the number almost nobody prints.

The three products with records on all three platforms make the point sharper still. Quartile, with 868 reviews spread across the three, lands within about three tenths of a point of itself everywhere. AiHello, with 36 across the three, spreads by more than a point. BidX, with 33 across the three, spreads by a point and a half — its highest and lowest averages cannot be reconciled at all, and any article quoting one of them as the rating has picked a side of a disagreement without mentioning a disagreement exists.

The pattern is not subtle: where counts are thin, the platforms disagree; where counts are large, they converge. Which gives you a usable rule. Below roughly fifty reviews, a rating describes who happened to write, not the product — read the text, ignore the average, and go and get references. Above a few hundred, on more than one independent platform, the average starts describing the company. Between those, treat it as one input among several.

And check whether the absence of a record means anything before you read anything into it. Sellozo's empty G2 profile and Scale Insights' empty one are not verdicts on either product; they are statements about where a self-serve company spent its review-solicitation effort. An empty profile is not a bad profile.

We made this exact error, and deleted the page rather than patch it

This is not a lesson we are passing on from a safe distance. We published a review page on this site that recommended a product on the strength of a headline score taken from a review-aggregation site — a confident figure, attributed to a named source, with a review count attached. It looked like exactly the sort of evidence this page is asking you to look for.

When we finally opened the platforms that aggregate was supposedly built from, one of them read very differently and another carried no reviews at all. The aggregate could not be reconciled with either primary record. Because the rating was the argument on that page, there was no honest patch available: correcting the number inverted the recommendation, the summary, the verdict and two FAQ answers. The page came down.

The mechanism is worth naming precisely, because it is not really about reviews. Three separate times now we have taken a third party's confident number and built a paragraph on it without opening the record that issues it — once with a competitor's contract terms read from a superseded cached version, once with a vendor's discount read off the wrong billing tab, and once with this. The surfaces were different every time. The mistake was identical.

So the rule we run on ourselves is narrower than a rule about reviews: if a number decides what a page recommends, open the source that issues it, not the page that quotes it. A checklist can confirm a figure has a source attached. Nothing can confirm the source is right except going there. Apply that to us as readily as to the affiliate roundups — including to every figure on this page, all of which name their platform and their date so you can.

The dates are the second tell, and almost nothing carries one

Check any Amazon PPC review for a publication date. Most do not carry one, and many that do carry a year in the title that was updated without the content being touched.

Here is a live example you can check yourself. Search for Adtomic pricing and the top results will tell you it costs $199 a month as an add-on, that a 2% fee applies above $10,000 of monthly spend, and that you can pair it with a $39 Starter plan. We opened Helium 10's own pricing page on 21 August 2026. There is no Starter plan. There is no $199. There is no $10,000 threshold. The word Adtomic does not appear on that page at all — the product is now Helium 10 Ads, the dedicated Adtomic pricing URL returns a 404, and the current footnote reads that Diamond customers incur a 2% management fee on PPC spend managed through Helium 10 Ads. The tool did not get worse. The review did.

Worth adding, because it is the sort of thing we would rather other people did not have to catch us on: our own earlier version of this argument implied the change made the tool more expensive. It mostly did not. Two per cent of $10,000 is $200, which lands within a dollar of the retired $199 add-on, so at the spend level where most people met that figure the cash difference is nearly nothing. The rename is a dating problem, not a pricing scandal, and killing our own gotcha is worth more than landing it.

The same trap catches vendors on their own domains. AiHello's current page prices its Smart tier at $199 per marketplace per month plus 2% of ad spend, while older AiHello landing pages remain reachable carrying a completely different ladder — flat tiers banded by spend, no per-marketplace fee, no percentage. Nothing was taken down; the new page went up alongside the old ones. Those legacy pages are where most of the cheaper AiHello figures in comparison posts come from, and a writer who lands on one has no signal it is superseded.

A price in a review is a rumour. Open the vendor's live pricing page, and if you land on something that looks orphaned, go back to the navigation and find the page the site itself links to. If there is no price at all — Pacvue's and Intentwise's pricing paths both return a 404, and Quartile's lands you back on its home page — that is the finding, and it is a useful one stated neutrally.

What the reviews consistently get right, and the one test they cannot run

Cynicism has limits. Some judgements recur often enough, from enough independent directions, to be treated as reliable even though no single source is trustworthy on its own:

• Helium 10's research tools are the best in the category. Nobody argues this, including its competitors. The advertising module is the weaker half.
• Perpetua's interface is the nicest to live in, consistently, including from people who left it.
• Scale Insights gives the most granular control and asks the most of you in return. Reviewers who like it enjoy building rule sets. Reviewers who dislike it do not.
• sellerboard is the best value in profit analytics and does not pretend to be an ads platform.
• Enterprise tools have enterprise onboarding. Pacvue and Quartile reviews agree on capability and agree on the implementation cost in hours.
• Rules engines create work. Almost every long-term review mentions maintaining the rules. Almost no vendor page does.

Then there is a question no reviewer can answer for you, because the answer lives in your catalogue: how much of last quarter's ad spend went to search terms that produced no orders at all, and can this tool find it and act on it without you writing the query?

We ran that question across our whole book rather than on one account, because one account can always be called an outlier. In Orbit, over 47 managed brands and 3,086,624 search terms that took a click from 1 May 2026 onward, the answer was 48.5% of all search spend — $4,962,963 of $10,243,379 — landing on terms that returned no orders, and 0.9% of 891,585 terms producing 80% of the sales. Most tools can show you your own version of that once you know to ask. Very few volunteer it, and no review will tell you whether yours will, because no reviewer ran it on your catalogue.

One more question worth asking of any written assessment, including the ones vendors send you. A client once returned a strategy document to us with the observation that it read like an audit nobody had read end to end, and that the issues in it should have been caught in the account weeks earlier rather than written up as findings. That is one enterprise client and one document, and both halves of it were correct. It changed how we work, and it left behind a question a buyer can use on anybody: ask who read this before it was sent, and what they changed.

How to choose, and where we fit

Sort by the constraint that is actually binding, not by star rating.

• Money is tight and you mostly need to see profit. sellerboard. Publishes its full price, priced on orders rather than spend, one-month trial, no card.
• You want exact control and enjoy building rules. Scale Insights. Per-ASIN pricing, 30-day trial, no card.
• You already pay for Helium 10. Use Helium 10 Ads on Diamond before buying anything else, and price the 2% managed-spend fee properly first — above modest spend it is larger than the subscription.
• Walmart matters as much as Amazon. Teikametrics, which publishes both halves of its pricing and both billing bases.
• You want it hands-off and the interface matters. Perpetua. Get the unquantified percentage in writing before you sign.
• You have a genuine enterprise catalogue. Pacvue. Budget for procurement and for implementation hours.

And the honest version of our own entry, since we are one of the interested parties described at the top of this page. Dr. PPC is not software you review, it is work you buy. It reads the whole account, writes a strategy per product against that brand's real unit economics, and proposes each change with its evidence, a measurement plan and a rollback trigger — you decide whether changes wait for a click, run with the big ones queued, or run inside agreed guardrails. Inventory, pricing, launches and creative always come to a person. $300/month plus 3% of ad spend, capped, month to month, first 30 days free, with Orbit included. Full Circle, a full-service Amazon management company with $500M+ in managed revenue across 100+ brands, operates it — and if what you want is that whole team on listings, catalogue and supply as well as media, Full Circle is the purchase rather than any product.

Two more redirects rather than a link list. If the money is leaking through fees, storage and stockouts rather than through bids, Dr. Stock is the right tool and no review on this keyword covers it. If sponsored ads are already efficient and growth has stalled anyway, the question is reach — Dr. DSP as a product, or reMKTR to buy the same capability as a service. We have no public review record at all, incidentally, which on the evidence of this page makes us the hardest vendor here to check. That is a gap on our side, not a virtue.