Frase Review: What the Evidence Couldn’t Settle

What Frase is, and what the vendor claims

Frase is subscription software that turns a keyword into a research document. Feed it a term, and it reads the top results for that term, pulls out how those pages are structured and which questions they answer, and assembles a brief plus an outline you can hand to a writer. Paste a draft back in and a 0-to-100 content score tells you how much of that recommended coverage you have hit. A bundled AI writer can generate the prose inside the same workspace. What makes Frase a research tool rather than a scorer or a generator is where the value sits: in compressing the reading-and-planning stage, not in the number it prints or the draft it writes.

Frase’s own marketing now reaches past that, and it reaches in a direction worth flagging. In an April 2026 post on its blog, the company argues that traditional content-optimization scoring is losing relevance as AI-powered search grows, cites an unlinked forum thread of people questioning tools in its own category, and uses that framing to introduce newer features that track whether your content gets surfaced by AI answer engines (EV-12). That is a vendor repositioning its own product, with no independent data attached, so I read it as a marketing claim and hold it there. The honest summary of Frase’s pitch is that the research-and-brief engine is the real thing people buy, the content score is a coverage guide the company itself is quietly walking back, and the writer is the part the evidence treats hardest. The sections that follow take those in order, starting where every buyer starts.

Frase pricing after the rebuild, and what a brief really costs

Every price here traces to one dated capture of frase.io’s pricing page on July 20, 2026 (EV-01), not to the older tables that still circulate and openly disagree with one another. Frase lists four plans, and the meter that decides value is the monthly “articles” allowance, where one article is a full brief-and-optimize workflow on a single piece. Two other meters sit alongside it: audit pages, for re-scoring content you have already published, and AI generations, which the bundled writer spends.

That last column is the number every rival review skips. Divide each plan by the articles it includes and Starter works out near $4.90 an article, Professional drops to about $3.23 as its allowance grows faster than its price, and Scale lands near $2.99 (EV-01). At the point of use that per-article rate is reasonable, and a 12-month benchmark run inside live client accounts called Frase the best-value pick in its category on that basis (EV-11). Read the ceiling, though, not just the rate: Starter’s ten articles a month is a real cap. If you publish weekly across two sites you clear it before mid-month, and the fix is a jump to the $129 Professional tier, not a stretch of the cheaper plan. Price the tier your actual output needs, not the sticker that looks friendliest.

The free trial, and a plan history that keeps changing

If you have read an older Frase review, its pricing is probably wrong now, and not by a little. A practitioner testing Frase in April 2026 reported a $15-a-month entry plan (EV-04). Two other dated sources from the same stretch of 2026, tested.media in April and a named 6-month user in June, both describe a “Basic” tier at $45 a month (EV-11, EV-08). A hands-on test from August 2025 was working off a “Team Plan” at $115 a month (EV-09). The page I captured in July 2026 shows none of those names and a Starter tier at $49 (EV-01). So within roughly a year Frase’s entry point has been reported at $15, $45, $49 and $115, under at least three different naming schemes. The takeaway is narrow and useful: trust the live pricing page over any review’s number, this one included, and confirm the figure yourself before you pay. On the trial specifically, most coverage describes a short free trial capped at a handful of articles, but I could not lock current trial or refund terms to a dated vendor capture this run, so I will not state them as fact. Check them on the live site first.

The lifetime deal: real history, stale offers

Frase ran an AppSumo lifetime deal years ago; the listing dates to a 2020 Black Friday promotion, and buyers from that run paid once and expected to keep access (EV-07). It is long expired as a live offer, so a page advertising a Frase lifetime deal today is selling you a memory. The more useful part is what happened to the people who bought the real one. Of the 166 reviews on that AppSumo page as of July 20, 2026, three, dated across February to April 2026, report their lifetime accounts losing paid access during a relaunch the reviewers call “V2,” one noting the company closed its user group before announcing the change (EV-07). That is three dated accounts out of 166, from a 2020 early-adopter cohort that is a separate purchase channel from today’s subscribers, so it is a flag, not a rate. But it is a pointed one: if the lifetime deal is what draws you, the dated evidence shows even its original buyers fighting to hold what they were sold.

Content briefs: the feature every source agrees on

This is where Frase earns its keep, and it is the one point the whole record lines up on. The research engine does in about a minute what a writer would otherwise spend an afternoon on: it reads the pages already ranking for your keyword, extracts their heading structure, and pulls together the questions real searchers are asking, then packages all of it into a brief and outline. In a disclosed 12-month benchmark across live client accounts, tested.media clocked Frase’s brief generator as the fastest in the category, at 30 to 60 seconds a topic (EV-11). A practitioner running identical outlines across five sites and 63 keywords over 25 days credited Frase with pulling together everything the top-ranking pages cover, generating briefs that would take a person roughly two hours by hand (EV-04). A rival tool’s own six-month comparison, which had every incentive to knock Frase, still conceded that it automates the research entirely, reading the top results and lifting their structure and questions (EV-03). And a named reviewer six months and more than 100 documents in singled out the research-backed question generation as the standout feature of the whole product (EV-08).

Read those four together and the pattern is clear: independent testers, a rival vendor and a long-term paying user, spanning 2026, all agree that the research-and-brief stage is what Frase is genuinely good at. If your bottleneck is the blank-page research grind, the afternoon-compressed-to-a-minute trade is real, and it is the reason to look at Frase at all. The catch, which the same sources are just as clear about, is that a brief is not a finished article, and Frase’s attempt to close that last gap is the weakest thing it does.

The bundled AI writer, and its honest limits

Frase does not stop at the brief. Its bundled AI writer, gated on the Pro-and-up tiers, will generate the draft too. The dated evidence on that writer is consistent, and it is not flattering. In a hands-on test from August 2025, a reviewer generated sample content and called it solid but a little uninspiring, a decent starting point rather than a finished piece (EV-09). The 6-month user who praised the research engine so highly turned flat on the writer in the same review, saying the AI assistant “is not the best” (EV-08). And the sharpest data point is a published one: in a one-month test, a reviewer put an article drafted with Frase up against a manually written piece on a similar topic and tracked both. The Frase-drafted article ended the month ranking for 5 keywords at an average position of 23, while the human-written one ranked for 17 at the same average position and pulled 30 clicks (EV-10). The same review logged unfinished paragraphs, occasional bizarre wordings and missing external references in the AI draft.

Two honesty notes keep that from becoming a beating. The ranking test is a single article over a single month on a similar rather than identical topic, and its underlying data originates in a 2021 test refreshed in 2026, so treat the 5-versus-17 gap as one dated account, not a controlled result. And no AI writer is exempt from the field’s core failure: these models produce fluent text that can be quietly, factually wrong. A 2025 peer-reviewed study of user-reported AI errors found factual incorrectness the single most common complaint type across the tools it examined ( Scientific Reports, 2025). That is a field-wide finding, not a Frase-specific rate, and no such measured test of Frase exists in this record. But it is the reason a Frase draft needs a fact-check pass, not just a style edit, before it goes live. The practical read is simple: use Frase’s writer for a rough first draft you then rewrite and verify, and lean on the brief engine for the part it is actually built to do. If drafting itself is your real bottleneck, a tool built for generation beats an optimizer’s bolt-on, and the Surfer SEO review and NeuronWriter review both land on the same split for their own bundled writers.

The content score, and the pivot Frase is making away from it

Frase’s headline number is a 0-to-100 content score that climbs as your draft covers more of what the top pages cover. It is tempting to read a high score as a ranking forecast. The dated evidence says do not. In a disclosed study from May 2025, an independent team ran 20 keywords through five optimization tools, Frase among them, and used rank-correlation math to test how closely each tool’s score tracked the pages’ real Google positions. Frase landed in the weaker-correlated group, and the study illustrated the deeper problem with a screenshot of a document scoring a full 100 after suggested keywords were stuffed in with no real draft behind them (EV-02). A separate six-month practitioner test put it more bluntly still, describing the score-to-ranking link as feeling like essentially noise (EV-03). A 12-month benchmark was kinder but still rated Frase’s scoring accuracy behind a rival’s (EV-11). So the honest use of the score is as a coverage checklist against your competitors, not a promise: a draft sitting at 90 that will not rank is the tool working as designed, because the number measures term coverage and not the many other things that decide a result.

Here is the part that makes this section unusual, and the reason it belongs under a “what the evidence couldn’t settle” title. Frase itself appears to agree the score is not the future. Its own April 2026 blog post argues content-optimization scoring is losing relevance and pivots toward features that track whether your content gets cited by AI answer engines, the AI-search-visibility angle competitors are also chasing (EV-12). The trouble is that no independent test of any such Frase feature exists in this record, and a practitioner testing the tool that same month found it still lacked AI-visibility tracking altogether (EV-04). So the score you are paying for is one the vendor is stepping back from, and the thing it is stepping toward is unproven and, on the dated evidence, was not fully shipped when it was being marketed. That gap is genuinely open, and I am not going to close it with a guess.

What dated reviews, ratings, and support reports show

With no Reddit or Quora thread reachable this run, community sentiment rests on two rating platforms I could open. On Capterra, Frase averaged 4.8 out of 5 across 335 reviews on the day I captured the profile, July 20, 2026 (EV-05, EV-06): a platform sample on a site where vendors can invite reviewers, so I weigh it as one signal, not a verdict, and the dates inside it matter more than the average. A director in e-learning, writing in July 2025, said the team got productive with most features almost immediately (EV-06), the ease-of-adoption note that recurs in the tool’s favor. A 2022 reviewer praised the research-and-comparison workflow but flagged the pricing as a stretch for smaller businesses (EV-06). Pulling the other way, a marketing CEO in October 2023 found Frase buggy and glitchy next to competitors, with non-responsive buttons and an AI writer that malfunctioned, and reported a refund request declined shortly after the trial ended (EV-05). That last review is nearly three years old, so I flag it as historical: the product has plausibly moved on, and it is one dated experience among 335, not a pattern.

Support is the thread worth pulling across those accounts. The declined-refund complaint from 2023 (EV-05), the disappointing support experience logged in the one-month ranking test (EV-10), and the AppSumo cluster describing a poorly communicated 2026 account migration (EV-07) are three separate, dated low points on service rather than a measured support benchmark. None generalizes to a rate, but together they are enough that I would test Frase’s responsiveness during a trial before committing at the Professional or Scale tier. On the Reddit gap specifically: the only pointer I found was Frase’s own blog paraphrasing an unlinked forum thread (EV-12), which is the vendor characterizing a conversation I cannot open, so I do not treat it as community evidence at all. If a genuine thread surfaces on a later pass, it goes into the record then, dated and linked, not before.

Where this review stops short. A few limits sit under everything above. Rating platforms attract complaints more readily than quiet satisfaction, so the frictions here likely weigh heavier than an average subscriber feels, and the flip side is that a 4.8 Capterra average across 335 reviews (EV-05) reflects those 335 raters, not a contentment level for the whole paying base. No Reddit or Quora primary opened this run, and with G2 and Trustpilot both blocking the fetch, whatever sentiment sits only on those platforms is missing here rather than weighed, and the secondhand rating figures that circulate for them go uncited because I could not confirm them myself. The AI-writer verdict rests on dated first-person accounts and one small, imperfectly controlled ranking test (EV-10), not a measured accuracy study, because none specific to Frase exists. And every price, allowance and quote was true of the page I captured on July 20, 2026, which can change the day after.

Who Frase is worth it for

Frase’s real question is not “is it good” but “good at the part you need,” and the evidence sorts three buyers cleanly.

A solo blogger writing to their own site is a decent fit if research is the grind you hate. Starter’s $49 a month and its roughly $4.90-an-article rate (EV-01) buy the brief engine that testers rate the category’s fastest (EV-11), which is genuinely worth it if planning is what stalls you. Go in knowing two things: the ten-article cap is real, and you are buying the research, not the writer, so budget your own editing time on anything the AI drafts (EV-09, EV-10). If a pure coverage check is all you want, a cheaper scorer covers that alone for less.

A freelance writer billing per deliverable gets the strongest return from Frase, with one specific caution. The brief engine’s two-hours-to-one-minute research trade (EV-04) is billable time you keep, and a clean, sourced brief is a real thing to put in front of a client. The risk sits in the writer: a stiff, generic first draft shipped to a client is a different stake than a rough draft on your own blog, and the dated accounts say Frase’s writer produces exactly that kind of first pass (EV-08, EV-09). Point Frase at the research and outline, do the drafting and fact-checking yourself, and it earns its place in a freelance stack.

An agency or content team is where the ceilings bite and the overlap question matters. Professional’s 40 articles or Scale’s 100 (EV-01) have to be priced against your real publishing calendar, not the cheapest tier that looks like it fits, and a team already running Surfer or NeuronWriter should weigh the redundancy honestly before paying for a second tool in the same lane. The bigger agency-specific gap is lifecycle: a six-month practitioner test found Frase has no post-publish monitoring, no Search Console link, and no alert when a live page starts slipping (EV-03), so if you need one tool to carry an article from research through decay-watching, Frase covers the front half and hands you off for the rest. And if seat-based lifetime access is part of your plan, read the AppSumo migration reports first (EV-07).

Verdict: Frase versus Surfer, NeuronWriter, and what stays open

Is Frase legit? Yes, plainly: a real product from an established company, with a research engine that independent testers, a rival vendor and long-term users all rate highly. That is why it lands at 3.5 out of 5 rather than higher or lower. The briefs are a genuine strength, the writer and the score are genuine weaknesses, and the price sits in between, fair per article but capped tighter than the sticker suggests.

The two tools most buyers weigh Frase against are the other members of its trio, and the split is about jobs, not winners. Surfer SEO is the premium end: its Content Editor is the piece its own users pay for, and it is built around scoring and refining an existing draft rather than researching one from scratch. NeuronWriter is the budget scorer, cheaper than both if a coverage check is really all you want, and in the one independent five-tool test that measured all three, its content score lined up with rankings better than Frase’s did (EV-02). Frase’s answer to both is the research and brief work it does faster than either (EV-11). The recurring one-liner across independent 2026 comparisons holds up: Frase owns research and briefs, Surfer owns scoring and editing, NeuronWriter owns the low price. Each pairing deserves a full side-by-side rather than a paragraph buried here, so I am keeping this a verdict on Frase alone; for now the NeuronWriter vs Surfer SEO comparison covers the research-versus-scoring axis Frase sits on. Named alternatives outside the trio, both surfaced by dated tests here, are Clearscope, which one benchmark rated more accurate on scoring (EV-11), and MarketMuse, which a 6-month user found sharper on keyword suggestions (EV-08). If your shortlist instead includes an all-round marketing writer, that is a different category from an SEO-research tool, and the Jasper review covers that job.

And to cash the title honestly, here is what this review could not settle. Whether Frase’s new AI-search-visibility push actually surfaces your content in AI answers: unproven, and on one dated test not yet shipped when it was being marketed (EV-04, EV-12). Whether lifetime-deal buyers keep the access they paid for, with dated 2026 accounts saying some did not (EV-07). What Frase actually costs a year from now, given an entry price four dated sources put at $15, $45, $49 and $115 inside a single year (EV-01, EV-04, EV-08, EV-09, EV-11). And whether a high content score does anything for rankings, which the evidence says is weak and which the vendor itself is walking away from (EV-02, EV-12). A generative summary will flatten all four into a single number; the review exists because they do not flatten.

Buy Frase if the research grind is your bottleneck and you want briefs and outlines automated fast. Look cheaper or elsewhere if a coverage check is all you need, or if you are counting on the bundled writer for publish-ready drafts. Don’t expect the content score to guarantee rankings, or the AI-search-visibility features to be proven yet. For the wider field, see how the writing tools we cover compare, and the dated reports behind this and every review live in our logged evidence record.

FAQ