Software reviews that get the pricing right
A one-person review desk covering SaaS, AI, hosting and marketing tools. Every price here is checked against the vendor's own pricing page and stamped with the date it was verified — not copied from a press release or a competitor's blog.
When a vendor changes its terms or the evidence shifts, the review changes too, including the score.
What changed when we re-checked
Most review sites never revisit a score. These are tools we rated highly and then marked down — with the reason, and a link to the current review.
Emergent
Credits are charged for failed attempts as readily as successful ones, and the $200 tier works out 33% more expensive per credit than the $20 one.
StealthGPT
Independent testing still flagged its humanised output, and the published pricing didn't reconcile across the vendor's own pages.
Plesk
A roughly 26% price rise in January, and the entry plan now ships only WordPress Toolkit Lite rather than the full version.
Reclaim.ai
Per-seat prices rose across every tier, the Lite plan tightened to roughly one habit, and AI actions became quota-metered.
Scrivener
Still the best tool for long manuscripts, but the half-price route closed when NaNoWriMo shut down, and the iOS app hasn't moved in a long while.
No score here was changed at a vendor's request. Several of these tools pay commission on referrals, which is why the reasons are stated plainly enough to check.
How a review gets made
Three habits do most of the work. None of them are glamorous.
Read the vendor, not the blogs
Pricing comes off the vendor's own page and help documentation. Most tool comparisons that rank well are published by a company selling a competing product, so they don't count as sources here.
Publish the contradictions
Vendors routinely disagree with themselves — a pricing page saying one allowance, the help centre saying another. When that happens the conflict goes in the review instead of being quietly resolved in the vendor's favour.
Separate tested from reported
Every review says which claims come from hands-on use and which come from documentation or other people's testing. No invented benchmarks, and no first-hand claims about tools that weren't bought.
Recent reviews
Each one carries the date its pricing was last verified.
Snov.io: finder, verifier and sender on one flat plan
Unlimited seats at $99 against roughly $99 per user elsewhere. LinkedIn automation is a separate charge the pricing page understates.
Read the review
Analytics
2026
Databox: a real alternative to Looker Studio?
Pulls Google Ads, HubSpot and Stripe into one dashboard. The free tier is tighter than it looks once you count data sources.
Read the review
Agency software
2026
GoHighLevel: can one platform replace an agency stack?
The white-label layer is the product. Everything else is adequate rather than best-in-class, and that trade is the whole decision.
Read the review
Ad creative
2026
QuickAds: generating ad creative without a designer
Good enough for volume testing, weaker for brand work. What the output actually looks like before you commit budget.
Read the review
Productivity
2026
Wispr Flow: better than Apple Dictation?
System-wide dictation that formats as you speak. Where it beats the built-in option, and where the subscription stops being worth it.
Read the review
Health tech
2026
Superpower: is the annual blood panel worth it?
What the membership actually includes, which markers are covered, and the costs that sit outside the headline price.
Read the review
What this site is, and what it isn't
Worth being specific, because plenty of review sites aren't.
What happens here
Always- Tools get bought or trialled where that's practical, and the review says so when they weren't
- Prices are read off the vendor's own page and dated, then re-checked when the vendor changes something
- Contradictions in a vendor's own documentation get published rather than smoothed over
- Scores get lowered when the facts stop supporting them, commission or not
What doesn't
Never- Hands-on testing that didn't happen, described as though it did
- Precise benchmark figures with no stated method behind them
- Reviews left sitting on last year's pricing
- Any pretence of being a team or an agency — this is one person, and the reviews are better for being read that way
Want your software reviewed?
Send the product, a test account and the pricing page. You'll get a full review built the same way as everything else here — verified pricing, the limitations stated, and a score that reflects what the tool actually does.
What's on offer is a careful, current review — not a guaranteed score, and not approval of copy written for you. If a review would be unflattering, you'll hear that before it goes up.