What "verified pricing" actually means when you do it honestly

There’s a phrase that shows up on a lot of affiliate review sites: “Pricing verified [month] [year].” Sometimes it’s tucked into the footer. Sometimes it’s a small badge next to the price card. The implication is that someone went and checked, recently, and the number is current.

Here’s what I’ve learned in the last few months: that phrase, on most sites, means almost nothing. Sometimes the date is auto-generated by a plugin that updates whenever the page is touched, regardless of whether anyone checked anything. Sometimes the verification was real but happened in a quarterly batch where one person opened thirty pricing pages in a single afternoon. Sometimes it’s just decorative — there’s no process behind it at all.

I’ve been trying to do the version where it actually means something. It is significantly harder than I expected. It’s also an interesting window into why the affiliate review layer drifts the way it does.

When I’m building or updating a review on AI Freands, I open the app’s official pricing page in a browser. Not a screenshot, not a cached copy, not the price quoted in a competitor’s review article. The live page on the company’s domain.

I write down every tier and every number. Free, basic, premium, whatever they call them. I note what’s included on each, what’s locked behind which paywall, and what features the marketing copy claims versus what users actually report seeing. If there’s a regional pricing difference — which there often is for AI products — I check at least the US, UK, and EU prices. I write the date on every check.

If the app has a free trial, I look at how it actually ends. Some “free trials” auto-renew aggressively. Some require credit card upfront. Some don’t actually unlock the features the marketing implies. These details belong in the review, but they only exist in the review if someone walked through the trial flow.

If the app has hidden costs — image generation credits, voice minutes, premium character unlocks — those go in too. The base subscription price is rarely the actual cost of using the product the way the marketing implies.

This whole process, for one app, takes about ninety minutes if I’m being thorough. For thirty-plus apps, that’s roughly forty hours of pricing work alone, before any actual reviewing of features or testing the product. Now multiply that by every refresh cycle. The economics of doing this right are brutal, which is exactly why most sites don’t.

The most consistent pattern is that pricing changes more often than the affiliate layer admits. AI products are in a phase where companies are still figuring out their economics. They raise prices when they realize their cost structure is wrong. They drop prices when a competitor undercuts them. They add tiers, remove tiers, rename tiers. They quietly adjust what’s included on the free plan when their conversion math changes.

A few specific patterns I keep running into:

Quiet price increases. A platform updates its pricing page, raises the basic tier by two or three dollars, and doesn’t announce anything. Existing affiliate articles still cite the old number. New users hit the pricing page and see something different from what brought them there. Refund-friction is high enough that most just pay.

Removed tiers. A platform sunsets a low-cost entry tier and pushes everyone to a higher one. The affiliate articles still describe the old tier as the “starting price.” When users go to sign up, the option doesn’t exist. They pay more or leave.

Region-specific pricing that articles don’t mention. A platform charges different rates in different markets, sometimes with a significant spread between regions. An article quoting “$9.99/month” might be technically correct for the US and dramatically wrong for users elsewhere. Most articles don’t bother saying which region the quoted price applies to.

Trials that aren’t trials. “Free 7-day trial” sometimes means a 7-day window in which you can cancel before being charged for a year up front. The framing in the article is “free trial.” The framing on the pricing page is “annual subscription with 7-day refund window.” These are not the same product.

I’m not naming specific apps because the patterns are universal across the category, not specific to one brand. But every cycle I check, I find at least one of these on at least one platform I cover.

There’s a reason the cycle of “open pricing page, copy number, paste into article” has won across most of the affiliate review layer: it’s the only way to keep up with the volume of content that ranks on Google.

If you do pricing verification the slow way, you can maintain maybe forty review articles. If you do it the fast way, you can maintain four hundred. Google rewards quantity in a lot of niches, especially commercial-intent ones. The site that has more pages on more keywords captures more search clicks. The site that has more clicks captures more affiliate revenue. The site that captures more revenue can afford more writers and more software and more SEO consultants. The arms race compounds.

The slow way doesn’t compound the same way. It compounds in trust, but trust is a long-term asset and Google’s ranking signals don’t always reward it directly. So the publishing layer of an entire category can drift, over years, into a state where every site is wrong in the same predictable ways, and the user gets caught in the middle.

The interesting thing is that the affiliate networks themselves know this. Most companies running affiliate programs would prefer their partners describe the product accurately, because incorrect descriptions cause refund requests, support tickets, and brand damage. But the program incentive is volume, not accuracy. So you get what you incentivize.

I can’t promise that AI Freands will have perfect pricing on every app every day. The category moves too fast and I’m one person doing this alongside writing and other site work.

What I can promise is that:

When I write or update a review, the pricing in it was checked on the official pricing page, not pulled from another article. The date in the verification footer is the date the check actually happened. When pricing changes between cycles, the article gets a banner noting what changed, instead of the old number being silently overwritten.

When pricing on an app feels misleading — trial framing, regional spread, hidden costs — the review notes it explicitly, even if it makes me look less optimistic about the app than the affiliate marketing wants me to be.

When I cycle back through the catalog and find that I’m wrong about something, I publish the correction with a date stamp on it. Not as a quiet edit. As a noticeable note.

This is a low bar. I think we used to clear it routinely as an industry, before scale broke the economics. I’m not interested in pretending the bar is higher than that. I’m just interested in actually clearing it.

Affiliate publishing has a credibility problem that nobody inside it really talks about. We optimize for traffic, the traffic optimizes for content velocity, and the content velocity has a maximum quality ceiling that’s lower than what most readers assume.

The result is a review layer that produces a lot of words and not very much information. A reader can spend forty minutes reading “Best AI Companion Apps of 2026” articles across five different sites and end up less informed than when they started, because the same outdated facts get repeated five times in slightly different prose.

I don’t think the answer is to write less. I think the answer is to be honest about how the cycle of checking and updating actually works, what it costs, and what corners get cut when it’s done at the volume the industry runs on.

That’s the version of affiliate publishing I’d like to see more of. It’s the version I’m trying to build at AI Freands. The verification practice is one piece of that. The next post will probably be about another piece.

— Derek