Most affiliate sites recommend whatever pays the most. We sort the list the other way around: commission rate is the last thing we look at, after a tool has already earned its place for reasons that have nothing to do with what we get paid. That ordering costs us money on specific tools, on purpose. Here is the actual process a tool goes through before it gets a link, and the reasons we pull tools back off the list later.
The filters a tool has to pass, in order
Before a tool gets a /go/ redirect and a spot in an article footer, it runs through four checks. They are sequential — fail one and it never reaches the next, no matter how good the payout is.
We have to have actually used it. Not opened a demo, not read the docs — used it on real work long enough to hit something annoying. Every tool we link to has a paragraph in our inventory that names a specific workflow we ran through it: importing a list, shipping a client site, wiring up a first sponsorship. If we can't write that paragraph from memory, the tool isn't ready. This single filter removes more candidates than the other three combined, because most of what lands in our topic queue is stuff we've only read about.
The affiliate program has to exist and pay reliably. This sounds obvious and it's where a surprising number of tools die. We've had programs close their application window mid-evaluation (Notion's PartnerStack program is sitting in exactly that state right now, which is why you won't see a Notion link from us even though we use and like it). We check the program is open, that it pays through a system we already reconcile — Reditus, PartnerStack, or a direct deal — and that the cookie window and payout terms are written down somewhere we can point to. A tool with no working program just gets mentioned by name, no link.
It has to fit the person reading the page. Pickuma is split into four audiences — developers, PMs, junior devs, and investors — each with its own section. A tool that's great for one can be noise for another. We don't drop a generic "productivity" recommendation into a page about quant tooling because the commission happens to be high. The tool has to match the audience the article was written for, or it doesn't go in that article.
The link has to be swappable. Every affiliate destination lives behind our own /go/<slug> redirect, never a raw URL pasted into the post body. That's an editorial requirement as much as a technical one: it means we can repoint, pause, or kill a link in one place the moment a program changes, without editing a single article. If a tool can only be promoted through some hardcoded tracking URL we can't wrap, that's a strike against it.
Commission rate never moves a tool up the list — it only ever breaks ties between two tools that have already passed every filter above. If you ever see us lead with a tool's payout as the reason to use it, treat that as a bug and tell us. The whole point of this ordering is that the recommendation has to survive even if the commission were zero.
What gets a tool pulled back off the list
Getting on the list is only half the job. Tools change after we recommend them, and a stale recommendation is worse than no recommendation. A few things trigger an immediate review:
Link rot. Affiliate URLs break more often than you'd expect — programs migrate platforms, slugs change, redirects start 404ing. We audit our redirects on a schedule, and a dead destination gets the tool's status flipped before a reader ever hits the broken link. The /go/ wrapper is what makes this a config change instead of a content migration.
The program shuts down or goes paused. When a program closes, we mark the link paused in our database, which makes the redirect return a 410 instead of bouncing a reader to a dead deal, and we strip the tool from the footer and tool pages so it stops being suggested. The article text may still mention the tool — we just remove the financial relationship cleanly.
Pricing or positioning shifts. If a tool we praised for a generous free tier quietly guts it, or repositions away from the audience we recommended it to, the recommendation no longer matches reality. That's an edit, sometimes a removal. We log these in the article's changelog so you can see what changed and when.
Every article that was written with AI assistance carries an explicit note saying so, and tools we have a paid relationship with are disclosed. We'd rather lose the click than have you find out later that a "recommendation" was just an ad we didn't label.
The one tool this whole process is built around
The "we have to use it" filter is the hardest one to fake, so the clearest example is the tool we run our own publishing on. Our newsletter goes out through Beehiiv, which means every claim we make about it — the import flow, the referral program, which growth channels actually convert to paid — comes from our own dashboard, not a press kit. That's exactly the bar we hold every other tool to.
We pass the same test on to you: if a tool here doesn't have a paragraph explaining what we actually did with it, assume we haven't earned the right to recommend it yet — and it won't be on the list.
Originally published at pickuma.com. Subscribe to the RSS or follow @pickuma.bsky.social for new reviews.
Top comments (0)