Prala Radar
Short briefs on what is changing in affiliate marketing and the tools around it. One page, updated as things happen, rather than a hundred thin ones.
The bulk sender rules are still catching affiliate senders out
Gmail and Yahoo's sender requirements have been in force since early 2024, and they are still the most common reason an affiliate business quietly stops reaching inboxes. Three of them do most of the damage.
Authentication. SPF, DKIM and a DMARC record on the sending domain. Missing DMARC is the single most frequent gap.
One-click unsubscribe. A real
List-Unsubscribeheader, not only a link in the footer.A spam complaint rate under 0.3%. This is the one that bites people who bought a list, because complaints from strangers are charged against the domain you also use for people who asked to hear from you.
What to do: check your sending domain against the guidelines this week. All three are configuration, not strategy, and none of them will announce themselves when they fail — the mail simply stops arriving.
INP is the responsiveness metric now, and landing pages feel it
Interaction to Next Paint replaced First Input Delay as a Core Web Vital in March 2024, and the change matters more than a metric swap sounds like it should.
FID measured only the delay before the browser started responding to the first interaction. INP measures the full time from an interaction to the next frame painted, across the whole visit. Pages that scored well on FID because they responded quickly and then did something slow now score honestly.
For a landing page, the usual culprits are the same handful of things: chat widgets, consent banners with heavy scripts, and analytics tags that run work on click.
What to do: run your highest-traffic landing page through PageSpeed Insights and look specifically at INP on mobile. If it is over 200ms, the fix is almost always removing a third-party script rather than optimising your own.
"Scaled content abuse" is the policy that matters for AI writing
Google's spam policies do not prohibit AI-written content. They prohibit producing content at scale primarily to manipulate rankings rather than to help people — which is a description of a method and a motive, not of a tool.
The practical distinction for an affiliate business: a model helping you write a piece you would have written anyway is fine. Generating forty near-identical pages targeting keyword variants is the thing the policy names, and it is enforced whether a human or a model produced them.
What to do: apply one test to anything you publish. Could a reader get this answer from the first three results already ranking? If yes, publishing it adds nothing, and the policy question is academic because it will not rank either way.
Affiliate disclosure rules apply to every surface, including DMs
The FTC's endorsement guidance is old enough that people assume they know it, and the details most often missed are the ones about placement.
A disclosure has to be clear and conspicuous, close to the recommendation itself rather than in a bio, a footer, or behind a "more" link. It applies in a direct message and in an email as much as on a public post. And on video, a spoken disclosure counts where an on-screen one might be missed.
The guidance is written in terms of what an ordinary reader would notice, which means "technically present" is not the standard.
What to do: look at your last ten pieces of promotional contact — including the ones sent privately — and check the disclosure is in the message rather than adjacent to it.
FAQ rich results are gone, and the markup is still worth keeping
Google stopped showing the FAQ rich result entirely, announced in May 2026, and removed the supporting documentation the following month. The feature had already been restricted to a narrow set of authoritative sites since 2023; this is the end of it rather than a further narrowing.
The predictable response is to strip
FAQPagemarkup out. That is the wrong conclusion. The rich result was one consumer of the markup. The structured question-and-answer pairs are still the cleanest machine-readable statement of what a page answers, they are still valid schema.org, and Google is not the only thing reading a page any more.What to do: keep the markup and stop counting on the SERP real estate. If your FAQ section existed to win the expanding box rather than to answer questions people actually ask, that is the part worth rewriting.
Google says llms.txt does nothing for Search, which is not an argument against it
Google added a clarification in June 2026 that is worth reading precisely. Its guide on generative AI features now states that you do not need
llms.txtfiles or other "special" markup to appear in Google Search, including its AI surfaces, because Search does not use them — and that maintaining one will neither help nor harm your rankings.That is a statement about Google, not about the file. Google is explicit that keeping one for other services and systems is completely fine, which is the actual reason to have it: the assistants people increasingly ask for recommendations are not Google, and several of them do read it.
What to do: if you have an
llms.txt, keep it and stop counting it as SEO. If you were about to build one in the hope of ranking, build something else. The case for it is agent readability, and that case stands on its own.
Run all of this in one place
Prala is the AI-native CRM built for affiliate marketers: leads, pipeline, follow-up, bookings and content in one hub. Fourteen days free, no card.