Three Affiliate Approvals, and the Two We Will Not Promote
We were approved for three affiliate programmes in twelve days. Two of them sell things we know nothing about, so they get no coverage. Here is the test we apply, and why refusing is the part worth writing down.
In the twelve days to 25 September we were approved for three affiliate programmes.
One sells AI product photography for online stores. One sells jewellery. One is a retail brand we had not heard of before the invitation.
Only the first will ever appear in anything we publish, and the reasoning is worth setting out — partly as a disclosure, and partly because “what do you refuse” is a more informative question about a publisher than “what do you recommend”.
The test
One question, applied before anything else:
Would we write about this if there were no commission?
Not “is it a good product”. Not “would our audience click”. Whether the subject sits inside something we actually do, and whether we would have an opinion worth reading if the money did not exist.
The AI imagery programme passes. We build online stores, product photography is a real problem for the retailers we work with, and we have a view on where those tools help and where they do not. The piece we published about that category opens with a disclosure, states plainly that we have not run a controlled test, and spends most of its length on what the tools cannot do — including the consumer-law exposure of a generated image that misrepresents goods.
The other two fail at the first question. We know nothing about jewellery retail. We would have nothing useful to say, and anything we did say would be a product description with a link in it.
Why the refusals are the interesting part
An affiliate arrangement is a financial interest in a reader’s decision. That is not disqualifying — it is how a great deal of genuinely useful review content is funded — but it changes what the reader needs from you.
What they need is to know which way your interest runs, and to see that it has limits. A publisher who covers everything they are approved for has demonstrated that the approval is the criterion. A publisher who declines most of them has demonstrated something else.
The practical version: the value of a recommendation is proportional to the things you could have recommended and did not.
★ Insight ─────────────────────────────────────
The structural risk in affiliate content is not the obvious one. It is that the programmes you are approved for start shaping what you write about — not through any decision to sell out, but because an approval creates a topic that is now easy to justify. The drift is gradual and invisible from inside, which is why the test has to be applied at approval rather than at publication. By the time you are drafting, the topic is already chosen.
─────────────────────────────────────────────────
What we disclose, and where
Three rules, and they come from the same place as the advertising rules we wrote about last week — a relationship that could influence a recommendation has to be disclosed where the recommendation is made.
At the top of the piece, not the bottom. A disclosure a reader meets after forming a view is decoration. Ours sits above the first paragraph.
Naming what we have and have not done. “We joined this programme in September and have not run a controlled test” is a materially different statement from a review, and the reader is entitled to know which one they are reading.
On the post, not only on a policy page. We also link the disclosure page from every affiliate review, which was a gap we found and fixed this week — five review posts had no link to it at all.
The uncomfortable arithmetic
The two programmes we are declining are almost certainly the more lucrative ones. Consumer retail converts; explaining what an AI imagery tool cannot do does not.
That is the trade, and it is worth naming rather than presenting this as costless principle. A publisher optimising for affiliate revenue would cover all three, and would be right to, on those terms.
We are not optimising for that. The commercial model here is that people who read this eventually hire one of our divisions, and that only works if what we publish is the same thing we would say in a meeting.
What we do with the approvals we keep
Declining two does not mean the third gets a free pass. The piece we published in that category does three things we would do again.
It opens with the disclosure, above the first paragraph, where the reader meets it before forming a view.
It states what we have not done. We have not run a controlled test against a studio shoot, so nothing in the piece reports a result we did not measure. That sentence costs credibility in the short term and is the reason the rest is believable.
It spends most of its length on limits. What the tools cannot do, where the marketplace rules forbid the output, and the consumer-law exposure of a generated image that misrepresents goods.
A piece structured that way converts poorly compared with a straightforward recommendation. It is also the only version we would be comfortable having a client read after buying on our advice.
The question for any business doing this
If you run content with affiliate links — and a great many small businesses now do, often without a policy — one question is worth answering in writing before the next approval arrives:
What would we decline?
If there is no answer, or the answer is “anything irrelevant” without a definition of relevant, then the programme list is your content strategy and nobody decided that.
Ganda Tech Services runs web, cloud, mobile and content operations for a group of Australian brands. Where we hold an affiliate relationship with a product we write about, it is disclosed at the top of the piece.