I think we might be building affiliate websites backwards.
For years, the obvious model has been to start with the products.
Build a shop. Build a comparison page. Build a product grid. Put ten things on a page and then write enough words around them to persuade Google that the page is useful.
I have done plenty of that myself over the years. In fact, I recently wrote about why affiliate marketing has become my first love again.
But with search changing, I think there is a much more interesting way to approach it now.
Do not start with the product. Start with the question.
Answer it properly.
Then, if there is a product that genuinely helps, put it inside the answer.
That sounds like a subtle difference, but I think it changes almost everything about how an affiliate site should feel.
The article should still work if nobody buys anything
This is probably the easiest test.
If I remove every affiliate link from an article, is the page still useful?
If the answer is no, I have probably built an advert rather than an answer.
This matters even more as search moves towards AEO, or Answer Engine Optimization. I have written before about why AI search has made the web feel a bit like the early days again, because the discovery layer is changing.
People are increasingly asking complete questions rather than typing a couple of keywords and working through ten search results themselves.
That means I want my websites to contain genuinely useful answers to those questions.
If somebody wants to know why their tan has stopped getting darker, what a particular book is about, how something works or which option suits a particular problem, answer that first.
The commercial opportunity comes afterwards.
Sometimes there will be a brilliant product match.
Sometimes there will not.
That is fine.
I would rather have no affiliate link than a stupid one.
We have already seen this journey happen
This is not entirely theoretical either.
I recently watched a visitor arrive on Sahara Sands from ChatGPT, read an informational page, explore related guides, search the site, find a product and eventually click through to shop.
That was fascinating to me because the AI had effectively handled the discovery.
The website then took over.
It answered the question, helped the visitor understand the subject and eventually led them towards something they might want to buy.
That feels much more interesting than trying to build another online shop and hoping somebody discovers it.
But somebody still has to find the right products
This approach creates another problem.
If the article comes first, somebody still needs to work out which products genuinely belong inside it.
Doing that manually is fine when you have 20 products.
It becomes fairly tedious when an affiliate merchant has 10,000, 20,000 or 40,000 products sitting inside an Awin data feed.
That is what I have been working on recently.
I have started connecting Awin merchant feeds properly into Control Room.
Not by blindly importing everything.
That would just move the mess from Awin into my own database.
The useful bit is being able to understand the feed first.
Control Room can read the merchant's category structure, show me what is actually inside it and tell me how many products sit underneath each branch.
So rather than saying, "Here is a massive merchant feed, good luck," I can drill down into the bits that actually make sense.
If I want jackets, I can import jackets.
If I want watches, I can import watches.
If I am working with a jewellery merchant, I can choose the relevant jewellery categories instead of importing every single product they happen to sell.
The feed gives me scale.
I still keep editorial control.
Awin is the source, not the brain
This became quite an important distinction while I was building it.
I do not want Control Room to become an Awin application.
Awin is one source of product data.
Once a product comes into Control Room, the rest of the system should not particularly care where it came from.
It could be Awin.
It could be TikTok Shop.
It could be one of my own products or services.
What matters is that Control Room knows what the product is, whether it is currently available, whether I can promote it and where the live destination is.
That gives me one product system rather than lots of disconnected little affiliate systems.
Then Affiliate Brain gets involved
This is where it becomes much more interesting than a feed importer.
Once I have written an article, I can run Analyse product opportunities.
Gemini reads the article and judges relevance.
But it does not get to decide whether a product is actually live, in stock, available to promote or has a valid affiliate destination.
That always comes from the live database.
Gemini decides whether the product makes sense in the article. Control Room decides whether the product is actually usable.
I really like that separation.
AI is very good at understanding meaning.
It is much less sensible to ask it to guess commercial facts that I already have sitting in a database.
So Affiliate Brain can look at the context of an article, compare it with the products available to that site and identify places where a commercial recommendation genuinely makes sense.
If there is nothing useful, it does not need to invent something.
This is where AEO and affiliate marketing start fitting together
This is probably the bit I find most interesting.
I do not want to write an article about tanning ingredients, books, clothing, websites or anything else as though I started with a product and desperately wrote 1,000 words around it.
The information comes first.
Then Affiliate Brain looks at what I have written and asks whether there is a relevant commercial opportunity sitting naturally inside the answer.
That feels like a much better way to combine AEO and affiliate marketing.
Answer the question first.
Monetise where the answer genuinely creates an opportunity.
Perhaps the future affiliate site does not really look like a shop at all.
Perhaps it looks like a very good information site where the shopping part appears quietly at exactly the point where it becomes useful.
And it does not have to be somebody else's product
One thing I have also started doing is putting some of my own 1J2 Services into the same product system.
That means Affiliate Brain can judge those using exactly the same idea of relevance.
For example, if I am talking about indexing problems, technical SEO, structured data, search visibility or whether a website is properly set up for answer engines, a Website Visibility Audit is a pretty obvious fit.
If I am talking about a page getting traffic but failing to turn those visitors into enquiries or sales, with weak calls to action, poor trust signals or obvious friction, that sounds much more like a Landing Page Conversion Review.
If the problem is an ageing WordPress installation, plugin problems, PHP compatibility, SSL, backups or general site health, that is what the WordPress Health Check is for.
And if you simply have one stubborn problem with your website or online systems and want somebody to get involved for an hour, there is 60 Minutes of Technical Help.
Those are very different problems, which means Affiliate Brain has something meaningful to judge rather than seeing four variations of "Adam does website stuff."
Go on, hover over one
This is where I can actually show off what I have built rather than just write about it.
Once Affiliate Brain has analysed this article, some of those product or service references can become the new hover links.
So go back and hover over one.
You should be able to see the relevant product or service without immediately being thrown off the article.
There is a preview, some useful information and then a View product option if you actually want to go any further.
I much prefer that to taking a useful sentence and turning half of it into some horrible raw affiliate URL.
The commerce is there.
It just does not have to shout.
Then 1jump2.com handles the ugly bit underneath
The final bit of this was dealing with the affiliate URLs themselves.
Awin links are useful. They are not particularly pretty.
And more importantly, I did not really want the merchant destination permanently baked into every article across every site.
So the public product links can now go through 1jump2.com instead.
The reader sees a short, clean link. Underneath, Control Room checks the product and works out the current destination at the moment somebody clicks it.
If the affiliate URL changes later, I do not have to hunt through old articles replacing links.
The public link stays clean and stable while the destination behind it can change.
It also means I do not need a second database full of duplicate short-link records for products that Control Room already knows about.
The click gets recorded too, and from there I can get into the existing visitor analytics rather than building yet another reporting system.
That is really all the reader needs to know about the plumbing.
This is really about connecting the pieces
I have written before about how the website was never really the finished product, and this is another good example of what I mean.
None of these individual things is particularly extraordinary on its own.
Awin has product feeds.
AI can understand an article.
Databases can tell me whether a product is available.
Websites can display product previews.
Redirect systems can track clicks.
The interesting bit is joining them together.
A merchant can give me a catalogue containing tens of thousands of products.
I can bring the useful parts into Control Room.
I can write an article because somebody genuinely wants the answer to a question.
Affiliate Brain can read that answer and identify products that actually belong there.
Control Room checks that those products are genuinely usable.
The article presents them as useful hoverable recommendations rather than ugly tracking links.
And 1jump2 handles the commercial journey underneath without taking over the content.
That, to me, feels much closer to where affiliate publishing should be heading.
Useful content first. Commerce when it helps. Infrastructure underneath.
Not another shopping website.
Just a better answer that happens to know what you might need next.
