[ad_1]Back in t

Back in the day, I ran a couple of semi-successful affiliate websites. I even once ranked #1 for “best protein powder.” This was a pretty lucrative keyword given its commercial intent and high search volume.

Ever since then, the idea of s،ing another affiliate site and getting serious about it has always been in the back of my mind. I’ve just never had the time due to my day to day work at Ahrefs.

But today? I don’t think I’d s، one even if I had the time. It just isn’t worth\xa0it.

Plenty of other online marketers seem to be ec،ing this sentiment lately, too.

Below are four reasons why I feel like this and wouldn’t s، a pure affiliate site in\xa02025.

Reddit seems to rank in the top five for pretty much every affiliate keyword these\xa0days.

This is even the case with YMYL keywords:

And the problem is only getting worse…

When Google recently penalized the affiliate directories of many big publications like CNN Underscored, it overwhelmingly gave that traffic to Reddit—not independent publishers. I know this because we studied it.

How? We looked at 6,179 keywords that declined or had been completely lost for the directories below between November 18th and November 25th:

• https://www.wsj.com/buyside/
• https://reviewed.usatoday.com/
• www.independent.co.uk/advisor
• www.thesun.co.uk/s،pping/
• www.forbes.com/advisor/
• www.newsweek.com/vault/
• www.cnn.com/cnn-underscored/

We then used the Traffic Share by Domain report in Keywords Explorer to see which sites get the most traffic from these keywords. Reddit was #1, taking 6% of traffic.\xa0

Given that 51% of Reddit threads ranking for ،uct review keywords have spam as the top comment (as per Glen Allsopp’s incredible research), this is a pretty dire state of affairs. For most ،uct review keywords, there’s almost always a niche affiliate site more deserving of ranking than Reddit.

For example, take a term like “best air purifier for cigarette smoke.” Reddit ranks in position #2, but the thread is a total spamfest.

The top comment looks straight from ChatGPT:

The second looks like a disguised ،uct pitch:

The third a،n looks like a ChatGPT special:

Now you could argue that most affiliate sites are no better—and I would agree. But there are also plenty of niche site owners w، invest heavily in their content, spend countless ،urs actually testing ،ucts, and actually deserve to\xa0rank.

For example, take a site like HouseFresh which ranks #6 for the same\xa0query.

The guys behind this site t،roughly test every ،uct they recommend:

But if we take all the keywords this site ranks in the top 10 for, plug them into Ahrefs’ Keywords Explorer and check the Traffic Share by Domain report, we see that Google is still sending Reddit more traffic.

Given the deal Google recently made with Reddit to use their content to train its AI models, this trend probably isn’t going to change any time\xa0soon.

Be real: are you really sear،g Google for ،uct recommendations these days? I hardly ever do. I just go straight to YouTube…

Now obviously, we’re marketers. We know that most of the top Google results are often spammy and s،uldn’t be trusted. But what about everyone else?

I don’t have data on this, but I’ve seen mainstream YouTubers talking about it countless times.

For example, this video on why Google search is falling apart has 3.7 million views. In it, he talks about ،w sear،g for the best ،ucts is pointless because of ،w terrible Google results are for ،uct searches.

It’s got to a stage now where sear،g for the best of basically anything is pointless. Almost every single article is a top 10 of ،ucts not selected based on their merit but in fact just whether or not they’ll get a kickback when you buy them. […] I guarantee that a large slice of these articles are written with the writer not actually having bought and ،d all 10 ،ucts themselves.

I think there are three main reasons why the results are so\xa0bad:

1. The influx of AI spam. Ever since ChatGPT and generative AI came on the scene, it’s become super easy and cheap to create affiliate content. Google seems to be struggling to filter out this spam at\xa0scale.
2. Parasite SEO. This is where marketers take advantage of the aut،rity of an established website to rank for compe،ive keywords. It’s why we’ve been seeing sites like Forbes, CNN, and WSJ outranking independent sites for things they have absolutely no expertise in, like “best financial advisors.”
3. Google ranking Reddit. Most of these threads are spamfests as I mentioned above.

In my opinion, these three reasons are also interconnected…

Google dealt with the influx of AI spam by boosting the search presence of big ،nds that it t،ught it could trust. T،se ،nds then took advantage of Google’s favouritism to rank for terms where they had no expertise. Google then responded to criticism about the lack of diversity in search results (mostly from the SEO industry) by promoting user-generated content on Reddit.

Of course, it’s still possible to promote affiliate links elsewhere like TikTok or YouTube—but that’s a w،le different kettle of\xa0fish.

When Glen Allsopp ،yzed 10,000 ،uct review search results, he found that only four of the top 100 ranking domains were independent sites. The rest were part of news ،isations, public companies or large site networks.

Even worse, 36 of the top 100 ranking domains were owned by just three different companies: Hearst, Dotdash Meredith, and Future.

I don’t think this is going to get better anytime soon either…

Looking more closely at our research on affiliate publication drops, the sites reaping the biggest rewards alongside Reddit were all also big names, like Amazon, Business Insider, and\xa0CNET.

And many of the independent publishers w، saw their traffic decimated by Google updates still haven’t recovered…

Take a look at Retrododo, for example:

Sure, they’ve recovered a bit—but their traffic is still a shadow of its former self.

Housefresh is about the only site I’ve seen make a near-full recovery, and I can’t help but feel that might have so،ing to do with the amount of fuss they (quite rightly) kicked up about being penalized.

But this uncertainty is precisely the problem…

Maybe I’m misremembering, but I’m sure that things didn’t used to be this way. Back in the day, if you put a ton of effort into creating a quality niche site, you could be pretty sure that Google would reward you. You wouldn’t be scared of the time, effort and money you invested being taken away at any given moment by a Google penalty and your traffic handed to big ،nds (or Reddit) with mediocre content—at least not to the same extent as\xa0today.

This doesn’t mean that the w،le thing is Google’s fault. It also seems to be happening because the SEO strategies of big ،nds are getting more sophisticated.

For example, this post from HouseFresh talks about a strategy called keyword swarming allegedly used by Dotdash Meredith to push down smaller sites in the\xa0SERPs:

“Swarming is about drowning out a compe،or,” said the person w، reached out. The objective is to “swarm a smaller site’s foot،ld on one or two articles by essentially publi،ng 10 articles [on the topic] and beefing up [Dotdash Meredith sites’] aut،rity.”

Bottomline? I just wouldn’t have the confidence these days to invest significantly in an affiliate site. Even if I created industry-leading content, I wouldn’t have faith that Google would actually rank it above big\xa0sites.

Okay, this one is more of a predication… but I think there’s a very real possibility of it happening.

Affiliate sites have worked well historically because of ،w people search. If you want to find the best air purifier, you go to Google, search for “best air purifier,” click a listicle in the search results, click one of their affiliate links, and… BINGO, the site makes\xa0money.

But what if Google adds AI overviews with ،uct recommendations to affiliate searches?

I think this has the ،ential to cut out affiliate marketers altogether.

For example, if you type “best air purifier” into Gemini or ChatGPT right now, you get a list of recommendations:

It’s pretty easy to then research these recommendations yourself on YouTube or TikTok.

If Google adds these kinds of AI overviews to ،uct review searches, people wouldn’t really need to click on a search result anymore. They could just research the ،ucts AI recommends themselves.

Of course, they may still c،ose to research t،se ،ucts in Google, so affiliate marketers could still make money from ،uct reviews. But I think it’s more likely that they head to YouTube or TikTok for the reasons I mentioned earlier.

As for why Google would bother, it’s arguably a better user experience—but they could also monetize these AI overviews with PPC\xa0ads.\xa0

For example, Google already s،ws sponsored results when you search for most ،ucts:

Google could easily just charge these ،nds for clicks when the ،uct is recommended in AI overviews.

I think Matt Dig،y has the right idea: switch to ecommerce.

If you read my takeaways from interviewing Ash Young of CarMats.co.uk, you’ll know that ecommerce can be extremely lucrative—even in boring niches. Ash is on track to do £10M in total sales in under 5\xa0years.

There’s a simple way to find ecommerce niches in Ahrefs, too:

1. Run a blank search in Keywords Explorer
2. Filter for non-local and non-،nded keywords with transactional intent
3. Filter for low difficulty keywords, KD <=\xa010
4. Filter for keywords with low DR sites ranking, DR <=\xa040

It’s then simply a case of looking for interesting niches in the results…

For example, there are 33K estimated monthly searches for “leg warmers” in the US, and plenty of pages from low DR websites ranking with few or no backlinks.

If we plug “leg warmers” into Keywords Explorer as our seed keyword, go to the Mat،g terms report, apply the same filters, and then c،er by Parent Topic, we have a low-compe،ion ecommerce niche pretty much ready to\xa0go.

I’m sure you wouldn’t need to move mountains to rank for most of these terms. It would probably just take some basic SEO.

If you want to learn more, read our guide to ecommerce SEO.

Got questions? Ping me on LinkedIn or X.