Amazon SEO in 2026, 2027 and Onwards

It's been a while since I wrote anything here, mostly because nothing I'd have written would have changed what we actually do. That in itself is the story. The tactics content mill has spent two years announcing that everything is different now, and meanwhile the thing that ranked products in 2018 is still the thing that ranks products in 2026.

But some things genuinely have changed, some things people think changed haven't, and a couple of things quietly died. So here's the honest state of Amazon SEO going into 2027, from someone who ranks products for a living rather than someone selling a course about it.

What has changed

Three things, really.

First, ads got expensive to the point of absurdity. CPCs have climbed every single year I've been doing this and in most categories the maths has now tipped. More on that below because it deserves its own section.

Second, AI search arrived. Rufus sits inside Amazon answering questions, and outside Amazon a growing chunk of buying decisions start in ChatGPT or Google's AI results rather than a search bar. This changes what your listing content is for. It used to be written for a human skim reading bullets. Now it's also source material for a language model deciding whether to mention your product at all. We cover how we handle that on the GEO side of the methodology.

Third, Amazon got better at killing the fake stuff. Review manipulation, rebate club junk, botted traffic. Good. Genuinely good, because it widens the gap between things that look like real demand and things that are real demand. Which brings me to what hasn't changed.

What hasn't changed

The algorithm's core question is the same as it's always been: when someone types this keyword, which product is most likely to convert into a sale? Amazon is a shop. It ranks what sells. Everything else, the title formulas, the backend keyword hacks, the A+ modules, is either a prerequisite or noise.

I've been saying the same line for years and it's still true: on page optimisation is the filter, not the lever. You need it done properly so Amazon knows what you are and indexes you for the right terms. But a perfectly optimised listing with no sales velocity on the target keyword goes nowhere. I've watched sellers pay agencies for their third title rewrite of the year wondering why nothing moves. Nothing moves because words don't move rank. Sales do.

What still works: keyword focused sales

This is the part of the job that hasn't needed updating, because it's not a trick, it's the signal itself. If real customers search your money keyword, click your product and buy it at full price, Amazon's algorithm does exactly what it was built to do and moves you up for that keyword. We call it Keyword Focused Sales and it's been the core of the methodology since 2018.

People sometimes ask whether the crackdowns on manipulation affect this. They don't, and it's worth understanding why. Amazon's enforcement targets fake signals: incentivised reviews, bot traffic, rebate schemes that produce refunded orders. A real person making a real search and paying full price for a product is not a fake signal. It's the exact behaviour the algorithm exists to measure. That's why across 2,500+ ranked products we've never had an account suspended, and it's also why every crackdown quietly helps us, because it removes the noise competing with genuine velocity.

One honest caveat, because I'd rather write it here than have someone find out later. The campaign gets you to page 1. Staying there is about your conversion rate at the new position, which means your reviews, your price and your listing quality. A strong product holds. A weak one drifts back down. It's why we assess every product before quoting and turn down the ones we can't win with, and it's why the first thing we send anyone is a free Revenue Gap Report rather than an invoice.

Amazon PPC is dead

As a growth strategy, to be clear. Ads still have a job: launch visibility, defending your own brand terms, mopping up long tail. But the idea of PPC as the engine that grows an Amazon business is finished for most categories, and the reason is simple arithmetic. CPCs have risen every year while conversion rates haven't, so the same sale costs more every year, forever, and the moment you pause, everything you paid for disappears.

The sellers doing well in 2026 treat ads as a tool with a defined job, not a growth plan. The growth plan is organic rank on the handful of keywords that actually pay for your category, which is precisely what a campaign buys once instead of renting monthly. There's a working example in the case studies if you want to see the numbers on a real product.

Reviews in 2026

Reviews matter, but not the way most sellers think. They're a conversion asset and a qualification floor, not a ranking lever. You cannot review your way up a results page, and buying reviews is the one thing on Amazon genuinely likely to get your account nuked, so don't.

What reviews do decide is whether a ranking campaign is worth running at all. If you're sitting at a 3.9 star average against page 1 competitors at 4.5, velocity will get you up there and your conversion rate will send you straight back down. Fix the product or fix the review base first. This is exactly the kind of thing we tell people at the report stage, sometimes to our own cost, because taking money for a campaign that won't hold is a great way to make one sale and lose a client.

Rufus, AI search and where this goes

Here's my honest read on Rufus and AI search generally. It changes discovery more than it changes ranking. When Rufus or ChatGPT recommends a product, it's drawing on the same underlying signals: what's visible, what converts, what people actually buy, plus the content on and around your listing. Which means two things.

One, your listing copy matters again, but differently. It's now feeding a model that summarises and compares, so factual, specific, structured content beats keyword stuffed fluff by a wider margin than ever. Two, being the product that demonstrably sells for a keyword still wins, because every recommendation system, human or AI, leans on bestseller status, rank and sales history as its trust signals. AI search is a new front door into the same warehouse.

We've folded this into the methodology as GEO, generative engine optimisation, and the practical work overlaps heavily with doing Amazon SEO properly. Which is convenient, because it means the answer to "what should I do about AI search" is mostly "the fundamentals, done well, plus some structural content work" rather than a new retainer for a new acronym.

Why we still do this

Because it still works, and because almost nobody else will say the quiet part out loud: rankings come from sales, so if you want to rank, engineer the sales. Everything else in this industry is either a prerequisite dressed up as a strategy or a way to bill you monthly for activity.

Same approach since 2018. 2,500+ products ranked. Zero accounts suspended. If you want to know what page 1 is actually worth for your product, in pounds and pence per month, that's exactly what the free Revenue Gap Report tells you. We'll also tell you if your keyword isn't winnable yet, and why, which is occasionally worth more than the campaign would have been.

Get my free Revenue Gap Report →

Questions this didn't answer are probably in the FAQ, and the full mechanism is written up plainly on the Keyword Focused Sales page.