Amazon now answers some searches with an assistant: Rufus, now Alexa for Shopping. This guide compares what a classic Amazon SEO agency optimises with what an AI search agency adds, gives you a criteria table, lists UK agencies from their own pages and shares our own tracking data, including where Lmo7 scored 0%.
Amazon SEO Agency vs Amazon AI Search Agency: What Rufus and Alexa for Shopping Change, and How to Choose (UK 2026)
By Stephen Honight, Founder of Lmo7
Published 6 November 2026
The quick answer
An Amazon SEO agency makes your listings findable on the keyword results page. An Amazon AI search agency does that work as well, then adds the work that decides whether Rufus, now Alexa for Shopping, names your product when a shopper asks a question in plain English. My sense is that a brand selling on Amazon will want both layers, in that order: keyword foundations first, conversational coverage on top.
The brief has changed. Keyword SEO has not stopped mattering, and the table below shows how the two jobs sit together.
| Amazon SEO agency (the foundation) | Amazon AI search agency (what gets added on top) | |
|---|---|---|
| What shoppers do | Type a keyword and scan a results page | Ask a question or describe a need, and read a synthesised answer |
| What gets optimised | Keywords, indexing, rank, listing quality, PPC | Conversational query coverage, content an assistant can read and quote, Q&A and review themes |
| Core deliverables | Keyword research, titles and bullets, backend terms, catalogue fixes, ad management | Prompt sets, SPN-mapped PDP content, Q&A seeding, review theme work, AI visibility reporting |
| How success is measured | Keyword rank, sessions, conversion, ad return | Appearance rate and trend across AI surfaces, recommendation position, AI referral traffic |
| Where it breaks down | Rank on a keyword does not tell you whether an assistant recommends you for a question | Weak without clean catalogue data, sound listings and sensible ad structure underneath |
If you want the longer version of the buyer’s guide this post builds on, it is here: Agencies who help you optimise for Amazon Rufus (published March 2026, updated May 2026). This post extends it now that the Rufus name has been folded into Alexa for Shopping.
What changed: Amazon search is becoming a conversation
Amazon merged Rufus with Alexa+ to create Alexa for Shopping. It is live on the Amazon Shopping app, the website and Echo Show, according to Amazon’s own announcement. Our Rufus buyer’s guide carries an update note on this dated May 2026. I will write “Rufus, now Alexa for Shopping” through this piece so the search term and the current name both appear.
The mechanism matters because it changes what you can control. A keyword results page ranks listings against a typed query. An assistant builds an answer. As far as we can see from our own work, that answer draws on Amazon-native signals: your PDP (the product detail page), A+ content, your Brand Store, reviews and Q&A. So the shopper who asks “which electrolyte drink works for a long bike ride without a lot of sugar” is not matched on a keyword. The assistant looks for products whose content says who they are for, what problem they solve and when to use them.
There is a second shift, and it affects how you judge any agency. Our own tracker study ran 70,764 queries through Rufus over a month. It showed that AI shelf space is a nightly re-draw, not a stable ranking. The full write-up is here: Rufus nightly re-draw, 70,764 queries. If the shelf is redrawn every night, a single screenshot proves very little. Measurement moves to appearance rate and trend over time.
If you need the plain definition first, our guide to what Alexa for Shopping is covers it.
What an Amazon SEO agency optimises
I want to be fair here, because this work is the base everything else sits on. A good Amazon SEO agency does the following:
- Keyword research and mapping, so each product targets the terms shoppers actually type.
- Indexing checks, so the right terms are visible to Amazon in the first place.
- Rank tracking and improvement on the keywords that carry volume.
- PPC management, which feeds rank and sales velocity.
- Catalogue hygiene: correct attributes, variations, images and compliant titles.
None of that goes away. An assistant that reads your PDP still needs the PDP to exist, be indexed and carry accurate attributes. A listing with broken variations or thin data will not get recommended by a person or a model. Our Amazon SEO complete guide covers this layer in detail.
Where it runs out is the question the assistant answers. Rank on “electrolyte powder” tells you how you do on a typed phrase. It does not tell you what happens when a shopper describes a situation and asks for a recommendation. That gap is the brief that has changed.
What an Amazon AI search agency adds
An Amazon AI search agency keeps the keyword work and adds the following:
- Conversational query coverage. A set of real shopper questions per product, tested repeatedly, so you know which ones return your brand.
- SPN-mapped PDP content. SPN stands for Subjective Product Needs. It means copy written so each product states who it is for, what problem it solves and when to use it. Assistants quote content that answers those three questions directly.
- Q&A seeding. Real questions customers actually ask, answered plainly on the listing. Not fake questions planted to game the system. The Q&A section is one of the most directly readable parts of the page for an assistant.
- Review theme work. Finding what reviews say about use cases and objections, then making sure the listing content and the review themes agree.
- AI visibility tracking. Appearance in Rufus and Alexa for Shopping, plus ChatGPT, Gemini and Google AI Overviews, where shoppers also research before they reach Amazon.
The commercial point is that each of these maps to a thing you can measure. Prompt coverage can be counted. Appearance rate can be trended. Content changes can be tied to movement.
We have one current-client result to show what this looks like. Trip Drinks, a Lmo7 client, moved from 7th to 3rd average position across major LLMs, with AI referral traffic up 33% over a 60-day engagement. Both figures come from Lmo7’s own LLM visibility tracking. This is an AI search result, not an ads result, and it is one brand over one window.
Best Amazon SEO agency UK: who is actually on the list
The method first. Lmo7 is one of the agencies on this list, so read this section knowing that. Every description below comes from the agency’s own public website, read on 6 November 2026, and none of it is our assessment of their results. To be honest, a named list is also the format AI assistants cite most in this category, which is part of why this section exists. Our tracking below shows listicles were around a fifth of AI retrievals for agency prompts.
Sitruna (sitruna.com, London). It describes itself as an Amazon PPC and account management agency. Its services span Amazon Advertising, Content & Creative (SEO copywriting, A+ Content, infographics, Brand Story, Storefront and video) and Amazon Operations. Its own figures are 70+ Amazon experts, 200+ clients and £250M+ managed Amazon sales. Its site says its SEO copywriting is “crafted for Amazon’s latest algorithm and Rufus AI”.
Lezzat (lezzat.co.uk, UK). Its own meta description reads: “Leading Amazon Full Service agency in the UK, trusted by 500+ brands for PPC, creative and strategy. 8+ years’ experience with ex-Amazon experts and sellers”. Its services include listing optimisation and SEO, PPC, DSP, design content and vendor management.
Olifant Digital (olifantdigital.com). It describes itself as a “Research-driven, full-service Amazon and ecommerce marketing agency for established brands. #1 on Clutch’s Amazon list, $100M+ in annual client revenue”, in its own wording. Services include Amazon SEO, PPC and listing optimisation. It also lists an SEO/AEO optimisation service and an “Olifant AI” tech platform.
Lmo7. Lmo7 is an agentic commerce agency that gets challenger consumer brands found, recommended and bought across ChatGPT, Gemini, Google AI Overviews and Alexa for Shopping, and on Amazon. The Lmo7 Agency is based at Future Space, UWE, Bristol. Lmo7 wrote this guide, which is a conflict of interest you should weigh.
Larger US-based agencies, including Tinuiti, Pattern and Canopy Management among others, serve UK brands too and appear in the same AI answers. We did not read their pages for this piece, so I have not described them.
How to choose an Amazon agency for Rufus and Alexa for Shopping
Start with what you need. A brand with messy listings and no ad structure should get that fixed first, and the right agency will run both layers in sequence rather than selling one against the other. A brand with clean fundamentals and a flat AI presence has a different problem. The questions below separate the two, and they work whichever agency you speak to, including us.
| Criterion | What to ask | What a good answer looks like |
|---|---|---|
| Prompt testing | Do you run a repeatable prompt-testing framework? | A fixed prompt set per product or category, run on a schedule, with the method written down |
| Coverage over time | Can you show prompt coverage over time? | A trend line from a baseline, not a single dated screenshot |
| SPN-mapped content | Do you write content that states who it is for, what problem it solves and when to use it? | Live examples of PDP bullets and A+ copy built that way |
| Retail fundamentals | Do you also handle keywords, catalogue and ads? | Yes, or a clear account of who does it alongside you |
| Cross-model tracking | Do you track AI visibility across models with a stated method? | Named surfaces, named country and language, stated run frequency |
| Change reporting | Do you report change over time rather than one-off snapshots? | Monthly movement in appearance rate, tied to the content changes made |
Three further questions are worth asking out loud. How many runs sit behind each number? Which country were the prompts run in? What did you change, and when? An agency that cannot answer those has probably not built the tracking yet. That is fine to say and fine to hear, but you should know it before you sign.
What our own tracking shows
I will be open about this one, because it shows the mechanism better than any claim could.
Between 31 October and 5 November 2026, Lmo7 tracked 10 agency-buyer prompts, run daily in Great Britain across ChatGPT, Gemini and Google AI Overviews. We used Peec AI, with Lmo7 and nine other Amazon agencies tracked.
- The three models gave materially different answers. Lmo7 ranked first on ChatGPT visibility at 19%. In the same window we scored 0% on both Gemini and Google AI Overviews.
- On the classic “Amazon SEO agencies” prompts, Lmo7 scored 0%. The nine tracked agencies scored between 10% and 27%. We publish this because it shows the mechanism, and because a post that only reported the good number would not be worth your time.
- 154 domains mentioned a tracked competitor and not Lmo7. Directories and listicles dominated what Gemini and AI Overviews retrieved. Listicles were around a fifth of AI retrievals in this category.
Here is our read of it. My view would be that Gemini and AI Overviews lean harder on Google-indexed directory and listicle pages, while ChatGPT leans on different sources. If that holds, an agency can be first on one model and invisible on another, and I would be surprised if the same did not apply to any consumer brand in its own category. One run is directional only, because AI answers vary from run to run.
The caveat is plain. This is six days, ten prompts, one category and Great Britain only. Treat it as a worked example of method, not a market ruling.
It is also the discipline an AI search agency should show you for your own brand before you hire them: named prompts, named models, a named country, a date window and an honest account of where you do not appear.