Overview

AI is everywhere in 2025 – from generative chatbots writing copy to algorithmic assistants in the Amazon marketplace. But for Amazon agency professionals, a pressing question looms: Is AI truly transforming how we drive results, or is it just another shiny object pulling us away from core strategies? In a recent MerchantSpring webinar, host Paul Sonneveld (Co-founder of MerchantSpring) sat down with Chris Turton, Managing Director at Ecommerce Intelligence (UK), to cut through the hype. Turton, known for his no-nonsense approach, offers a contrarian perspective on AI for Amazon agencies – challenging the “silver bullet” narrative and refocusing on what really moves the needle for clients.

In this comprehensive recap, we’ll synthesise their insights into a thought leadership narrative. We’ll explore why many Amazon AI tools and promises don’t yet live up to the buzz, how Amazon’s own generative AI initiatives (like Rufus and Cosmo) impact shopping, and where agencies should double down instead. You’ll also find practical guidance on balancing innovation with fundamentals, real examples of AI helping (and not helping) agencies, and an expert outlook on the road ahead. Whether you’re an agency leader feeling pressure to adopt the latest AI tool or a skeptic wondering if it’s all smoke and mirrors, read on for a grounded take on AI’s role in the Amazon agency world.

The Allure of AI: Productivity Booster or Overhyped Panacea?

AI’s allure in the Amazon sphere is undeniable. Many vendors promise that a few clicks in a tool powered by machine learning can supercharge an agency’s entire workflow – from automating listing creation to managing ads. However, Turton argues the biggest misconception in the Amazon agency world is the idea that AI is “all-encompassing” – a magic wand for the entire end-to-end process. It’s tempting to believe an AI tool can handle everything and instantly make brands successful, but reality is far more nuanced.

“People are always looking for that quick fix… somehow there’s this concept that AI can empower the entire funnel… just click a few buttons and you’re covered. That’s not the case.” – Chris Turton

In truth, current AI applications tend to excel at narrow tasks – often the surface-level, generative tasks like drafting content or creating images – but fall short on deeper strategic or analytical work. For example, Amazon’s own AI Creative Studio can generate lifestyle product images from a single photo, a clever time-saver for ads. Generative AI content tools (from ChatGPT to copywriting bots) can spit out product descriptions or ad copy in seconds. AI image generation can produce infographics or enhanced photos on the fly. On the surface, these are impressive and can boost productivity. Chris Turton acknowledges these “really smart, clever” innovations.

The problem? Many AI-driven promises don’t hold up in practice. Agencies and brands chasing the latest AI tool often discover huge gaps between hype and reality. There’s a disparity, as Turton puts it, between what generative AI produces and the rigorous, tactical needs of running a profitable Amazon business. An AI might generate a slick-looking listing or a pile of data, but does it understand your client’s brand voice, optimise conversion deeply, or align with real shopper intent? Not yet – and treating AI as an instant fix can distract from the hard work that truly drives ROI.

Shiny Object Syndrome: Picking Useful AI vs. Hollow Hype

In a world of new AI tools popping up weekly, how can agencies tell which are actually useful? Turton’s advice: start with the problem, not the tool. Rather than adopting AI for AI’s sake, identify a specific pain point in your workflow or a repetitive task consuming your team’s hours. If an AI application clearly solves that problem faster or better, it’s worth exploring. If not, beware the shiny object.

Many AI tools for Amazon sellers being touted focus on top-of-funnel tasks: for instance, chat-based agents that summarise reports, or generators that crank out content. These can be time-savers, but Turton notes a common thread: a lot of them handle things that are relatively easy or already “macro.” For example, numerous tools use GPT-like agents (e.g. custom ChatGPT instances or Anthropic Claude) to churn out listing text or analyse basic reports. Yet “they’re all very top-line”, often neglecting the deeper layers of Amazon's selling – like nuanced conversion optimisation, profit analysis, catalogue policing, or brand strategy. In Turton’s experience consulting with 70+ brands, clients rarely ask “What’s your AI stack?” They ask, “How will you improve our ROI, protect our IP, and grow our sales?” – fundamentals that no standalone AI has mastered.

So when evaluating the next “revolutionary” AI product for your agency, ask: Does it tangibly solve a priority issue or just sound cool? Turton suggests a tried-and-tested approach: pilot new AI processes internally on a small scale, measure results, and see if it truly streamlines work without sacrificing quality. It’s easy to burn hours fiddling with a flashy tool only to realise later you could have done the task manually in half the time. 

Useful AI applications tend to integrate naturally into your existing workflow to eliminate drudgery or surface insights you’d miss – for example, using ChatGPT to quickly summarise a large chunk of performance data and bullet out key trends (something Turton’s team finds AI very good at). Another example is feeding an AI tool with bulk review data to extract common customer pain points, which can inform product improvements faster than manual analysis. These targeted uses of AI in an Amazon agency can indeed boost productivity.

On the flip side, hype-driven tools promise end-to-end automation but often require extensive human cleanup or simply don’t address the real bottlenecks in an agency’s service. Turton cautions that if a tool’s value proposition is fuzzy and it’s not immediately clear how it improves a specific workflow or KPI, it’s likely more noise than signal. In short, prioritise AI that augments your team’s strengths, not tools that replace the very expertise clients pay you for.

Amazon’s Own AI: Rufus & Cosmo – Revolution or Sideshow?

No discussion of AI in the Amazon ecosystem is complete without examining Amazon’s homegrown AI initiatives. At Amazon Accelerate and in recent rollouts, Amazon has heavily touted features like Rufus, a generative AI shopping assistant, and Cosmo, a new AI-driven search model aimed at understanding shopper intent. These developments have sparked plenty of buzz in the seller community. Should agencies be overhauling SEO and content strategies to appease Rufus and Cosmo? Turton’s take: Proceed with caution and perspective.

Rufus – Amazon’s chat-style shopping assistant – was introduced as a “personal shopping assistant” within Amazon’s mobile app and site search. In theory, a shopper can ask Rufus conversational questions (e.g. “What’s the best laptop for graphic design under $1000?”) and get personalised product recommendations or answers drawn from Amazon’s vast catalogue and the web. It’s Amazon’s answer to the conversational AI trend, bringing a ChatGPT-like experience into e-commerce. 

Cosmo, on the other hand, is an AI engine behind Amazon’s search that uses a common-sense knowledge graph and large language models to better infer what customers mean when they search. For example, if someone searches “pregnancy shoes,” Cosmo might intuit they need comfortable, flat footwear for pregnant women, even if the listing doesn’t explicitly say “for pregnancy.” Essentially, Cosmo works to match products to implied needs and contexts, moving beyond literal keyword matching.

This all sounds revolutionary – and eventually, it might be. But how much is it changing shopper behaviour today? Thus far, the impact seems limited. Turton points out that we simply don’t have strong evidence yet of significant uptake. Early data suggests that few shoppers are actually using Rufus regularly. For instance, one agency’s analysis (cited by Turton) estimated Rufus handled only about 13% of Amazon searches shortly after launch – and even that number hasn’t been updated with any growth. 

Prime Day saw a spike in Rufus usage (likely due to Amazon’s promotion of it), but there’s little sign of sustained engagement afterwards. In fact, independent observations indicate that perhaps only 3 out of 100 purchases involve Rufus at all – a mere ~3% of transactions. Such low adoption means Rufus is far from transforming how the majority shop on Amazon at present. Even Bloomberg’s tech columnists have “found it lacking”, noting instances where Rufus gave irrelevant or incorrect answers. Usability experts at Nielsen Norman Group also reported that most users “didn’t even notice” the Rufus chat feature on Amazon’s interface, as it was tucked away and not part of their normal shopping habit. All this suggests that Rufus, while innovative, is more of a beta experiment than a game-changer in 2025.

Cosmo’s influence is harder to measure directly since it works behind the scenes. It represents Amazon’s shift toward AI-driven, intent-based search, somewhat akin to how Google evolved to understand natural language queries. Cosmo aims to connect the dots between a vague query and relevant products using learned common-sense relationships. For example, it might know that “TV bracket” in industry jargon corresponds to “TV wall mount” for a consumer search (synonyms understanding), or that someone searching “running shoes” in July might be training for a marathon, thus surface marathon-friendly products. These are promising enhancements. 

However, Turton emphasises a critical reality check: Amazon’s ranking algorithm still fundamentally cares about revenue and performance. AI or not, Amazon will not show a product – no matter how semantically relevant – if that product generates less downstream revenue (via price x conversion) than another. “Amazon is always going to rank products based on the referral fees, ad fees and FBA fees it makes… It’s not going to recommend a product with no reviews or sales just because an AI thinks it’s a perfect fit for the query”, Turton notes. In practice, that means sales velocity, conversion rate, and fulfillment metrics remain king, even as AI is layered on.

For agencies, the takeaway is: don’t overreact to Amazon’s AI hype with knee-jerk strategy pivots. Yes, keep an eye on how features like Rufus and Cosmo evolve. For instance, ensure your product listings contain rich information and context (dimensions, use cases, clear benefits) so that any AI – whether it’s a chatbot or an algorithm – can easily identify why your product is a good match for certain needs. In other words, solid listing optimisation (good titles, bullet points, and descriptions with natural language that explain the product’s use) will serve you well in an AI-driven search environment. This is basically what sellers should be doing anyway for traditional search. 

If you already craft listings focusing on real customer questions and benefits rather than just stuffing keywords, you’re likely in good shape for Cosmo’s intent mapping. But don’t pour dozens of hours into, say, tailoring content specifically to “please Rufus,” when its usage is minimal. As Turton advises, agencies must be “careful about overreacting to the hype over real customer adoption”. The core principles of driving ranking – sales consistency, conversion optimisation, earning strong reviews – haven’t been upended overnight by AI.

In fact, chasing the Rufus/Cosmo hype could lead some agencies to overlook the basics. There might be a temporary window of arbitrage (for example, if one discovered a quirk in how Rufus presents answers and gamed content for it), but those loopholes will close as Amazon refines the models. And if the AI shopping experiment “bubbles” then fizzles (a scenario Turton thinks is quite possible, akin to the dot-com bubble bursting before true web innovation resumed), agencies that neglected fundamentals will be worse off. Instead, the smart play is to monitor AI developments but prioritise strategies that improve real performance metrics. If or when Amazon’s AI features gain substantial traction, you’ll have the data and solid foundation to adjust in a measured way.

Client Pressure: “Do We Need to Adopt AI ASAP?”

With all the buzz, many agencies are feeling pressure not just internally but from their clients. It’s easy to imagine a client executive reading a headline about AI or attending a conference, then asking their agency, “What’s our AI strategy? Are we using Rufus and the latest tools? We don’t want to miss out!” Turton’s experience, interestingly, is that clients rarely phrase it that way. He notes that no client has ever come to him asking for a rundown of what AI tools his agency uses. Instead, they ask for outcomes: better ROI, higher sales, stronger brand protection.

However, some clients do bring AI-generated materials or ideas, for instance, handing over a product listing draft written by ChatGPT, expecting the agency to simply fine-tune or accept it. In those situations, education and refocusing on goals is key. Turton’s approach is transparent: he explains to clients why the agency isn’t blindly jumping on every AI trend and ties it back to the client’s own success metrics. For example, he might say: “We could spend 6 hours implementing this new AI-based SEO trick, but we’d rather spend those 6 hours improving your top 20 listings’ conversion rates – which we know will yield a tangible lift in sales.” By translating it into the language of ROI and opportunity cost, clients quickly understand that it’s not about missing out, it’s about prioritising proven impact over experiments.

“It’s all about setting expectations. We tell them: we measure success by your market share, your BSR, your profits – and right now, there’s little evidence that chasing [AI chat features] will move those needles. We’re keeping an eye on it, but we’re investing your time and money where it counts.” – Chris Turton (on responding to clients’ FOMO around AI)

Another tactic is to share data or reputable insights on AI uptake (or lack thereof) to assure clients they’re not falling behind. For instance, mentioning that initial reports show only a tiny fraction of shoppers use Rufus regularly, or that major tech publications like Business Insider and Bloomberg have flagged the limited usefulness of Amazon’s AI assistant, can add credibility to your counsel. The key is to acknowledge the client’s concern (AI is exciting and worth watching) but then to reframe the conversation to business value. Often, once clients hear that your agency is indeed researching and even testing AI behind the scenes – just not at the expense of their current results – they’re satisfied. You might even walk them through a small success, e.g., “We used an AI tool to analyse your last 6 months of reviews and discovered customers keep mentioning ‘packaging issues’ – we’re addressing that in your content and support.” This shows you’re not anti-AI; you’re pragmatic about it.

Practical AI Wins: Where Automation Helps Agencies Today

Despite his contrarian stance on the hype, Turton is not anti-technology. In fact, his agency has quietly woven in AI where it genuinely adds value. Where does AI shine for Amazon agencies in 2025? According to Turton, some of the best use cases are in data crunching and initial content drafting – essentially, AI as your junior assistant, not your strategist.

1. Data Analysis & Reporting: Agencies drown in data – search term reports, campaign metrics, inventory reports, reviews, pricing trends, and more. This is where AI can act like a supercharged analyst. Turton’s team, for example, uses tools like Claude (an AI model by Anthropic) or ChatGPT to digest large spreadsheets or datasets and spit out human-friendly summaries. They might input a table of a client’s Subscribe & Save data or dozens of ad performance metrics, and prompt the AI to highlight notable patterns (e.g., “Which discount codes yielded the highest subscription uptick?” or “Summarise key trends from these 500 customer reviews.”). 

The AI’s ability to “simplify data and really break it down” saves hours and helps surface insights quickly. Similarly, AI can merge data from multiple sources (say, sales and inventory) and identify correlations that would be tedious manually. The result isn’t presented directly to clients without vetting, but it gives the strategists a head start. This is AI augmenting human analysis – freeing the team to focus on interpreting results and planning actions.

2. Content Drafting (with Human Editing): Turton distinguishes between “creating” and “optimising” content. AI is already quite good at the former: generating a first draft of a product listing or ad copy. For instance, using Amazon’s own beta AI listing tool or external ones like ChatGPT, an agency writer can get a 150-word description that covers basic features in seconds. Turton notes this can “get the creative up and launched faster”, which is valuable when you’re onboarding a new product line and need to list dozens of SKUs quickly. 

However, the real work – optimisation – starts after that draft. The agency team will A/B test and refine the AI-written content over weeks, tweaking titles, bullets, and images to improve click-through rate and conversion rate. The initial AI draft is just a starting point, not the final word. Without ongoing human-led optimisation, AI-written listings often don’t move the needle. Turton has even fielded prospects who tried a fully AI-generated content approach and saw flat results: “I let AI handle it… and I’m not getting anywhere,” one seller admitted, underscoring that a continuous human strategy was missing. The lesson: AI can accelerate routine content generation, but it cannot (yet) replace the creative strategy and testing that optimises that content for Amazon’s algorithm and real shoppers.

3. Automating Repetitive Tasks: Think of all the small, tedious tasks that eat up an agency’s day – checking for listing suppressions, monitoring for unauthorised sellers, pulling routine performance reports, etc. Many of these can be streamlined with scripts or AI. Turton’s team leverages tools (sometimes even simple scripts connected to APIs) to automate things like weekly account health checks or basic listing audits. While not “AI” in the flashy sense, even rule-based automation is part of the smarter toolset. More advanced AI could handle tasks like bulk translating listings for international marketplaces or adjusting bids based on predefined ROI rules (some agencies use AI-enabled PPC management software to do dynamic bidding). 

The key, again, is oversight: fully autonomous AI PPC tools have emerged, but Turton notes he’s “not heard any good words about them” so far – often they overspend or miss nuances, and clients revert to human managers. Instead, his agency might use an AI-enhanced tool to suggest bid changes, but an expert reviews and approves them. This human-in-the-loop approach ensures quality control while still benefiting from AI speed.

By focusing on these pragmatic applications, Turton’s agency has indeed become more efficient over the past two years, “running faster, quicker, smarter, deeper” in certain areas. For example, tasks like building a monthly performance summary that used to take a full day of manual Excel work can now be done in an hour with AI assistance, giving the team more time to devise a strategy. The net effect: AI is a valuable assistant, but not the strategy driver. It helps with the heavy lifting of data and draft content, while the agency team applies the polish, context, and strategic direction. This balanced adoption is likely why Turton’s firm continues to grow ~20% year over year without ballooning headcount – they’re leveraging tech, but wisely.

Maintaining the Human Edge (While Embracing AI Wisely)

One of the overarching themes in the webinar was balance. In an era where AI can do more and more, what is the role of human experts in Amazon agencies? Turton’s answer: an absolutely critical one. He warns that agencies branding themselves as “all-AI” or trying to automate everything risk losing the plot. The fundamentals of ecommerce – understanding consumer psychology, crafting brand stories, problem-solving supply chain or account issues – cannot be fully offloaded to AI. At least not yet, and likely not for a long time, because these require creativity, empathy, and strategic judgment.

Agencies should continue to invest in human talent and upskilling teams on both core Amazon skills and AI literacy. Turton shared how his team continually adapts their processes and encourages learning new tools – with a guiding question: “What can we do right now to make this more efficient?” If an AI or automation can reduce grunt work, they’ll test it. But equally, they train staff on interpreting AI outputs and not becoming over-reliant. A junior analyst might use an AI to generate a report, but needs to learn to verify and cross-check the recommendations.

This not only ensures accuracy but also helps the professional growth of the team in an AI-enabled world. Junior team members get to handle more advanced analysis once freed from rote tasks, and senior members focus on higher-level strategy rather than micromanaging spreadsheets. In essence, AI can elevate human roles if implemented thoughtfully – it takes away some of the mechanical aspects of the job, allowing humans to concentrate on creative and strategic contributions.

Turton also highlighted that, for all the talk of AI, clients still highly value the human touch. Regular communication, transparent reporting, and the feeling that a knowledgeable person is watching out for their business are a huge part of client retention. An AI might send an automated report, but it won’t get on a call and confidently answer why sales dipped last week (at least not in a way most clients would feel comfortable with). Trust and relationships are built by humans. Thus, agencies should see AI as a tool to enhance their service delivery, not a replacement for client interaction.

This philosophy also safeguards agencies against the day AI does become more powerful. By being “AI-augmented” rather than AI-dependent, an agency can always differentiate itself with human expertise.

As Turton quips, “The value isn’t about our AI stack; the value is in the ROI we deliver.”

And that ROI comes from creative strategy, cross-channel thinking, and sometimes gritty hands-on work – things an agency gets paid for precisely because they’re not simple to automate.

Future Outlook: Staying Sober in the AI Gold Rush

Peering 2–5 years into the future, it’s clear AI will advance and likely become a standard part of doing business on Amazon. Turton readily admits that AI will integrate more fully into agencies in the next few years, becoming table stakes in areas like campaign optimisation, demand forecasting, and even creative generation. The current wave of tools is just the first generation. However, he expects a turbulent journey. There could well be an “AI bubble” that bursts: too many companies and agencies throw money and faith into immature AI solutions, only to face disappointment and pull back. That shakeout would leave the survivors – the genuinely useful AI applications – to slowly mature and take hold. In other words, a repeat of the dot-com era cycle: hype, crash, then real innovation rises from the ashes.

For Amazon agencies, Turton’s advice is to remain tech-curious but level-headed. He envisions the successful agency of 2027 as one that has seamlessly woven AI into its operations in dozens of small ways, boosting efficiency and insight, but still driven by human-led strategy and creativity. These agencies will use AI not as a marketing gimmick but as a quiet competitive advantage in delivering results. In contrast, agencies that today claim “100% AI-driven” may either course-correct to a more hybrid model or risk fading if their results don’t back the hype. (Notably, Turton mentioned pure AI-driven Amazon PPC management services that emerged recently often saw clients revert to human managers due to subpar results – a telling sign that going all-in on AI without a human touch can backfire.)

One heartening point is that clients seem to recognise authentic value. As Turton noted, clients ultimately care about growth and ROI, not whether you used a fancy algorithm or manual elbow grease. Five years from now, clients will likely expect agencies to use AI where appropriate (just as they expect you to use any modern software), but they will still choose agencies based on expertise, trust, and track record. By focusing on those pillars and adopting AI in service of them (rather than as a substitute), agencies can thrive in the AI era.

“You won’t beat AI by ignoring it. You’ll beat your competitors by using it better, faster, and smarter than they do – while still doing the things AI can’t.” – Insight from an Amazon seller

As this quote underscores, the winning formula is AI + human, not AI vs. human. Agencies should aim to be augmented agencies – leveraging AI for Amazon advertising efficiency, for Amazon SEO research (e.g., finding semantic keywords), for spotting trends, etc., but also doubling down on human-only strengths like brand storytelling, creative problem-solving, and building client relationships.

Conclusion

AI is heralded as the next big breakthrough for ecommerce, and indeed, its potential is staggering. But as Chris Turton expertly reminds us, potential isn’t the same as present reality. For now, Amazon agencies shouldn’t lose sleep over AI taking their jobs or making their hard-won expertise obsolete. The real risk isn’t being replaced by AI; it’s getting distracted by it. Those who chase every AI fad might neglect the core strategies that actually drive client value – strong content, savvy advertising, diligent account management, and data-driven optimisations. In contrast, agencies that remain grounded, adopt AI selectively, and continue to prioritise client ROI will not only survive but thrive.

In practice, that means using AI as a tool in your arsenal – to automate the tedious 20% of tasks, to gain insights faster – but keeping your focus on the 80% that require human judgment. It means responding to client inquiries about AI with confidence and clarity, educating them on what truly matters. And it means staying curious: encourage your team to experiment with new AI features (perhaps Amazon will roll out more AI in Seller Central soon), but also to report honestly on what works and what doesn’t.

The bottom line for Amazon agencies: Stick to your fundamentals, and layer AI on top where it makes you better. As Turton put it, delivering great ROI and consistent growth for clients is the best sleep medicine – do that, and you “don’t have to lie awake at night about AI.”

Ready to dive deeper into this discussion? Watch the full webinar episode (Marketplace Masters: “AI in Amazon Agencies – Game-Changer or Shiny Distraction?”) featuring Chris Turton for an unfiltered conversation on AI hype vs. reality. You’ll catch even more candid insights and audience Q&A that we couldn’t fully capture here. If you found these insights useful, consider subscribing to our Marketplace Masters podcast for regular agency best-practice discussions. And of course, if you’re looking for a partner to help navigate Amazon’s fast-changing landscape (with a balanced approach to cutting-edge tools and proven strategies), contact MerchantSpring – we’d love to help you sharpen your edge.