Amazon Put Your Title on a Diet: What the 75-Character Limit Actually Means for Sellers

Okay, I need to talk about something that's happening on Sunday. And no, it's not a pleasant weekend event. It's Amazon telling every seller on the platform that the product title — the single field you've obsessively optimised since approximately the Obama administration — now has to fit inside 75 characters. Spaces included. That's roughly the length of this sentence right here.

If you're sitting there thinking "I'll deal with it next week," here's the part that should ruin your Saturday plans: any title still over the limit after the 27th doesn't get hidden. It doesn't get flagged. It gets rewritten. By Amazon's AI. Without asking you. The platform's models will quietly replace your words with their words, listing by listing, like a very efficient copywriter who doesn't know your product, your customer, or why you specifically chose "industrial-grade" instead of "heavy-duty."

This isn't a formatting tweak. It's a fundamental change in who controls what your product says about itself.

The Two-Room Apartment (Where You Used to Have a Warehouse)

Let's talk about what's actually changing, because the detail is where the strategy lives.

Your title caps at 75 characters. A second field — Item Highlights — takes up to 125 characters of the supporting detail that used to live in the title. Materials, use cases, comparison terms, all the stuff you previously crammed into one overflowing string like you were packing for a two-week holiday using only a carry-on.

Here's the bit that matters: Item Highlights isn't decorative. Amazon has confirmed it's searchable and displayed beneath titles in both search results and on product detail pages. Early testing suggests it's weighted more heavily than generic back-end search terms. So the keywords you move there aren't sacrificed. They're relocated.

Your old title was a warehouse — you threw everything in there and hoped the algorithm would sort through the pile. The new setup is two rooms with different purposes. A short, human-readable title carrying your brand and primary keyword, and a structured field for everything that used to make your title read like someone had a seizure on the keyword research tool.

One critical difference that's easy to miss: sellers in Brand Registry get a 14-day window to review, modify, and approve any AI-generated rewrite before it goes live. Sellers without brand registration? They appear to get no review window at all. For them, Sunday is simply a hard line after which the platform starts editing on their behalf. (More on this asymmetry in a minute, because it's doing a lot of quiet work.)

The Art of Deciding What Dies

For a decade, title strategy on Amazon rewarded a specific and slightly inelegant skill: cramming the maximum number of relevant search terms into a single field while staying just inside the character limit and just inside the policy line. Entire agencies built workflows around it. That skill is now largely obsolete.

The replacement is quieter and harder to automate — the judgement to decide what survives compression. When only 75 characters remain, every term has an opportunity cost. That brand name nobody actually searches for? It's no longer a sensible default at the front of your title. It's 15 or 20 characters that could carry a converting keyword instead.

The consensus forming among experienced sellers is unsentimental: if people actively search your brand name, keep it up front — it earns a bold header and functions as a query in its own right. Everyone else is better served leading with the primary keyword and their single most differentiating attribute, then pushing the brand into Item Highlights. Whether that trade is right for your specific catalogue is exactly the kind of decision that cannot be delegated to a rewrite tool that has no idea which terms actually drive your impressions.

And I want to name something that's easy to smooth over in a "just adapt!" narrative: for industrial and technical sellers, this is a genuine loss. In categories where the title historically carried sizing, material grades, fitting types, and dimensional data — information that is defining rather than supplementary — 75 characters forces a buyer to click through to the detail page for what used to be visible in search results. That's a real cost. It falls unevenly. And pretending it doesn't exist doesn't help anyone figure out how aggressively to lean on Item Highlights to compensate.

The Great Flattening Reaches the Title

Regular readers will recognise the second-order effect immediately. We've spent much of this year talking about the Great Flattening — the way AI systems pull brand communications toward a shared average, so that content converges on a common register and the signals that once separated one brand from another quietly erode. Like watching every restaurant on the high street morph into the same grey-and-wood minimalist aesthetic, except it's happening to your listings.

Compression alone pushes titles toward sameness. When every listing in a category gets squeezed to the same skeleton — brand, product type, key attribute, size — the structural room for distinctiveness shrinks. Layer an AI rewrite on top, applied from the same underlying model, and the convergence accelerates. Sellers who let the deadline pass aren't merely accepting a shorter title. They're accepting a title written to look like everyone else's, by a system optimising for compliance rather than for their particular customer.

The honest counterpoint (and I do try to have these, even when they're inconvenient to my argument) is that a good number of operators welcome exactly this. Their case: keyword-stuffed titles had already flattened the category on their own — when everyone stuffs, everyone reads the same — and Amazon's indexing is now sophisticated enough to understand what a product is without the term appearing verbatim in the title. On this view, a ranking that depended entirely on long-tail keywords buried in a title string was a fragile position to begin with, and the limit simply exposes it.

Both things can be true at once. The change removes a crude lever and narrows the space in which a distinctive listing can distinguish itself. What it rewards is the brand that had something specific to say before the AI touched anything. Which is the same conclusion the flattening keeps returning us to, like a recurring dream where you show up to a party and everyone's wearing the same outfit. Including you.

Who Holds the Pen (Spoiler: Probably Not You)

Here's the part that should actually keep you up at night. The most consequential detail isn't the character count. It's the enforcement mechanism.

Eleven years ago, when Amazon imposed a 200-character limit, it policed the rule by suppressing non-compliant listings in search. Blunt, yes. But it left your words intact. This time, the platform isn't suppressing. It's rewriting. The models don't hide your title. They replace it.

That is a shift in who controls the presentation layer, and it doesn't sit in isolation. Think about the sequence: the 200-character limit in January 2025. Agentic capabilities rolling across Seller Central, turning the Seller Assistant from a tool that answered questions into a system that monitors accounts, manages inventory, and runs campaigns — generating, by Amazon's own figures, more than 12 million sales-ready listings with an acceptance rate above 90%. New AI-agent rules in the Business Solutions Agreement in early 2026. Discount and list-price presentation tightened in April. And now the title limit.

It's the same direction of travel, just arriving at different fields at different times: greater platform control over how products appear to shoppers, with AI as the instrument of compliance. Each step is individually reasonable. The pattern, taken together, is a marketplace steadily reclaiming the presentation layer from the merchants who use it.

Read that way, the 14-day Brand Registry review window stops being an administrative footnote and becomes a line in the sand. Registered brands retain a veto. Unregistered sellers are handed a result. For a marketplace that many smaller sellers already suspect is engineered to favour scale, that asymmetry reads as one more structural nudge — and a low-cost, high-consequence reason to complete brand registration that has nothing to do with the character count itself.

The Advertising Bomb Nobody's Defusing

None of this is cleanly separable from paid performance, which is why your media buyer should treat Sunday as an advertising event, not just a content one.

Amazon pulls heavily from titles for keyword indexing. Sponsored Products relevance depends in part on how a listing's data is structured. A title rewritten without seller input carries an unknown keyword configuration — one that may or may not preserve the long-tail terms your campaigns have been targeting. If the model removes or reorders those terms, the open question is whether the listing keeps its relevance signal through Item Highlights. And the precise weighting the ad system gives that field relative to the title? Simply not documented at this level of detail.

The practical implication is unglamorous and firm: your listing should be in its final, seller-controlled form before Sunday, not after. That protects organic ranking stability and — just as importantly — anchors paid targeting to a content structure you chose rather than one the platform generated. It's the difference between running campaigns against known variables and reverse-engineering what the AI decided on your behalf. (And if there's one thing I've learned about reverse-engineering Amazon's decisions, it's that the algorithm has never once texted back to explain itself.)

What You Should Actually Be Doing This Weekend

The work is triage, and it's finite. Start with the listings that carry the volume and the ad spend. Decide, listing by listing, which single keyword and which one differentiating attribute earn the scarce space in the 75-character title. Move everything else — deliberately — into Item Highlights, treating it as a structured field with its own priorities rather than a dumping ground for whatever the title couldn't hold.

If you're not in Brand Registry, that gap is now the difference between reviewing the platform's rewrites and inheriting them. It's worth closing on principle, even if the paperwork lands after Sunday.

And once the 27th passes, the job shifts from editing to watching: monitor what the AI proposes for anything left non-compliant, and compare ranking, conversion, and impression share before and after the change rather than assuming the rewrite is neutral. (It might be fine. It also might have quietly removed the exact term that was driving 40% of your organic impressions. You won't know unless you check.)

The Bottom Line

The title has quietly become the first field Amazon has fully reclaimed from sellers — the first place where the platform's models, not the merchant, get the last word by default. How Item Highlights ultimately behaves inside the ecosystem is still unsettled: how Rufus reads it, how it's weighted in sponsored auctions, how it surfaces across devices. Those answers will come only through deployment and testing.

What's already settled is the direction of travel. The presentation layer is being standardised, and the instrument of standardisation is AI. The 75-character limit isn't the end of that process. It's the clearest sign yet of where it's heading.

The sellers who do the triage themselves this weekend are the ones who keep a hand on the pen a little longer. Everyone else gets to find out what Amazon's AI thinks their product should be called.

Which, honestly, is the most passive-aggressive power move a marketplace has ever pulled. "Oh, you didn't update your title? Don't worry. We wrote one for you. You're welcome."

• Google adds a "How this ad was made" panel to My Ad Center so users can check if ads on Search, YouTube and Discover used AI. Meta tags AI ads under "About this ad", TikTok labels video, Microsoft reads C2PA credentials.

• Reddit ads exec Roelof van Zwol admits the company is intensely debating whether licensing content to OpenAI and Google undercuts its own targeting edge. His answer for now: licensed posts are not the engagement signal.

• OpenAI preps a screen free smart speaker as its first device, built with Jony Ive's LoveFrom, packing a camera, sensors, battery and moving parts. It unveils this year and ships 2027, pending Apple's trade secret suit.

• Essay argues agent workforces fail like human ones: token loops mirror meeting bloat, context hoarding becomes job security, evals replace OKRs. The trillion dollar prize goes to AI transformation firms, not neofirms.

• Axios, Forbes, Time and The Washington Post package AI visibility into GEO products for brands, though the metric stays unstandardized. Time sells bot traffic instead, sitting in TollBit's 98th percentile for crawler demand.

• Growth Memo maps an AEO ladder of Retrieved, Cited and Trusted after finding only 1% of users click AI Overview citations and 70.6% of AI referrals log as direct in GA4. Freeze 20 to 50 prompts for four weeks first.

• GPT-5.6 Sol tops Design Arena web design, 18 places above GPT-5.5, with embedding maps showing holes where purple gradients and bento layouts sit. It still overuses confetti in 26.5% of generations.

• Google rolls connected apps into AI Mode for US Search, letting users push groceries to Instacart, pull Canva templates and save playlists to YouTube Music without leaving the answer.

🛒 MESA - AI-driven workflow automation built for Shopify stores.

🌐 Perplexity - AI answer engine with inline citations from real-time web search.

🖥️ Cursor - AI-native code editor and pair programmer for shipping faster.

🎶 Artlist - Royalty-free music, footage and AI creative tools for video.

🤖 Lindy - Build AI agents that handle email, scheduling, lead research and follow-ups.

😍It was great, all I needed

🙂Good, not great

🙄It sucked