On the go? Listen to this roundup in audio format.
Welcome back to the Retail Roundup, your monthly recap of the biggest developments shaping retail media, marketplaces, and eCommerce strategy.
This month brought a concentrated set of AI updates, media platform upgrades, and Prime Day prep signals. Here's what matters and what to do about it.
- The AI Corner: Big Moves, Bigger Questions
- Amazon Launches Dynamic Canvas Inside Seller Central
- Amazon Ranked #1 in Omnichannel Advertising (Forrester Wave)
- Netflix + Amazon DSP Integration Unlocks Real Audience Targeting
- Amazon DSP Expands Performance+ Insights
- Prime Day 2026 Deadlines Move Earlier
Let’s dive in.
1. The AI Corner: Big Moves, Bigger Questions
Four major announcements set the tone this month, and while they look disconnected on the surface, they circle a single question: Is AI a place to transact, a place to discover, or a place to influence decisions?
- OpenAI is reportedly scaling back its agentic commerce and instant checkout ambitions
- A massive $50B partnership between Amazon and OpenAI signals deeper infrastructure collaboration via AWS
- Criteo joins ChatGPT’s ad pilot, bringing performance advertising into AI conversations
- Walmart integrates its “Sparky” assistant directly into ChatGPT
- Open AI Scales Back Shopping Plans
OpenAI is pulling back on its agentic commerce and instant checkout efforts. On paper the ambition made sense: if users are discovering products in ChatGPT, why not let them buy there too?
It didn't stick. The problem was partly technical: normalizing product data across millions of ecommerce sites is a massive challenge. But the deeper issue was behavioral. ChatGPT users were researching products to buy in the chatbot, but they weren't using it to actually make purchases.
We've seen this before. When Facebook and Amazon integrated to let users check out with Amazon directly inside the Facebook app, almost nobody used it. The feature existed, but that's not why people opened the app. Just because a platform can support a behavior doesn't mean users will adopt it. If it doesn't clearly make their lives easier or solve a problem, there won’t be uptake.
- Amazon and OpenAI Agree to $50B Investment
At the infrastructure level, Amazon and OpenAI have announced a $50B partnership, most likely centered around AWS compute.
This is easy to over-read. Despite speculation, this doesn't mean Amazon listings will start appearing in ChatGPT anytime soon. Amazon remains a walled garden - it has its own AI shopping assistant in Rufus and continues to prioritize its internal ecosystem.
This is mostly a technology partnership to enable a bunch of compute for OpenAI using AWS. We don't think that tomorrow we'll start to see Amazon listings in ChatGPT. That's still a long roadmap ahead of us, if it even happens at all.
Think of this as long-term technological enablement. It strengthens OpenAI's ability to scale but doesn't redefine the user experience yet.
- Criteo Joins OpenAI ’s ChatGPT Ad Pilot
Next, Criteo became the first adtech partner in ChatGPT’s ad pilot. This is where things start to get interesting. Under this model:
- Brands can run performance-driven ads inside ChatGPT conversations
- Ads are informed by Criteo’s commerce data and retailer relationships
- Users are reached while actively asking for product advice, comparisons, or recommendations
In theory, this is incredibly powerful: ads delivered at the exact moment of purchase intent. But the same behavioral question applies. Does this actually improve the user experience? If it doesn't, it risks becoming another feature that exists but never gets adopted.
- Walmart's Chatbot Just Moved Into ChatGPT
While OpenAI is pulling back on transactions, Walmart is leaning in, embedding its "Sparky" assistant directly inside ChatGPT. This "chatbot inside a chatbot" approach is a meaningfully different bet.
Instead of making ChatGPT the checkout layer, Walmart is inserting itself into the decision-making layer. It's less about owning the transaction and more about influencing it.
Pros
- AI is proving to be a powerful discovery layer. Users are clearly comfortable using tools like ChatGPT for research, comparison and product exploration. That’s a high-intent moment brands can tap into.
- Advertising is getting closer to intent than ever. With partners like Criteo, brands can reach users in real time during active decision making. That’s a step beyond traditional search or social targeting.
- Multiple strategies = more opportunity. With OpenAI, Amazon, and Walmart taking different approaches, brands aren’t locked into one path.
Cons
- User behavior doesn’t match platform ambition (yet). People don’t want to complete purchases inside AI (for now). That limits monetization options.
- Too many unknowns. Will ads feel helpful or intrusive? Will users trust AI recommendations enough to buy? Which platform approach will win? There’s no clear answer yet.
- Fragmentation across platforms. Different players are building different systems. This makes it harder for brands to standardize strategy.
- Data challenges are still real. AI depends on clean, structured and consistent data. Most brands are not fully there yet.
Tips
- Focus on discovery, not checkout. Right now, AI is strongest at influencing decisions, not completing transactions. Optimize for being recommended, not just purchased.
- Solve the iceberg problem. Most brands only show the tip: basic keywords, titles, and bullet content. AI needs the whole story. Turn customer reviews and Q&A data into structured insights a bot can actually use. When you give AI the full picture, it has the confidence it needs to recommend you.
- Clean up your product data. Make sure specs are accurate, attributes are consistent and no conflicting information exists. AI won’t recommend what it doesn’t trust.
- Prepare to win the recommendation slot. When AI suggests products, it shows a few options and the top one gets most of the clicks. That means you’ll need competitive pricing, clean inventory data and reliable shipping.
- Move fast. In the AI era, the brands that are punished the most will be the ones where content and website changes take months for approvals.
- Don't place a single bet. You can't bet on one retail channel or one LLM. Get the fundamentals in place so that as consumer behavior shifts, you're ready.
2. Amazon’s Dynamic Canvas: An AI-Native Workspace Inside Seller Central
Looking at old Seller Central sometimes felt like trying to read the Matrix in 1999. The endless columns, the numbers, the vibe. That's what Amazon has just replaced.
Dynamic Canvas is a new AI-powered workspace inside Seller Central. But it's not just a dashboard refresh. Under the hood, it's built on agentic architecture powered by Amazon Bedrock, Nova models, and Anthropic Claude. In practice, that means the system doesn't just display data, it reasons through it. Think of it, as a data scientist living inside your laptop who never takes a lunch break.
In practice, sellers can:
- Build custom views using 40+ pre-built templates (inventory health, ad efficiency, traffic, and more)
- Visualize data through heat maps, scatter plots, and donut charts
- Ask questions like "What product should I launch next?" and explore what-if scenarios
- Share view-only links internally or with clients - no export required
That last feature is an immediate workflow upgrade for agencies and brand teams presenting insights to stakeholders. And more broadly, this continues Amazon's pattern of building analytics natively, reducing reliance on third-party tools.
Pros
- Lowers the barrier to meaningful data analysis - no technical background required.
- Shareable links add an immediate workflow upgrade for presenting insights to stakeholders.
- Continues Amazon's pattern of building analytics natively, reducing reliance on third-party tools.
Cons
- Early-stage releases from Amazon often have rough edges - expect some instability before it matures.
💡 Tips
- You don’t have to be a math genius to start. Get in early and test pre-built templates against your existing reporting to identify gaps.
- Use the shareable link immediately for client or stakeholder presentations — it's a free workflow upgrade.
- This is likely the foundation for deeper AI integration. Keep an eye on what comes out next - early adoption will compound.
3. Amazon Named Leader in Omnichannel Advertising
Amazon was ranked as the leader in omnichannel advertising platforms in the latest Forrester Wave report, ahead of major players like Google and The Trade Desk.
Amazon Ads was among seven companies Forrester selected to research, analyze, and score based on their current offering, strategy, and market presence.
“Amazon Ads' ‘authenticated graph’ enables advertisers to deterministically reach approximately 90% of households across the US. Best-in-class audience creation and targeting capabilities are boosted by Ads Agent and Creative Agent, which lift media and creative SMEs’ workflow and analytical efficiencies.”
This is powered by:
- Amazon Audiences
- DSP integrations
- Signals across retail, media, and entertainment properties
This isn’t just a product milestone, it’s a perception shift. For years, many brands have treated Amazon as a bottom-of-funnel conversion channel with a playbook that looks like: Drive traffic via Meta, influencers, or Google and convert that traffic on Amazon. But this way of thinking is now outdated. Amazon has evolved into a full-funnel ecosystem, where brands can:
- Reach incremental new to brand shoppers via streaming TV across both Amazon owned and operated properties (Prime Video, Twitch) as well as third party supply platforms and publishers (e.g., Netflix, Roku, Microsoft, Spotify, NBC Sports, CNN, NFL Network, A&E, FOX, etc.)
- Customize audience creation and targeting using Amazon Marketing Cloud to leverage the billions and billions of shopping signals Amazon can provide
- Convert them either on-Amazon or on a brand’s .com with link-out DSP campaigns
- Retain them through loyalty programs (e.g., Subscribe & Save)
- Create lookalike audiences off of existing high CLTV customer cohorts to inform the next incremental audience segment
In other words, Amazon is no longer just where demand is captured, it’s where demand is generated and nurtured. But what makes Amazon uniquely defensible is data quality. Unlike platforms that rely on probabilistic signals, Amazon operates on real purchase behavior, actual browsing history, and logged-in user data. That creates a fundamentally different level of targeting precision.
Pros
- Independent validation strengthens the internal case for expanding Amazon DSP investment.
- ~90% authenticated STV household reach (everything except for YouTube and YouTubeTV) makes Amazon Ads a powerful upper-funnel buying platform.
Cons
- Growing recognition and budget share might push CPMs higher over time.
- Full-funnel measurement requires moving beyond ROAS - a shift many teams haven't made yet.
💡 Tips
- Use the Forrester ranking as internal leverage when making the case for Amazon DSP budget.
- Build a measurement framework that includes new-to-brand rate and customer acquisition cost - ROAS alone undersells DSP's value.
4. Netflix Taps Amazon’s Shopping Data
Amazon and Netflix have extended their partnership inside Amazon DSP, but the key unlock is not access to inventory. That part already existed. What’s new is this: Brands can now apply Amazon Audiences when buying Netflix inventory through DSP.
Phase 1 (already existed): Advertisers could buy Netflix ad inventory through Amazon DSP. Broad reach, but limited targeting refinement.
Phase 2 (new): Advertisers can now layer Amazon's first-party audience data on top of that Netflix inventory.
Those Amazon audiences are built from billions of real shopping signals — what users searched for, clicked on, added to cart, purchased, and how they engaged with ads over time. This is "the real juice," as Pat Petriello put it: access to Netflix audiences was high-value on its own, but being able to apply Amazon targeting within that inventory is the whole ballgame.
In practice, this means brands can now:
- Target users who browsed their category but didn't purchase
- Reach shoppers who clicked on a product detail page in the last 6 months
- Build highly specific audience segments using AMC data
- Exclude past purchasers - ensuring campaigns are truly incremental, not retargeting existing customers
- And most importantly: Exclude past purchasers. This way brands can ensure campaigns are truly incremental, not just retargeting existing customers.
This also signals something bigger. The question for media planning is evolving: from "which keywords should I bid on?" to "which audiences should I reach, and where do they spend time?" and eventually "which shows and platforms drive the strongest response from my audience?" Netflix is just the start. Expect Amazon audience targeting to expand across more streaming platforms, more publishers, and the broader open web.
Pros
- Among the most precise upper-funnel CTV targeting available today.
- Audience negation makes spend truly incremental - a standard most CTV buys can't meet.
- One platform for buying, targeting, and measurement.
Cons
- Requires a mature AMC practice to build effective audiences - expect a ramp-up if you're starting from scratch.
- Premium streaming inventory means CTV-level CPMs.
Tips
- Start building AMC audiences now. The more historical data available, the stronger your targeting at launch.
- Prioritize audience suppression - negating past purchasers is the highest-leverage move for incremental reach.
- Run a small Q2 test before scaling. Measure new-to-brand rate and CAC, not just impressions.
- Build your audience architecture with multi-publisher deployment in mind - Netflix is just the start.
5. Amazon DSP Upgrades: Closing the Measurement Gap
Amazon has expanded the Performance+ insights available directly inside DSP, surfacing metrics that advertisers previously had to piece together manually. The two most important additions are:
- Time-to-conversion - how long it actually takes each tactic to drive a purchase
- Shopper audience segmentation - breaking down who is converting and when
Historically, accessing this level of insight required Amazon Marketing Cloud (AMC) queries, manual exports and reporting workflows, and stitching together multiple data sources. Now, these insights are becoming native to the DSP interface, particularly at the order and tactic level.
The most compelling use case is comparing time-to-conversion across all P+ tactics in a single view. That comparison matters because different tactics naturally reach users at different funnel stages: upper-funnel tactics have longer conversion windows, lower-funnel tactics shorter ones. Having that data centralized changes how you sequence campaigns, allocate budget, and set expectations with clients.
This is especially important as performance is no longer judged purely on immediate ROAS. Metrics like customer acquisition cost, new-to-brand growth, and assist value are becoming central to decision-making. Not every dollar is supposed to return immediately, and now that’s easier to prove.
This update also reflects a broader philosophical shift. Amazon DSP has traditionally leaned toward a “black box” model, where automation handles optimization, but transparency is limited. With these new insights, Amazon is moving toward a “glass box” approach: You still benefit from automation, but now you can see who's being targeted, when they convert, and how different tactics compare.
Amazon essentially had to build this. If you're telling brands to invest in Netflix inventory and Amazon audiences, you also need to be able to prove it worked. Without credible measurement, the pitch isn't complete.
Pros
- Time-to-conversion across all P+ tactics in one view enables smarter budget allocation without AMC expertise.
- Reduces manual reporting burden for agencies and in-house teams.
Cons
- More data is only valuable if teams have the bandwidth to act on it.
💡 Tips
- Pull the time-to-conversion comparison across your P+ tactics now. Wide variance signals a need to revisit how you're sequencing upper- and lower-funnel tactics.
- Add this to your standard reporting cadence. Reporting P+ results without conversion timing context is no longer defensible.
6. Prime Day 2026: The Calendar Shift Is Already Happening
Amazon hasn't announced official Prime Day dates yet, but the confirmed milestones already signal a structural shift in timing. The homework moved earlier, and brands that haven't started prep are already behind.
Key dates confirmed:
- March 24: Deal submissions opened
- End of April: Submit deals to capture the $50 per-deal discount
- May 26: Deal submission window closes
- May 27: AWD shipment deadline + minimal split FBA cutoff
- June 5: FBA shipments using Amazon optimized splits must arrive
The Biggest Change: Simplified Fees
Previously, different deal types had different fee structures and you had to do the math for each. This year, Amazon standardized it across Best Deals, Lightning Deals, and Prime Exclusive Discounts (PEDs):
- $100 upfront fee per deal
- +1.5% variable fee on sales
- Capped at $5,000 - high-velocity SKUs are protected from uncapped exposure
Stricter Pricing Rules
Eligibility requirements tightened:
- Deal price must be at or below your lowest price in the past 60 days
- Must also be at least 5% below your lowest price in the past 30 days
The practical implication: if you're planning a spring promotion in April, wrap it up by mid-April to preserve your 60-day eligibility window. Run a discount too close to Prime Day and you'll either miss qualification or be forced to discount even deeper.
Pros
- Simplified fee structure makes deal planning more predictable.
- The $5,000 cap protects high-velocity SKUs from uncapped variable fee exposure.
- PEDs allow up to 500 SKUs per promo - a structural advantage for large catalogs.
Cons
- Strict price history requirements limit promotional flexibility in the lead-up period.
- No confirmed date makes media budget planning imprecise.
💡 Tips
- Start now. Inventory timelines, deal submissions, and price stability windows all require immediate action.
- Stabilize pricing. End spring promotions by mid-April to protect your 60-day eligibility window.
- Prioritize Best Deals, then PEDs especially for large catalogs (Prime Exclusive Discounts (PEDs) let you run promos across hundreds of SKUs at once-huge win for bigger catalog brands).
- Lock in inventory deadlines. May 27 → AWD cutoff + minimal split shipments. June 5 → Optimized FBA shipments must arrive.
- Capture the $50 discount by submitting deal fees before the end of April.
Subscribe for More!
We publish a new Retail Roundup every month. Stay in touch on LinkedIn to keep track of the most important changes in the Retail media landscape.
Give It a Listen
You can tune in for the full interview with Armin Alispahic and Pat Petriello on the Ecommerce Braintrust hosted by Julie Spear and Jordan Ripley.
This show gives you access to the world's best brains when it comes to building momentum online for established consumer brands. Join in and listen to discussions with expert guests about e-commerce strategies, trends, and innovations.
Read more from this author: