Blog · Takes
Takes
Points of view from the Lily AI team on where retail discovery is going and what actually moves revenue.
TakesA buying standard for AI commerce
Four questions to ask before you fund any AI marketing tool. Every retail budget now carries an AI line item, and almost nothing on it arrives with a way to know whether it worked. Here's the standard, and the one shortcut to remember it: would it survive a finance review?
TakesThe most expensive thing in retail right now
Brands are pouring attention into a future that isn't generating revenue yet, while the surfaces actually producing revenue today get treated like settled infrastructure. A preview of my CommerceNext session with Ken Pilot and Noam Paransky.
TakesYou don't have an agency problem. You have an input problem.
A CMO fired three paid agencies in two years and decided you can't find a good one anymore. The real problem was upstream: the old agency edge has been commoditized, and the leverage has moved to the inputs only the brand controls.
TakesGoogle says conversational attributes are optional. History says otherwise.
Mobile-friendly, page speed, structured data: Google introduced each as optional, and each became the price of entry. Conversational attributes, its new AI-shopping feed fields, look like the next one.
TakesIt always turns out to be a feed
OpenAI spent a year pitching conversational shopping, then quietly rebuilt something that looks a lot like Google Merchant Center. Every platform that monetizes shopping ends up needing the same thing first: a feed.
TakesGoogle may have just created a billion-dollar optimization market
AI visibility struggled to attract serious budget for a simple reason: nobody funds what they can't measure. With Google's new AI Performance Insights, it's measurable now. So the money moves.
TakesAutomation commoditized everything except your feed
Google spent five years making paid search dramatically easier to run, which is exactly why it keeps getting more expensive to win. A classic Jevons paradox: automation standardized the edge brands used to build by hand. The one input it can't commoditize is your product feed.
TakesGoogle just told the market AI content isn't the moat. Structured product data is.
Google's official guide on optimizing for AI search quietly dismissed the llms.txt-and-content-variants playbook a whole vendor category was built on. For retailers, the moat isn't AI-optimized content. It's structured, machine-readable product data.
TakesStripe is building the rails. Your product data decides who wins on them.
A day at Stripe Sessions made one thing unmistakable: agentic commerce has moved from interesting future to build now. Stripe is building the payment rails. The brands that win on them will already have their product data structured for agents.
TakesBrands are paying for bad product data twice now
Adobe says 34% of retail product pages can't be properly read by AI, even as the AI channel converts 42% better. The cost of thin product data used to be invisible. Answer engines just made it impossible to ignore.
TakesNRF 2026 Takeaways: Retail Is Being Rewired Around AI Discovery
NRF 2026 made clear that AI is no longer a nebulous concept: it is already reshaping how retail teams approach discovery, personalization, and growth. Here is what stood out and why it matters.
TakesFrom Experimentation to Expectation: How AI Shopping Normalized Faster Than Retail Prepared
Over the course of 2025, AI-powered shopping moved from cautious experimentation to everyday utility, shifting how consumers search, discover, and decide what to buy.
Takes6 Things Every Retailer Needs to Know About Agentic Shopping Right Now
Consumers no longer type short keywords; they express intent in full sentences and AI agents reason through them. Six takeaways on the shift from ranking to reasoning.
TakesThe Agentic Wake-Up Call for Retail: Why CMOs and CDOs Must Prioritize Product Data Now
Andrej Karpathy called this the “decade of agents.” For retail CMOs and CDOs it is a runway, not a buffer: agents only perform atop rock-solid, machine-readable product data.
TakesThe Agentic Commerce Era Has Officially Begun
Walmart has announced a partnership with OpenAI that lets shoppers discover and buy directly through ChatGPT. Product content optimization is no longer operational; it is existential.
TakesEvery Discovery Is AI Discovery
The way you search on Instagram isn’t the way you search on Amazon. Each query describes the same goal in a different language—every platform speaks its own dialect of intent.
TakesIn the Instant Checkout Era, Product Content Is Everything
Everyone’s buzzing about OpenAI’s Instant Checkout, but the real winners won’t be the ones with the flashiest tech stack. They’ll be the brands that finally fix their product content.
Takes7 Hot Takes on AI from Female Execs & Founders at Cannes Lions
At Cannes Lions this year, Lily AI CEO Purva Gupta joined a panel of women founders to discuss the AI ecosystem in advertising and media. Here are 7 hot takes from the session.
TakesQ2 2025 Retail Trends: 3 Opportunities for Google, Meta, SEO, AIO, AEO & GEO
Q2 2025 retail data across digital advertising and organic search was marked by both ups and downs. Synthesizing the quarter reveals three golden opportunities for marketers right now.
TakesAI Slop vs Creme-of-the-Crop: Optimizing Product Content Generation with AI
AI slop is low-quality, filler-heavy content generated with AI, and it is flooding the digital landscape. Why it matters for retail, and how to reach the creme of the crop instead.
Takes5 Times Google and Meta Disrupted Organic Traffic & SEO, and What’s Next for Retail
Brands are seeing 30% drops in organic traffic as AI Overviews fuel zero-click search. A look back at five times big tech disrupted SEO, and three predictions for retail.
Takes3 Ways GEO vs SEO Tactics Converge, and the Role of Product Content Optimization for Retailers
Retail executives keep asking how different GEO and AEO really are from traditional SEO. The answer is complicated, but content is where the two converge most tightly.
TakesRetailers, Is Your Product Content Ready for GEO?
Generative search is rewriting the rules of SEO right before our eyes. Winning now means being included in the answer itself, with accurate, on-brand, consumer-relevant information.
TakesWhat Do Retailers Need to Know About Google AI Mode’s Query Fan-Out Technique and Thematic Search?
Google's AI Mode and its thematic search patent threaten keyword-based SEO, but they also open a pivotal opportunity for retailers who invest in richer, consumer-centric product data.