Andy Ellwood on Building Stretch After Basket

Andy Ellwood on Building Stretch After Basket

Andy Ellwood returned to grocery technology because the problem survived his first company. Basket had already tried to make grocery prices easier to compare. Years later, the data environment had changed, but households still had to reconcile price, distance, brand preference, availability, and time.

In episode 99 of Venture Step, Ellwood explains why that combination drew him back. The conversation also reveals a more demanding standard for grocery technology. A useful product cannot merely display a low price. It has to show which product, at which store, at what time, under which conditions, with enough context for a person to decide whether the trip is worth making.

Basket proved that the problem could outlive the company

Ellwood says he began thinking about grocery price transparency in 2013 and later co-founded Basket. In the interview, he describes a large crowdsourced catalog and hundreds of thousands of families using the product. Those scale figures are his recording-time account and have not been independently audited by Venture Step.

His current founder history lists Basket from April 2014 through December 2019 and labels it acquired. It describes the company as a smart shopping list built to bring grocery pricing transparency to shoppers. The same page claims weekly savings above 30 percent. That is a first-party historical claim, not a current performance benchmark or an independent finding.

The more durable point is that Basket did not settle the underlying problem. Grocery prices remained fragmented by retailer, store, channel, membership, promotion, package, and location. A shopper could still see an attractive item price without knowing whether the entire basket, trip, or delivery would be cheaper.

Ellwood says the pandemic changed how grocery data was structured and made a second attempt materially different. That is his explanation for timing, not a technical account of every data source. Still, it separates persistence from repetition. Returning to the same market can make sense when both the unresolved need and the enabling conditions have changed.

flowchart LR
    A["Basket tests grocery transparency"] --> B["Company ends, problem remains"]
    B --> C["Retail and data environment changes"]
    C --> D["Stretch starts with a grocery list"]
    D --> E["Price, distance, preference, and availability"]
    E --> F["A household decides which trip is worthwhile"]

The final decision is the product. The list and prices are inputs.

Stretch starts with the list a household already has

Stretch's current website describes the service as an AI-powered grocery list maker. It says the product can scan lists, receipts, or notes, track nearby stores, learn from repeated use, and show prices and locations before a shopper leaves home. Its current US App Store listing names the product “Stretch: Save on Groceries,” identifies Stretch Collective as the developer, and lists the app as free and available for iPhone.

Those are company and platform descriptions, not a Venture Step recommendation. The company does not promise in its legal terms that every displayed price is current or complete. Stretch's terms of service say its data depends on information displayed or provided by retailers and manufacturers. Prices and coupons can change, and users remain responsible for verifying an offer before purchase.

That limitation is not incidental. A comparison product sits between a volatile retail system and a household decision. It has to resolve misspelled list items, package differences, brand preferences, store locations, promotions, missing products, and data freshness. A confident answer can hide uncertainty unless the product makes those boundaries visible.

Dalton's test exposed the importance of distance

In the episode, Dalton describes using Stretch in New York City. He says the onboarding took about four minutes and asked about household preferences, including brands. The result surfaced stores and prices, but his practical reaction was not simply to choose the cheapest basket.

Distance mattered because he did not have a car. A store farther away could require more walking, transit, or an additional trip. Ellwood calls this “proximity pricing preference,” a useful name for a familiar decision. The amount a person is willing to travel for a lower price varies by location, mobility, schedule, weather, and basket size.

That observation prevents a price comparison from becoming false precision. Ten dollars of item savings may be attractive for a weekly stock-up and irrelevant for a forgotten ingredient. A delivery option may cost more in fees but be preferable for someone with limited mobility or caregiving responsibilities. A familiar nearby store may reduce the chance of substitutions and a second stop.

The right comparison is therefore not “Which store has the lowest displayed total?” It is “Which available option best fits this household's basket and constraints today?” [[How to Compare Grocery Prices Across Stores]] turns that question into a repeatable method.

The pricing controversy needs exact language

The episode discusses a 2025 Consumer Reports investigation conducted with Groundwork Collaborative and More Perfect Union. According to the Consumer Reports report, 437 volunteers in four cities ran coordinated Instacart shopping sessions. The investigation found that many checked items appeared at more than one price during simultaneous comparisons. It reported item differences as high as 23 percent and used an observed basket spread to estimate a possible annual household impact.

The annual figure was an extrapolation, not a bill observed for every family. Instacart's published response said the tests were short-term and randomized, that retailers controlled pricing strategies, and that personal, demographic, and individual behavioral data were not used to assign item prices. The company rejected the annualized estimate and disputed descriptions of the tests as dynamic or surveillance pricing.

The observed variation and the causal explanation are different evidence questions. Simultaneous price differences show that some shoppers were offered different values. They do not, by themselves, show which data determined assignment. [[Algorithmic Grocery Pricing - What Shoppers Can and Cannot See]] separates price testing, dynamic pricing, personalized pricing, and surveillance pricing without treating one phrase as a substitute for proof.

New York's attorney general later demanded information from Instacart. That January 2026 action was an inquiry and warning about the state's disclosure law, not a final finding that personal data caused the price differences in the study.

The FTC also announced a separate $60 million Instacart settlement. That matter concerned allegations involving delivery advertising, refunds, satisfaction guarantees, and subscription enrollment. It did not settle the Consumer Reports price-testing dispute. Combining the two events would produce a cleaner story and a less accurate one.

A digital pantry is a trust relationship

Ellwood describes a product that becomes more useful as it learns what a household buys. In the interview, future possibilities include alerts and integrations. Those recording-era ideas should not be presented as current features unless the product documentation confirms them.

The live privacy record shows why the distinction matters. Stretch's current privacy policy says the service may collect shopping lists, searches, purchase history, receipts, preferred stores, linked loyalty-account data, general or precise location, device identifiers, demographics, income, and budget information. It describes uses that include personalization, eligibility, analytics, location services, offers, advertising, and marketing. The policy also describes jurisdiction-dependent access, deletion, correction, opt-out, and limitation rights.

This does not prove that every listed category is collected from every user. A privacy policy defines permitted practices and possible data flows, not a usage log. It does establish that a “digital pantry” can become a detailed household record. Product usefulness and data governance must be evaluated together.

[[What a Trustworthy Digital Pantry Requires]] maps the product identity, price freshness, preference correction, consent, retention, and explanation layers that belong behind a simple list.

The ten-year test explains the return

Ellwood says he developed four non-negotiables before becoming a founder again. The one he explains most fully is that the work had to address a problem he could think about for ten years. Grocery transparency passed partly because he had already spent more than a decade returning to it.

That is a commitment test, not proof of demand or virtue. A founder can remain fascinated by a weak market, and a valuable problem can still produce a poor company. Ellwood also uses the familiar “painkiller versus multivitamin” analogy to ask whether someone will interrupt the day to solve the problem. The analogy is useful when it forces a founder to identify urgency, but it is not a scientific rule. Preventive products, infrastructure, and long-horizon services may matter before users feel immediate pain.

The stronger version combines commitment with evidence. A founder should know whose problem this is, what people do now, why the situation has changed, what can be tested, what economics may work, what harm the product could create, and what result would justify stopping. [[How to Choose a Problem Worth Ten Years]] makes those questions explicit.

The useful promise is a more inspectable decision

Stretch's current website promises to do much of the searching before the shopper leaves home. Its terms also tell users to verify prices and offers. The space between those two statements is where trustworthy grocery technology has to operate.

A good comparison should identify the exact item, reveal what is missing, date the price, distinguish promotion from base price, account for distance and fees, and let the household correct a preference. It should also explain what information is collected to make the result more personal.

That is a more modest promise than perfect grocery truth. It is also more useful. The product does not need to make the same decision for every household. It needs to make the evidence behind each household's decision clearer.

If this conversation is useful, continue with [[How to Compare Grocery Prices Across Stores]]. Readers interested in evidence systems may also like [[Derek Hales on Building NapLab from an Apartment to a Testing Operation]], where a different consumer market raises the same question: what work should sit behind a confident recommendation?

About Andy Ellwood

[[Andy Ellwood Guest Profile|Andy Ellwood]] is the founder and CEO of Stretch. His current site identifies him with the company, while his founder history records Basket as an earlier grocery-transparency venture. Episode 99 focuses on his return to the market, the limits of price comparison, and the discipline of choosing a problem that can hold attention for years.

Sources and editorial notes

This article was written from the preserved episode 99 transcript and checked against Andy Ellwood's current site, founder history, Stretch's website, terms, privacy policy, and App Store listing.

The pricing discussion uses the Consumer Reports investigation, Instacart's response, the New York attorney general's inquiry, and the FTC's separate settlement record. Company history, product behavior, and scale claims are attributed where independent verification was unavailable or unnecessary. AI assisted with research organization and drafting under editorial review. Publication remains unauthorized.