Who the Machine Names When a Shopper Wants a Toy Store: Local Visibility in a Gift-Driven, Seasonal Category

Introduction

A toy is rarely bought for the buyer. It is bought for a birthday, a holiday, a new baby, a niece the shopper sees twice a year. That one fact, that the purchase is a gift for someone else, shapes every step of how a toy is searched for, chosen, and paid for, and it decides where a local toy shop is found or lost. The category is also sharply seasonal. Much of the year settles itself in the ten weeks before December 25, and the visibility a shop earns in October is the visibility it banks in December.

The common story about this category is a story of loss. Toys R Us collapsed in 2018, Amazon and the big-box chains absorbed the volume, and the neighborhood toy shop was supposed to go with it. The establishment count says otherwise. The United States held 61,067 hobby and toy stores in 2026, up 4.9 percent on the prior year (IBISWorld), and US retail toy sales grew 6 percent to 30.3 billion dollars in 2025 (Circana). Independents did not vanish. They regrouped around the one thing a warehouse cannot ship: curation, expertise, and a place a family walks into. The question this study asks is narrow and testable. When a shopper turns to a machine and asks for a toy store, who does the machine name, and where does the independent shop actually stand?

The terrain is mode-specific because the buying mode is mode-specific. A gift is a decision made under uncertainty about another person, on a deadline, with social stakes attached, and that pressure pushes shoppers toward trusted recommendation and toward a seller who reads as an expert. The same pressure splits the demand cleanly. A shopper who already means to buy online types a product name and lands on a marketplace. A shopper who wants a place, an experience, or advice types "toy store near me" and enters a local answer marketplaces do not occupy. This study measures that second surface, the local one, because that is the surface a physical shop can win.

Data and method. The evidence is primary and dated. We captured real Google output for four shopper prompts across five metros, classified every named result, measured demand and its month-by-month shape from search-volume data, and audited a small sample of real independent shops for machine-readiness. A follow-up grid then put the same question to three synthetic answer engines. Every figure below carries the date it was measured and an evidence tier. Where an engine could not be captured, it is named as uncaptured, not estimated.

Background and literature

Economic theory classes goods by how a buyer can judge quality. Nelson (1970) separated search goods, whose quality is verifiable before purchase, from experience goods, whose quality is known only after use. Darby and Karni (1973) added credence goods, whose quality the buyer struggles to judge even after use. A toy chosen as a gift sits across all three at once. The buyer cannot fully experience it, because it is for someone else, and cannot always judge whether a five-year-old will love it or whether a developmental claim is real. That is why third-party signals, reviews, staff recommendation, and a shop that reads as an expert carry unusual weight in this category.

The gift itself has a deep literature. Belk (1979) framed gift giving as symbolic communication in which the giver invests part of the self in the chosen object, so the purchase is judged as much as a statement about the relationship as it is a product. Sherry (1983) modeled the exchange as a staged social process, with real cognitive work in the search phase and both parties reading signals as the gift changes hands. Wooten (2000) showed that this work is felt as anxiety, a self-presentational unease that rises when the giver is unsure of the recipient or the social stakes are high, precisely the near-relative, holiday, and new-baby occasions that drive toy demand. Each strand points the same way: the gift shopper is buying reassurance as much as an object, which rewards any seller who spares them the burden of deciding.

The size of the informational problem is itself quantified. Waldfogel (1993), in the classic study of the deadweight loss of Christmas, estimated that holiday gifts destroy roughly a tenth to a third of their retail value because givers cannot match a recipient preferences as well as the recipient could, the loss widening with social distance. The finding reframes expert curation and recommendation not as a soft nicety but as a mechanism that closes a measurable value gap, which is the specific service a knowledgeable specialty shop sells.

Curation also answers a failure of abundance, though the evidence demands care. Iyengar and Lepper (2000) showed that larger assortments can lower the likelihood a shopper buys at all, the choice-overload effect. The later meta-analysis by Scheibehenne, Greifeneder, and Todd (2010), pooling 50 experiments, found the mean effect near zero but highly variable, concentrated in exactly the conditions the gift shopper occupies: low prior expertise and high uncertainty about what to want. A marketplace of millions of listings maximizes selection and, for the uncertain gift-giver, can maximize paralysis, while a specialty shop of a few hundred deliberately chosen items sells the opposite proposition, a decision made easier. This is the theoretical spine of the independent toy shop after Toys R Us, and trade reporting describes the same repositioning around specialty, developmental, and nostalgic assortments (The Toy Book; NWI Times).

How shoppers move between channels is the research-shopper phenomenon. Verhoef, Neslin, and Vroomen (2007) documented that buyers routinely research in one channel and purchase in another, the pattern now called webrooming when the search is online and the visit is in a store. For a local toy shop the machine answer is that research step, and it decides whether the shop makes the shopper visit list before a foot ever crosses the threshold.

The value of that answer is measurable. Luca (2016) found that a one-star improvement in a Yelp rating moved revenue by 5 to 9 percent for independent restaurants, direct evidence that platform-mediated reputation has commercial force for small, non-chain businesses. Toy shops live in the same review-signaling economy, and the credence-good frame predicts the effect should run at least as strong where quality is hardest to verify.

Finally, the surface itself is moving. Aggarwal and colleagues (2024), in the Generative Engine Optimization study presented at KDD 2024, showed that inclusion and prominence inside generated answers respond to identifiable content and source signals, not to classic ranking alone. As shoppers begin to ask an assistant rather than scroll a list of links, the question shifts from where a shop ranks to whether a machine names it. This study measures both the classic surface and the earliest state of the generative one for this category.

Findings

The measured picture contradicts the loss narrative on its central point and confirms it on a subtler one. On the surface a local shop can actually win, the Google local pack, independents dominate. Across the five metros the map 3-pack was 87 percent independent shops (47 of 54 slots) and held zero online marketplaces. Amazon, Target, and Walmart appeared in no local pack at all. The big-box threat to a toy shop is real, but it does not live inside the local answer. It lives in the separate demand of shoppers who never search locally in the first place.

The chains that did surface in the pack were a narrow set. POP MART, a fast-expanding collectibles chain, recurred across the near-me packs in New York, Chicago, and Atlanta, and flagship or franchise names (FAO Schwarz, a Learning Express franchise) took the remaining chain slots. The intrusion tracked metro size: New York packs were 67 percent independent, while Dallas and Phoenix packs were 100 percent independent. Chain pressure on the local pack rises with density, and the mid-size metro stays almost entirely an independent surface.

The second contest is organic composition, and here the independent has a real problem. Of 319 organic results across the 20 pages, only 48 percent were independent shop websites, while 46 percent were aggregators, social platforms, and directories, led by Yelp, YouTube, Facebook, Instagram, Reddit, TimeOut, TripAdvisor, and Yellow Pages. Chains and marketplaces took 3 percent each. A shopper who trusts the map pack meets independents. A shopper who scrolls into the blue links meets intermediaries, who then decide in their own ranked lists which shops to surface. The directory is not the competitor. The directory is the gatekeeper.

The third contest is the generative answer, and for this category it has barely begun. Google AI Overviews were requested on all 20 pages and fired on only 1 (5 percent). That single Overview, on the query "independent toy store new york", is instructive out of all proportion to its frequency. It named Mary Arnold Toys, Kidding Around NYC, and Playing Mantis, cited their own websites alongside a map source, and opened by praising "highly curated, specialty items you won't find at big-box chains". When the machine chose to write a paragraph about this category, it named independents and it rewarded curation. The generative surface is not yet the battleground here, but its earliest behavior favors exactly the shops the loss narrative wrote off.

The fourth finding is the shopper's own language, and it is blunt. Demand is enormous and plain. "Toy store near me" and "toy stores near me" each drew 246,000 US searches per month, "toy store" drew 165,000, and "toys near me" drew 49,500. The trade's own vocabulary is a rounding error beside it: "independent toy store" drew 210, "specialty toy store" drew 210, "specialty toy retailer" drew 10, and "curated toy shop" returned no measurable volume. The gap between how a shop describes itself and how a shopper searches is roughly 1,171 to 1. A shop tuned for the word "independent" is tuned for a door almost nobody knocks on.

The fifth finding is the shape of the year. Demand for "toy store near me" peaked at 450,000 searches in December 2025 and fell to a 165,000 trough in January 2026, a 2.7-times swing, with the December figure running 1.8 times the twelve-month average. "Toys near me" traced the same arc (110,000 in December against a 49,500 average). Visibility in this category is won or lost on a calendar. The map-pack position a shop holds entering October is the asset that pays out in December, and a shop that starts fixing its local surface in November has already missed the season it set out to fix for.

The AI answer engines: who ChatGPT, Perplexity, and Gemini name

The Google round left the chat engines uncaptured. On 2026-07-23 we closed that gap for three of the five metros, capturing three synthetic answer engines (ChatGPT, Perplexity, and Gemini, all with web search enabled) across two shopper prompts, a reputational one ("best independent toy stores in [metro]") and an intent one ("where should I buy a gift toy in [metro]"), in New York, Dallas, and Chicago, at k=2 runs each. All 36 captures returned. The grid is small and single-day, so the reading is directional, and because these answers are synthetic, non-deterministic, and drift over time, the same prompt tomorrow may return a different list.

The headline is that all three engines name independent local shops, and they largely name the same ones. On the reputational prompt every engine, in all three metros, led with independents and often opened by explicitly framing them against the chains: Kidding Around, Mary Arnold Toys, Playing Mantis, and Toy Tokyo in New York; The Toy Maven, Toys Unique!, Froggie's 5 & 10, and Dallas Vintage Toys in Dallas; Timeless Toys, Building Blocks, Rotofugi, and *play in Chicago. A shared independent core recurred across all three engines in each metro, so the local independents that win the Google map pack are also the ones the chatbots name. Online marketplaces (Amazon, Target, Walmart) were essentially absent from the chat answers, as they were from the local pack.

The engines diverge sharply on where they source that list, which is the part that matters to a shop. ChatGPT cited the shops' own websites almost exclusively (boomerangtoys.com, toysunique.net, timelesstoys.com, rotofugi.com), with a shop's Facebook page as the only intermediary, and its two runs were near-identical: for a shop, this is the best case, because the citation is its own front door. Perplexity named independents too but derived them from directories and local media (Curbed, Vogue, Time Out, CBS News, the Dallas Observer, Yelp, Reddit, and city parenting guides), so the same intermediary gatekeepers who control the organic surface also assemble Perplexity's shortlist. Gemini produced the longest and most detailed lists, but every one of its citations was masked behind a vertexaisearch.cloud.google.com redirect, so its sources cannot be classified at all; we read Gemini only by the businesses named in its text, and that opacity is itself a finding.

Two qualifications sharpen the reading. First, the buying-mode split the study already theorizes appears inside the chat answer: on the intent prompt ("where to buy a gift toy") all three engines tilted toward national flagships and brand stores (FAO Schwarz, the LEGO Store, American Girl, POP MART, a re-opened Toys R Us) alongside the independents, exactly as the reputational-versus-transactional distinction predicts. Second, cross-run stability was good but not perfect: ChatGPT was near-identical run to run, while Perplexity and Gemini held their top names but reshuffled the tail of their longer lists. The net reading strengthens the study's thesis on the local surface and carries its warning into the new one. The chatbots do name independents, but Perplexity routes through the same directories that gatekeep organic, and Gemini will not show its work, so the shop that becomes a clean, well-reviewed, machine-legible local entity is the shop each engine can most reliably find and cite.

Discussion

Read through the buying-mode theory, the data resolves into a coherent account of the category. The gift frame (Belk, 1979; Sherry, 1983), the anxiety it carries (Wooten, 2000), and the credence problem (Darby and Karni, 1973) make trusted recommendation and expert curation the decisive product. Waldfogel (1993) put a number on why: a gift destroys value when the giver cannot match a recipient preferences, and curation is the mechanism that narrows the gap. Choice overload sharpens the same point, with a caveat the evidence demands, since the pooled effect is real mainly for low-expertise, high-uncertainty buyers (Scheibehenne, Greifeneder, and Todd, 2010), which is exactly the gift shopper facing a marketplace of millions. The market reorganized around that advantage after Toys R Us, and the machine, when it writes about the category at all, echoes it word for word: curated, specialty, not the big-box chains. The independent toy shop is not fighting the algorithm. On the local surface, the algorithm is largely on its side.

The threat is therefore misdiagnosed by the loss narrative. Marketplaces are absent from the local pack because they compete for a different shopper, the one who never searches locally and goes straight to a product page. That leakage is real, and it is where feed and shopping work earns its keep, but it is not a local-visibility loss because the shop was never in that answer. The losses that are local-visibility losses are three, and they are specific: a shop with an incomplete profile that does not make the pack at all, the two-in-twenty case where no pack returned; a shop out-signaled inside the pack by a chain like POP MART that manages its local presence aggressively; and a shop buried in an organic surface that is half Yelp, Reddit, and Instagram, where the intermediary, not the shop, controls the ranked list the shopper reads.

Reading the classic and synthetic surfaces together sharpens the picture rather than blurring it. The same independent core that wins the map pack is the core all three chat engines name, so entity strength compounds across surfaces instead of splitting between them. The divergence is in the sourcing, and it is a warning. ChatGPT cited the shops' own sites, which is the shop's front door doing the work; Perplexity reached the same independents through Curbed, Time Out, and Yelp, the very intermediaries that gatekeep organic; and Gemini masked its citations entirely. A shop that is a clean, well-reviewed, machine-legible local entity is the one every surface can find, cite, and agree on, while a shop that exists only through directories inherits the directory's advantage on the surfaces still under construction.

This maps directly onto the RavenEye model. The Visibility Corpus is the terrain map this study is a single tile of: for a given category and geography, where does the attention actually sit, which surfaces decide, and who currently holds them. Search Surface Optimization is the method that works the surfaces the map identifies as decisive, which for toys means the Google Business Profile and the local entity first, review depth second, and the machine-readability of the shop's own site third, rather than a scattershot of tactics aimed at surfaces that do not govern the outcome. The Machine-Readiness Score is the single measured read of a shop's standing across those surfaces, so that seasonal work can be timed and its movement verified rather than asserted.

The generative surface deserves a measured word. It is tempting to over-read a single Overview, so we will not. The reading is that AI Overviews are not yet a material surface for local toy queries (they fired 5 percent of the time), that the standalone chat engines, captured separately on 2026-07-23 and reported below, do behave differently on this question, and that the one Overview we did capture is a directional signal, not a trend, that the generative layer inherits and amplifies the same entity and curation signals that already decide the local pack. A shop that becomes a clean, well-reviewed, machine-legible local entity is preparing for both the surface that governs today and the one that is arriving.

Implications for retailers

For an independent toy shop, the data argues for calm and for sequence rather than for panic. The local answer is winnable and largely yours already. The work is to be eligible for it, to hold it against the one or two chains that contest it, and to time the effort to the calendar the category runs on.

The first move is the cheapest and the highest-impact: make the Google Business Profile complete and correct, because in this study the profile, not the website, is what earns the map-pack slot. The second is to make the shop's own site legible to a machine as a local business, since none of the audited independents did this and it is the clearest untaken advantage in the category. The third is to treat reviews as the credence-good signal the theory says they are, earned from real customers only, because in a gift category the shopper is buying reassurance as much as a toy. And the fourth is to do this work before the season, not during it, because the December peak pays out the position a shop built in October.

Google rankings and AI answers are decided by the engines, not by any firm. What can be done is to engineer every signal that legitimately moves, measure the starting position, and verify the change. That is the work.