Beautiful Curly Mehas a storefront problem, not a traffic problem.

“Customer Reviews” and “Find us in Store” occupy 40% of the primary navigation. The store locator actively hands a converted visitor to Target, where the brand earns wholesale margin instead of retail margin and loses the customer record, the email, and the repeat purchase.

Evidence Fig 1.1 · header link extraction, 77 anchors

Speaking bookings, school partnerships, doll sponsorship, wholesale, general contact and the blog all sit in one menu. Each is a different visitor with a different job. Wholesale — a B2B revenue line — is three levels down inside a menu whose label promises a contact form.

Evidence navigation tree above

Bundles → “Gift Sets and Bundles”, Curl Care → “Satin Bonnets”, Confidence Courses → “Confidence Courses”, and Find us in Store → “Store Locator” all resolve to the identical URL as the item above them. The menu looks deeper than it is, and every extra row costs scanning time.

Evidence navigation tree above · 4 duplicate parent/child hrefs

“Books & Puzzles” is the parent category in the menu and holds 10 products. Its child “Books by Zoe” holds 13. A shopper who narrows from the parent to the child sees more products, which is the opposite of what narrowing means.

Evidence /collections/booksandpuzzles 10 · /collections/books 13

Sixteen links stacked in a single left-aligned list — Shop, Book Zoe to Speak, Join our Email List, Find us in Store, Wholesale, Search, FAQs, Contact, Returns, Shipping, About, Careers, Partner, Reviews, Terms, Blog — with no headings, no logo, no mission line, no social icons, no newsletter, and roughly 75% of the footer width left empty. There is no privacy policy link, which matters more than usual for a brand marketing to children.

Evidence Fig 1.2

The country/currency selector renders every option as a real anchor with href="#". That is 237 of the footer’s 254 links, and the bulk of the homepage’s 361 links. A keyboard or screen-reader user reaching the footer must traverse all of them to get to “Terms of Service”, and every crawler spends its budget on them.

Evidence footer anchor extraction — 254 links, 237 currency options, 16 navigation, 1 Shopify

The donation programme appears as a product URL nested under Shop → Dolls, as a full-bleed homepage band, as a counter on the product page, and as three bulk SKUs in the catalogue. It has no section, no landing page in the nav, and no consistent name (“GIFT-A-DOLL”, “Sponsor Dolls for Underserved Girls”, “Giving Back”, “GiftADoll”).

Evidence navigation tree · Fig 3.6 · products.json (6 donation SKUs)

Every collection returns an empty body_html. There is no category copy anywhere on the site: nothing to orient a shopper landing from search, nothing for Google to rank, nothing to differentiate “Beautiful Black Dolls” from “Plush Dolls” from “Doll Bundles”. On /collections/all the description slot instead holds an operations note — “All discount codes will be applied at checkout” — which is also what feeds the page’s meta description.

Evidence collections.json — 55/55 with 0 characters of description · Fig 4.1

They are live URLs with no inbound link from the header, footer, or any menu. Several hold real merchandising value — educator-products (15 SKUs), gifts-under-25 (14), best-sellers (21) — and are simply unreachable. The rest are expired campaigns that dilute the crawl and can be reached from Google.

Evidence 12 collection URLs across header + footer vs 55 in collections.json

zoes-course, freebies, not-on-sale, new-birth-toy-drive and doll-bundles-1 all return zero products. Anyone reaching them sees an empty grid.

Evidence per-collection product counts, all 55 collections

“Spring Savings” (sale) holds 29 of 46 products. “Give the Gift of Confidence” and “Give the Gift of Self-Love” hold 28 each. The menu’s “Sale” points at a seventh set, bcmsale, with 24. Roughly two thirds of the catalogue is permanently discounted across six competing sale collections, which trains customers to wait and makes the Labor Day banner meaningless.

Evidence product counts — sale 29 · selflove 28 · give-the-gift-of-confidence 28 · bcmsale 24 · evanabirthdaysale 25 · holidaydeals 23

Five products are typed Doll and four Dolls. Digital goods are split three ways across Digital, Digital Download and Course. Two products have no type at all, and one is typed warranty — the shipping-protection app’s SKU, sitting in the merchandisable catalogue. Product type is what powers automated collections and filters, so every one of these splits breaks browse.

Evidence products.json — Bundle 9, Book 6, Doll 5, Dolls 4, Accessories 4, Bonnet 3, Puzzles 3, Digital 2, Classroom Bundle 2, (empty) 2, Course 1, Digital Download 1, Donation 1, Gift Card 1, Tees 1, warranty 1

bfcm2024-sale is still attached to roughly 20 products. Alongside genuine merchandising tags sit app and ops flags — NOREVIEW, insurancetype-fixed, simplyinsurance, LINKSCOURSE, LINKSSTEMREADY, _For The Classroom, zuridolls — with inconsistent casing throughout. Tag-driven collections and filters inherit all of it.

Evidence products.json tag extraction, 46 products

inspire-her-book-puzzle-bundle, gift-a-book, curly-confident-gift-card, denim-jumper-set-for-18-dolls, lilac-denim-tutu-dress-bow and four others ship with one photo. A bundle with one photo cannot show what is in the box, which is the only question a bundle buyer has.

Evidence products.json image counts

Six donation SKUs sit in the shoppable grid typed Dolls and Donation. “Gift A Doll | Support Girls in Underserved Communities” appears between a puzzle and a colouring book with a struck-through ₱5,100 price and a pink “Save 49%” flag. A charitable gift presented as a discounted product undercuts both the gift and the product.

Evidence Fig 4.1, row 2 · products.json (giftadoll, gift-a-book, gift-50-dolls, gift100dolls, gift-200-dolls…, gift-dolls-to-girls-links-campaign)

Shopify’s complementary-product endpoint returns an empty set for Bella, the bestseller. Every cross-sell on the site is algorithmic, which is why the “related” list for an 18-inch doll leads with two bundles that contain the same doll instead of the outfits, bonnets and books that would raise the basket.

Evidence /recommendations/products.json?intent=complementary → []; intent=related → bella-doll-confidence-bundle, bella-doll-and-book-bundle …

Terse legacy handles (belladoll, madamvp, bcmbook, ailab, gift100dolls) sit beside 60-character SEO slugs; six carry a redundant beautiful-curly-me- prefix. Three collections have handles that contradict their titles — hair-care-2023 is titled “Books & Puzzles – 2023”, sale is “Spring Savings”, bryant-park-winter-village is “BCM Pop Up”. Homepage bundle links also use collection-scoped product URLs (/collections/giftsets/products/…), creating a second address for the same product.

Evidence products.json + collections.json handles · homepage anchor extraction

The slideshow section contains zero characters of text. Headline, offer, terms and the “Shop Now” pill are all rasterised into a single JPG, wrapped in an anchor with no accessible name. Nothing is selectable, translatable, indexable or readable by assistive technology, the text cannot reflow on any viewport, and updating the offer requires a design tool. Desktop and mobile use two different creatives, in two different brand colours (pink and blue), with two different wordings of the same promotion.

Evidence Fig 3.1 (desktop) · Fig 5.1 (mobile) · section innerText = "" · 1 anchor, no accessible name

All ten .logo-bar__item wrappers compute to opacity: 0, and none of the inner image-element nodes ever receive the aos-animate class that would reveal them. We confirmed this after direct scroll-to, after incremental scroll past, and after scrolling down and back. The images themselves are at full opacity; the parent is what hides them.

Evidence Fig 3.2 · computed opacity 0/10 items in three scroll states · aos-animate 0/10

The only h1 on the page is the logo, “BEAUTIFUL CURLY ME”. The document outline then jumps to “GIVE THE GIFT OF CONFIDENCE!”, with “CURRENCY” — the footer selector — appearing as a peer heading. 29 of 41 images carry no alt attribute, and 29 of 361 links have no accessible name at all.

Evidence heading + image + anchor extraction, homepage

Nine tiles across two sections point at seven collections under five different names, sharing at least one identical photograph. Seven of the nine white labels fall below the 3:1 minimum for large text; eight of nine fall below the 4.5:1 required for text at their actual size.

Evidence Figs 3.3–3.4 · contrast table above

A 765px YouTube embed and a 1,586px Loox review carousel account for 26% of the page. Neither contributes a single crawlable word, neither is styled to the brand, and the Loox carousel is the only place on the homepage where the brand’s 191 reviews appear.

Evidence iframe measurement — YouTube zvtsK8_Q0SY 1360×765 · Loox carousel 1360×1586

The announcement bar is a two-slide carousel: “BOGO50 SALE FOR LABOR DAY WKND! NO CODE NEEDED” and “FREE DOMESTIC SHIPPING for orders $60+”. A visitor may never see the shipping threshold, which is the single most reliable average-order-value lever a store this size has.

Evidence announcement section slide extraction

Every card on the homepage and in every grid has a quantity stepper and an Add to Cart button, and none has a rating, a review count, an age range, or a doll size. The card is optimised for a shopper who has already decided, on a site whose whole job is helping them decide.

Evidence Fig 3.5 · Fig 4.1 · 0 rating elements across 40 collection cards

“No code needed.Savings at checkout” (missing space, baked into the hero image). The newsletter block renders its button label twice — “SUBSCRIBE SUBSCRIBE”. The founder’s age is given as 7 here (“Founded by a 7-year-old girl”) and differently elsewhere on the site. The Gift-A-Doll band opens on a negative — “DON’T WANT/NEED A DOLL?”.

Evidence Fig 3.1 · Fig 3.6 · section text extraction

A 40-product catalogue is split into five paginated screens. Seeing everything the brand sells requires four extra page loads, and nothing carries the shopper from one page to the next — no infinite scroll, no “load more”, no per-page control.

Evidence Fig 4.1 · “40 products” · pagination 1 2 3 … 5

Across a catalogue that spans 18-inch dolls, plush dolls, picture books, journals, jigsaw puzzles, satin bonnets, doll clothing, tees, digital downloads, courses and classroom bundles, the shopper can filter by in-stock or out-of-stock. There is no filter for age, price, product type, doll size, or format — and no way to answer “what do you have for a six-year-old under $30?”, which is the single most common question a gift-buyer has.

Evidence Fig 4.1 · filter group extraction, /collections/all and /search

Bella carries a 4.9 rating from 191 reviews. Zero rating elements render on the collection grid, on the homepage rows, or on search results. The brand’s strongest conversion asset exists only on the product page and inside an iframe carousel.

Evidence Fig 4.1 · 0 rating nodes across 40 cards · Loox rating widget present on PDP only

The page a shopper lands on from “Shop” is titled “All”, and the copy beneath it reads “All discount codes will be applied at checkout. Bundles are already discounted as reflected.” That text also feeds the page’s meta description and its structured data, so it is what appears in search results.

Evidence Fig 4.1 · <h1>All</h1> · collection body_html

Searching “doll” returns 31 of 46 products, because the word appears in nearly every description. The results page is headed “SEARCH”, offers the same single Availability filter, and provides no type or category grouping to recover from the flood.

Evidence /search?q=doll — “31 results found” of 46 products

One row of the grid contains a cut-out doll floating on white, a 3D book mockup with a drop shadow, a flat cover scan, a lifestyle photograph with the words “GIFT-A-DOLL” set into the image, a packshot with an award seal composited on, and a boxed product with a retailer logo burned in. Crops, scales, backgrounds, lighting and baked-in graphics all differ, so the grid reads as a marketplace rather than a brand.

Evidence Fig 4.1 · Fig 3.5 · Fig 5.2

On a 375×812 viewport the button is at 956px — 144px past the first screen — and there is no sticky add-to-cart bar to compensate. The entire first screen is a single product photograph with no title, price, rating or price context.

Evidence Fig 5.3 · measured offsets, clean page load

12,122px on mobile, 10,172px on desktop. A 2,598px Loox review iframe is 21% of it; four separate brand-story sections between the buy zone and the reviews add another 1,790px. Nothing is collapsed, and the sections are not ordered by what a buyer needs next.

Evidence section stack (§03) · 18 sections measured

Benefits, mission copy, care instructions and materials run together in one block with literal asterisks for bullets. There is no specification panel — no doll height, age range, materials, what’s in the box, or wash instructions in a scannable form — even though every one of those facts is buried in the paragraph.

Evidence Figs 5.1–5.2

The product has no category path because it was not reached through a collection. There is no way back up to “18-inch dolls” or “Dolls”, no sideways move to a sibling, and no signal to search engines about where this product sits.

Evidence Fig 5.1 · Fig 5.3

The rating widget renders five small stars and “(191)”. The score itself — 4.9, which is in the page’s structured data — is never displayed, is not repeated near Add to Cart, and does not appear anywhere else on the site.

Evidence Fig 5.3 · Loox widget data-rating="4.9" data-raters="191"

Directly beneath Add to Cart — the most valuable strip on the page — sit the word “GiftADoll” with no styling or link, and a two-column counter reading 3,599 “Dolls Donated” against a 5,000 “Quarter Goal”. The donation story deserves that prominence; presented without a sentence of context or a call to action, it reads as an unexplained dashboard.

Evidence Fig 5.1

The announcement bar offers free domestic shipping over $60. The product page, on a ₱4,200 / $75-class item, tells the shopper shipping is an unknown to be revealed at checkout. Nothing on the page connects the two.

Evidence Fig 5.1 · announcement slide 2

Mobile paginates twelve images with dots only, so there is no way to see what is in the set or jump to the packaging shot. Several images carry composited callouts (“18" Tall”) positioned so they are cut off at the viewport edge.

Evidence Fig 5.3 · 12 gallery nodes · products.json (9 images) vs 12 rendered

The Gorgias “Chat with us” pill floats over the quantity stepper in the buy zone, and over the first row of products on the mobile homepage. It is one of four Gorgias iframes mounted on every page.

Evidence live mobile capture, 375×812 · 4 Gorgias iframes in DOM

Bella’s related products lead with “Bella Doll Confidence Bundle” and “Bella Doll and Book Bundle” — two bundles containing the doll being viewed. With no complementary products configured (PA-09), the outfits, bonnets and matching books that would lift the basket never surface.

Evidence /recommendations/products.json?intent=related

Desktop: pink ground, “LABOR DAY WEEKEND SALE! / BUY ONE GET ONE 50% OFF Books & Puzzles / No code needed.Savings at checkout”. Mobile: blue ground, script lettering, “Labor Day Wknd Sale / BUY ONE GET ONE 50% off / ALL BOOKS AND PUZZLES / no code needed”. Two files, two colour systems, two wordings of one offer — and both must be re-exported to change anything.

Evidence Fig 3.1 vs Fig 6.1

Because cards are stretched to the tallest in the row, a two-line title leaves a large gap above its Add to Cart button while a four-line title has none. Scanning the column means reading past inconsistent gaps, and fewer products fit per screen.

Evidence Fig 6.2

When the longer promotional slide is showing, the message wraps to two lines and the “Shop Now” link is cut off mid-character by the bar’s fixed height.

Evidence live mobile capture, 375×812

The B2B channel is a Faire widget with two identical “Start buying wholesale” CTAs. It routes retail buyers off-site, gives away commission and the customer relationship, and presents a completely different visual identity on a beautifulcurlyme.com URL. Shopify B2B would keep the buyer, the catalogue and the brand in one place.

Evidence Fig 7.1

/collections/educator-products holds 15 products, including two purpose-built classroom bundles with teacher resources. It appears in no menu. Schools and districts are a natural, high-value, repeat-purchase audience for this brand, and the products for them are already built.

Evidence collections.json · navigation extraction

The menu’s “Confidence Courses and Resources” resolves to /collections/digital-downloads (4 products) while the collection actually named zoes-course is empty. Course delivery runs on Thinkific, outside the store, so course buyers never enter the customer record that drives repeat doll sales.

Evidence per-collection counts · products.json (confidence-course, tag Thinkific)