E-commerce Store Intelligence: Leads, Media Readiness & Accessibility Signals

This actor takes a list of online store domains (Shopify, WooCommerce or another platform) and returns one flat row per shop: the platform it runs on, how many photos its products have and whether they state their dimensions, which accessibility checks its HTML fails, whether its products have structured data, and who to contact.

Introduction

It is built for anyone who sells to online stores: accessibility remediation, SEO, product photography, 3D and AR. It shows which shops have the problem you fix before you spend a call finding out. The result is a list of online store leads, one row per shop, where each column is a reason to contact the shop or to skip it.

Different sellers work from different columns of the same run. Accessibility agencies use a11yChecksFailed, SEO agencies use schemaProductJsonLd, and anyone selling product media uses mediaReadiness. Everyone uses emails.

It analyses the shops you give it. It does not search for new ones.

What it finds

• Platform: Shopify, WooCommerce, PrestaShop, Magento, OpenCart, BigCommerce, Wix, Squarespace and others, and whether the shop already runs a 3D viewer such as model-viewer or Sketchfab (has3dViewer).
• Product photos and dimensions: the product count where the platform publishes one, the average number of photos per product, the share of products with three or more photos, the share that state their dimensions, and a 0 to 100 mediaReadiness score built from these.
• Accessibility: eight signals read from the HTML (page language, image alt text, form labels, a single h1, skip link, pinch zoom, generic link texts and page title) and a11yChecksFailed, the number that fail.
• Product structured data: whether the product page has Product JSON-LD (schemaProductJsonLd), and whether it gives a price, an image and a brand or product code (GTIN or MPN).
• Contacts: emails, with an MX check on the first one, phone numbers in international format and links to social profiles.
• Status: ok, unreachable, blocked or not-a-shop, plus warnings that say what could not be read and why.

Tutorial

1. Bring a list of domains

Anything with a host in it works: ledinis.lt, https://www.shop.lt/kontaktai, shop.lt/. Everything after the host is dropped, because the host is what you will join the results back onto.

2. Run it

The defaults are set for a first pass: 120 products sampled per shop and a hard limit of 25 requests per shop. Fifty shops take about a minute.

3. Sort by the column you sell against

Use mediaReadiness if you sell product photography, 3D or AR, a11yChecksFailed for accessibility work and schemaProductJsonLd for SEO. Filter on has3dViewer to drop the shops that have already bought what you sell.

4. Check status before you trust a blank

A blank cell can mean two different things. Shops with the status unreachable or blocked were never measured. An ok shop with an empty mediaReadiness is one whose catalogue could not be read. The warnings column says which.

Measured on 50 Lithuanian shops

One run on real domains with the default settings. These are measured numbers to plan a job with:

Wall clock	about 57 s for 50 shops at concurrency 8
Reached	48 analysed, 1 unreachable, 1 blocked
Requests	431 total, 8.6 per shop, 25 at the most (the cap)
Time per shop	6.6 s median, 22.7 s at the 95th percentile
Contacts found	an email on all 48, every one with live MX; a phone on 45, every number valid
Media readiness scored	45 of 48
Accessibility	2.4 of 8 signals failing on average; 11 shops failing 4 or more
Product structured data	23 of 48
Already running a 3D viewer	0 of 48

Readiness across the 45 shops that could be scored: ten under 25, nine between 25 and 49, eleven between 50 and 74, fifteen at 75 or above. In the best shop every sampled product had three or more photos and stated its dimensions. The worst published no sizes at all.

Phone numbers are parsed with libphonenumber, using the country named by the domain or by the page's language, and a number has to be written as a phone number. An earlier version matched any eight digits after an 8, and a quarter of what it returned were prices and product codes.

Three of the 48 could not be scored: no store API, no sitemap listing products, and no product page among the links on the home page and its first categories. Shops without a JSON catalogue are read from product pages, found through the sitemap or, if there is none, through the shop's own links. A page counts as a product only if it says so: structured data, Open Graph, or one machine-readable price next to a basket button.

On nine of the shops read this way the photo count could not be read reliably (one image in the structured data, and a gallery named by upload time). Their image columns are left empty instead of showing 1, and their score is based on dimensions.

Dimensions are looked for only in the text a shopper reads. In raw HTML, a page's own stylesheet (max-width: 900px; height: 1em) counted as a stated size, and an earlier version of this table was wrong because of it.

Pricing

$5 per 1,000 shops analysed: $0.005 for each shop that was reached and read. Unreachable and blocked domains, and sites that turn out not to be shops, are still recorded free of charge, so the dead entries in an old list cost nothing. The fifty shops above come to $0.24 at most. Apify adds its usual start fee of $0.00005 a run, and nothing else: compute is included in the price.

Input and output

Input

{
  "domains": ["ledinis.lt", "https://www.hovden.lt", "sofaforma.lt"],
  "maxProductsPerStore": 120,
  "maxRequestsPerStore": 25,
  "concurrency": 8
}

Output

{
  "domain": "sofaforma.lt",
  "finalUrl": "https://sofaforma.lt/",
  "status": "ok",
  "platform": "woocommerce",
  "has3dViewer": false,
  "productCount": 843,
  "productsSampled": 120,
  "estimated": true,
  "avgImagesPerProduct": 8.5,
  "pctWith3PlusImages": 96,
  "pctWithDimensions": 76,
  "mediaReadiness": 91,
  "schemaProductJsonLd": true,
  "a11yChecksFailed": 4,
  "emails": ["shop@sofaforma.lt"],
  "emailHasMx": true,
  "requestsMade": 4,
  "elapsedMs": 9639
}

productCount is exact only where the platform publishes a total, which in practice means WooCommerce. Elsewhere it is null. estimated is true whenever the percentages come from a sample. mediaReadiness gives a weight of one half to the share of products with three or more photos, three tenths to the share that state dimensions, and one fifth to the average photo count.

Actor recommendations

a11yChecksFailed does not measure compliance. The European Accessibility Act has applied to e-commerce since June 2025 and enforcement has begun, which is why this column is worth money. Most of WCAG cannot be checked without a rendered page, a keyboard and a person: contrast needs computed styles, focus order needs the Tab key, and only a person can judge whether alternative text describes its image. The column counts eight facts readable from the source, which is where failures cluster. A shop failing five of them is not compliant, and one failing none still needs a full audit.

An empty mediaReadiness means the catalogue could not be read. A 0 is a score for a catalogue that was read. In the measured run, three of the 48 shops had no score. Sorting blanks as zeros would put the wrong shops at the bottom of your list.

Raise maxRequestsPerStore only for shops with no store API. WooCommerce and Shopify publish JSON catalogues, so those shops are done in four or five requests whatever the limit. The limit only matters for other shops, where the figures come from opening product pages one at a time.

Do not raise concurrency to finish a big list sooner. Requests within one shop are sequential by design, with a pause between them. concurrency sets how many different small shops are visited at the same time, and those shops did not ask to be visited.

Use has3dViewer to filter shops out. A shop already running model-viewer, Sketchfab or a paid viewer has bought what a 3D vendor sells. Filtering those shops out is usually worth more than any ranking of the rest.

FAQ and support

Does it find shops as well as analyse them? No. It takes the list you give it. Store finders and lead databases already do that job well; this actor fills in what their lists leave blank.

Why does a shop show blocked? It answered with 401, 403, 407 or 429, or with a 503 from Cloudflare, usually because a firewall turns away automated visitors. The row records the refusal, so a blocked shop is not reported as unreachable.

Does it respect robots.txt? Yes, including a group that names shop-intel specifically. robots.txt is fetched outside the request budget, which only limits the pages taken from the site.

Why is productCount empty on a Shopify shop? Shopify's open catalogue endpoint lists products but publishes no total, so there is no count to report. The sampled percentages are still valid.

Related tools

• Website Lead Qualifier, for any business website or a Google Maps list: what to pitch, emails, tech stack.
• Lead Qualifier Chrome extension, the same kind of check for the one site open in your browser.
• Photo to 3D, product photos to web-ready 3D models at real size.
• Made by ITneeds.