Dell scraper: laptop and desktop price and deal watch (USA)

This Dell scraper reads the laptop and desktop listings on dell.com/en-us and returns one flat row per Dell product line: the starting price in USD, whether the line is on a deal and how big the deal is, rating and review count, model number, display size, the key-feature bullets, colours, the product link and the photo. Run it on a schedule and diff the dataset to get a Dell laptop and desktop price and deal watch: which XPS, Dell Pro, Alienware or Precision lines dropped, which went on a "$300.00 off select configuration" deal, and what is new.

No account, no API key, no login, no browser. One request returns 12 products.

Input

Field	Notes
categories	One or more Dell listings, one per line: laptop or desktop, or the page URL (https://www.dell.com/en-us/shop/laptop-computers/scr/laptop). To read one of Dell's filters, add its appref segment exactly as it appears in the page address: laptop/appref=alienware-product-line or https://www.dell.com/en-us/shop/laptop-computers/scr/laptop/appref=16-gb-memory. One appref filter per line. A filter Dell does not know returns no products and costs nothing. Only the dell.com/en-us laptop and desktop listings are supported: monitors and accessories are a different kind of Dell page, and another country's Dell site is a different store.
maxItems	Products to read per listing (default 100, up to 1,000). 12 per page. The whole laptop listing was 89 products and the desktop one 44 on 2026-11-02.
proxyConfiguration	Apify RESIDENTIAL proxy by default. dell.com refuses a request that does not look like a browser, and some addresses.

Example input:

{
  "categories": ["laptop", "desktop", "laptop/appref=alienware-product-line"],
  "maxItems": 100
}

Output

Prices are numbers in US dollars; currency is USD. One row is one product line, not one configuration: Dell sells a line such as "Dell 15 Laptop" (model DC15250) in dozens of configurations, and the listing shows the line once. product_id is the unique key within a category.

Field	Notes
category, refinement, position	The listing the row was read from (laptop or desktop), the appref filter if you used one (null otherwise), and the place in the listing counting from 1, as Dell's own "13 to 24 of 89" pager numbers it. Dell's default order drifts a little between requests.
product_id, offer_id, model	Dell's id for the line (the last part of its page address, dell15laptopdc15250), the offer code of the line's base configuration (dc15250_so_6; add it to the url for that configuration's page), and the model number Dell prints (DC15250).
name, url, image_url	The name as the tile shows it, the product line's page, and the tile photo.
price	The line's starting price: Dell's "Starting AT" price for the pre-configured base configuration with recommended services. It is already net of any deal on that configuration. A configuration you build on Dell's page costs more or less.
on_deal	True when the tile carries Dell's Hot Deal marker. It always came with a promotion line in the 133 rows checked. It is what Dell says, not a comparison against any price history.
promotion, promotion_amount_off	The promotion line as shown ($100.00 off select configuration) and its dollar amount. Null when there is no promotion, never 0. The saving is on a select configuration, which is not necessarily the one price is for.
deal_max_percent_off, deal_offer_count	The largest percentage off among the line's deal configurations (14.29 is 14.29% off), from the deal data the tile carries, and how many configurations are on a deal. Null and 0 when the line has no deal.
rating, review_count	The average star rating (1 to 5, one decimal) and the number of reviews. Both null for a line with no reviews: the tile then shows no rating (47 of 133 laptops and desktops).
display_inches	The size from the tile's "Display" line (15.6). Null when the tile prints none: all 89 laptops had one, 7 of 44 desktops (all-in-ones).
features	The tile's key-feature bullets as shown, in order (five on 130 of 133 rows, six on 3).
colors, badges	Colour names the tile offers (black, silver; empty when it lists none) and its badges (New; empty when none).
currency	USD.

Example row (real values, 2026-11-02, laptop listing):

{
  "category": "laptop",
  "refinement": null,
  "position": 6,
  "product_id": "dell15laptopdc15250",
  "offer_id": "dc15250_so_6",
  "model": "DC15250",
  "name": "Dell 15 Laptop",
  "url": "https://www.dell.com/en-us/shop/laptop-computers/spd/dell15laptopdc15250",
  "image_url": "https://i.dell.com/is/image/DellContent/content/dam/ss2/product-images/dell-client-products/notebooks/dell/dell-15-intel-3530/media-gallery/laptop-dell-dc15250nt-bk-plastic-gallery-1.psd?fmt=png-alpha&pscan=auto&scl=1&hei=402&wid=627&qlt=100,0&resMode=sharp2&size=627,402",
  "currency": "USD",
  "price": 599.99,
  "on_deal": true,
  "promotion": "$100.00 off select configuration",
  "promotion_amount_off": 100.0,
  "deal_max_percent_off": 14.29,
  "deal_offer_count": 1,
  "rating": 4.5,
  "review_count": 4305,
  "display_inches": 15.6,
  "features": [
    "Intel® Core™ Processors",
    "Windows 11 Home or Windows 11 Pro",
    "Built for everyday computing",
    "Enhanced usability for home learning",
    "Essential productivity features"
  ],
  "colors": ["black", "silver"],
  "badges": []
}

Honest nulls, and what the listing does not carry

A field the tile did not show is null, never 0 or "". features, colors and badges are empty lists when there are none.

• No list price. Dell's listing grid prints one price, the starting price, and no struck-through "was" price, so there is no original-price column and no computed one. promotion_amount_off and deal_max_percent_off are the savings Dell states on the tile.
• No specs table. The listing carries five or six marketing bullets, not processor, memory and storage as separate values, so features is text as Dell writes it. Per-configuration specs, stock and delivery dates are on the product pages, which this actor does not read.
• No sort or price filter. The sort values and price limits tried against the call this actor makes were ignored (the same list came back), so none is offered. Sort the dataset by price or deal_max_percent_off.
• Dell lists a few lines twice. On 2026-11-02 the 89 laptop tiles were 87 distinct product lines. A line is returned and charged once per listing.

Pricing

Pay per event:

• $0.002 per run (actor-start), charged only after your input is validated, so a run that fails on bad input costs nothing.
• $0.003493 per product written (product-scraped), i.e. $3.49 per 1,000 products. A row that is not written is not charged: a repeat of a product the same listing already returned, and a listing that fails, are free.

How it fetches

dell.com puts an Akamai edge in front of its pages. The listing page itself is 2.3 to 2.6 MB of HTML for 12 products and answered HTTP 403 ("Access Denied") to 26 of 32 requests from the machine this actor was built on (eight browser fingerprints, four requests each). Its own page-turning does not load another page: it calls Dell's JSON endpoint (POST /shopapi/plp/v1/getpartialvm/en-us/cnsr/stusa11000386), which answers the 12 product tiles and the pager (resultsCount, lastPage, pageStart) in 0.4 to 0.8 MB. This actor asks for that, with a browser's TLS fingerprint through Apify's RESIDENTIAL proxy, rotating the exit address and the browser fingerprint on every refused attempt. Plain HTTP clients without a browser fingerprint are answered 403.

If Dell's call is refused on a page after its retries (or answers "No Results" for a whole laptop or desktop listing, which only happens when the request is wrong), the actor reads that page from the listing page's own HTML instead (https://www.dell.com/en-us/shop/laptop-computers/scr/laptop?page=2), through the same parser: the same rows, positions and de-duplication, at 2.3 to 2.6 MB a page. The attempts are split between the two routes, so a page nothing answers still costs attempts requests, and the route that answered is tried first for the rest of the run. The two routes returned the same 13 products in the same order for a two-page listing read in the same minute. Dell redirects an unknown appref filter, and a second page of a one-page filter, to the unfiltered listing with HTTP 200; the page route checks which filter the page was rendered for and reads those as no results, never as the unfiltered products.

Things worth knowing, all measured on 2026-11-02:

• Paging and the end. The pager says how many results and pages there are; the actor reads page 1, then pages 2 up to the last, and stops. A page past the end is HTTP 200 with no tiles and the last page's number, which the actor treats as the end. A page that is not the page asked for is treated as a refusal.
• An unknown address is not an error on dell.com. The website serves the laptop list for /shop/monitors/scr/monitors with HTTP 200. The actor checks the category itself and accepts only laptop and desktop (laptops, desktops and Dell's laptop-computers and desktop-computers address names are accepted too).
• Filters. appref= filters agree with the website: alienware-product-line read 6 products and 16-gb-memory read 78 in both. A filter name Dell does not know returns no products, not the unfiltered list.
• Depth. 12 products per page and no page-size setting; at most 1,000 products per listing are read.
• Weight. About 0.7 MB a page, so a 30-product run is three requests and a few seconds.

If Dell changes the call or the tile layout, the run fails with a message instead of returning empty rows.