DBA Scraper: Denmark New Listing Alerts

Watch a search, get only what's new. This actor runs your search against DBA.dk's own search-results pages — Denmark's largest classifieds marketplace — and returns one clean row per listing: title, price, condition, brand, and a permanent link. Turn on monitoring and it becomes a watchlist: new listings and price drops since your last run, charged per new row, free on quiet days. Turn on "Alert on gone listings" and it is also a DBA sold items tracker: listings that disappeared since the last run, flagged "presumed sold" when the evidence supports it, plus a lifetime sell-through rate and the median asking price per search. Schedule it hourly and send the results to Discord, Slack, Google Sheets, or n8n.

Who it's for

A reseller or arbitrage buyer watching one category (phones, bikes, electronics) for underpriced or newly-posted listings wants a live feed of what's actually on DBA right now, without refreshing the search page by hand. Danes have been writing their own DBA-monitoring bots and posting them on GitHub for exactly this reason — this actor is that same job as a hosted, schedulable API instead of a script you maintain yourself. A market-research team tracking asking prices for a product category, or a Sheets/n8n workflow that reacts the moment something new is posted, gets the same shape here: paste a list of search terms, get back a flat row per listing, and pay only for listings actually returned — a search that finds nothing costs nothing.

This actor is DBA's member of the same Nordic classifieds family as blocket-listing-lookup (Sweden) in this catalogue — Blocket, DBA, Finn.no, and Tori.fi all run on the same Schibsted/Vend "recommerce" platform (Schibsted acquired DBA from eBay Classifieds Group in 2020), so the same search shape and monitoring model work the same way on each.

Why this one

• Monitoring first, not a one-off dump. Turn on "Only return results that are new, or dropped in price" and this actor becomes a watchlist: it remembers every listing ID it has already shown you, and only bills you for what's actually new or has dropped in price since the last run. A quiet run — nothing new, nothing cheaper — still finishes and costs only the base run fee.
• A sold signal, honestly labelled. With "Alert on gone listings" on, a remembered listing that has left DBA comes back once as a gone row with its last price and days listed, and gonePresumedSold when it was seen on 2+ runs in a row at or under the run's median price. It also keeps a lifetime sellThroughRate per search. Gone rows are ordinary listing rows; there is no extra event. See "Sold signal and sell-through" below for what "gone" can and can't tell you.
• Reads DBA's own structured listing data. Every base field comes from the seoStructuredData JSON-LD block DBA's own pages embed for search engines — not text scraped from rendered HTML — so price, currency, condition, and brand are exact, typed values.
• Never charged for a miss, or a quiet monitoring run. A search that finds nothing, or a scheduled watchlist run where nothing changed, still gets a row explaining what happened — and costs nothing. See "Pricing" below for the found-row rate.
• Honest about what DBA doesn't expose statically. DBA's search and item pages don't serve an original posted-time or seller identity in the static response (confirmed directly, not assumed) — this actor doesn't fabricate those fields. The search page does carry a sort timestamp that resets when a seller re-publishes an ad; the sold signal uses it internally to keep its order honest, and it isn't output as a posted time. See "Data & privacy" and the FAQ for what that means and what you do get from the optional details hop.
• Paste a keyword or your own saved search URL. Type plain keywords, or paste a DBA search-results URL you've already filtered on the site — its filters carry straight through.

What you get

One row per listing by default (turn off "Expand rows" in Advanced to get one row per search instead, with the listings nested under listings).

Field	Type	Description
query	text	The search you submitted, echoed back
found	boolean	true if DBA returned a real search-results page for this search
status	text	OK on a match; NOT_FOUND, BLOCKED, or BAD_FORMAT on a miss
message	text	Plain-English reason for a miss — only present when found is false
searchQuery	text	The keyword actually searched (parsed out of a pasted URL, if you gave one)
listingCount	number	Listings returned and billed for this search
totalAvailable	number	Total listings DBA reports matching this search, including ones not returned
truncated	boolean	true if more results existed than were returned
newCount	number	New listings this run (monitoring only)
priceDropCount	number	Price drops this run (monitoring only)
goneCount	number	Remembered listings detected as gone this run, whether or not "Alert on gone listings" returns them (monitoring only)
detailsCount	number	Listings this run fetched the item-page details for
monitorStatus	text	NO_NEW_ROWS on a quiet monitoring run, WATCHLIST_SEEDED on a silent first run, otherwise empty
sellThroughRate	number	Lifetime share of this watchlist's tracked listings that went gone, 0 to 1 (monitoring only)
medianPrice	number	Median DKK price of every listing this run read for the search, not only the rows returned (monitoring only)
listingId	text	DBA's own numeric listing ID
title	text	Listing title
price	number	Price in DKK, or empty if the listing has no price
currency	text	Always DKK
condition	text	New, Used, Refurbished, or Damaged, as DBA labels it
brand	text	Brand, when DBA's listing has one
imageUrl	text (link)	First listing image
description	text	The listing's blurb; upgraded to the full (contact-info-redacted) description when details are fetched
city	text	Details only. The seller's city
category	text	Details only. Full DBA category path
changeType	text	new, price-drop, seen, or gone (monitoring only)
isNew	boolean	true on this listing's first appearance (monitoring only)
previousPrice	number	The price last recorded for this listing, when it dropped (monitoring only)
priceDropPct	number	Percentage drop from the previous recorded price (monitoring only)
firstSeenAt	date	When this watchlist first saw this listing (monitoring only)
lastSeenAt	date	The last run that saw this listing: this run for a live row, the last sighting for a gone row (monitoring only)
daysListed	number	Days between the first and the last sighting (monitoring only)
lastPrice	number	Gone rows only. The last price seen before the listing went
gonePresumedSold	boolean	Gone rows only. true when it was seen on 2+ consecutive runs and its last price was at or under this run's median
url	text (link)	Permanent listing link
scrapedAt	date	When this row was fetched

Deselect any field in the Input tab's column picker to drop it from every row — query, found, status, message, and scrapedAt always stay.

Pricing

• Listing returned: $2 per 1,000 listings
• Listing details fetched: $1 per 1,000 listings

Plus a $0.002 start fee per run. Each event above is billed independently, only when it actually returns data — misses (found:false) are never charged.

listing-details only fires when "Fetch extra details per listing" is on, and only for listings actually delivered that run — a listing dropped by monitoring never triggers a details fetch, so you never pay for a detail hop on a row you didn't get.

Gone listings (with "Alert on gone listings" on) are ordinary listing rows: each one counts in listingCount and is billed at the same per-listing price. There is no separate event, and a gone row never triggers a details fetch or any other request. With the checkbox off, gone listings are still counted in goneCount and sellThroughRate but not returned or billed.

Monitoring / only new results

Turn on "Only return results that are new, or dropped in price" and a search line becomes a watchlist instead of a one-off dump:

• A listing counts as new the first time its DBA listing ID is returned by this watchlist.
• A listing counts as a price drop when its price falls by at least "Minimum price drop" (5% by default) since it was last seen — measured against the most recently seen price, so a listing that drops twice is compared against its latest price each time, not its original one.
• Already-seen, unchanged listings are dropped before you're billed. A quiet run — nothing new, nothing cheaper — still finishes and returns one free row with monitorStatus: NO_NEW_ROWS, so a scheduled task shows it actually ran. That row still carries goneCount, sellThroughRate and medianPrice.
• With "Alert on gone listings" on, listings the watchlist remembers that have disappeared come back once as changeType: gone rows. See "Sold signal and sell-through" below.
• Listings are de-duplicated by DBA listing ID across pages, so a listing that shifts across a page boundary while the run is paging is returned (and billed) once.
• The first run has nothing to compare against. By default it returns everything it finds as "new" (so you see your whole current search result once). Turn on "Seed silently" to instead have the first run bank everything quietly with zero rows and zero charge, and start alerting from the second run on — this is what stops a fresh watchlist from firing a hundred "new listing" messages into your Discord channel on day one.
• Sort is forced to DBA's own newest-first order whenever monitoring is on, so two runs of the same watchlist are always comparing the same ordering.

Schedule recipe. Save this actor as an Apify Task with deltaMode: true and a deltaName you choose, add an hourly (or daily) Schedule, and point the task's webhook at Discord, Slack, Google Sheets, or an n8n/Make webhook URL to get pinged only when something actually changes.

Price-drop window limitation. A price drop is only ever seen while the listing is still inside what a run actually fetches — if a listing falls out of your maxListingsPerSearch window or your price filters between runs, this actor has no way to notice it dropped. Keep searches reasonably narrow (a specific model, a tight category) rather than one huge catch-all if price-drop coverage matters to you.

Sold signal and sell-through

Honesty note first. DBA never confirms a sale. A gone row means the listing was delisted: most often sold, sometimes withdrawn, deleted by the seller or DBA, or edited so it no longer matches your search words or price filters. That is all "gone" claims. gonePresumedSold narrows it to the listings that were seen on 2 or more consecutive runs and whose last price was at or under the median asking price of this run's search, the pattern a quick sale leaves behind. It is still an inference, not a confirmed sale.

How gone is detected. A run only reads the top of the search, and a listing that slid below it is still for sale, so "missing from this run" is not enough. This actor calls a listing gone in two situations only:

• Complete fetch. The listings read reached DBA's own match count ("finder du N salgsannonce(r)") with no failed or blocked page. Every matching listing came back, so a remembered one that is missing is gone, wherever it used to sit. This is the common case for a narrow watchlist (a model, a price band) with up to a few hundred matches.
• Feed position. For a bigger search, monitoring always sorts newest first, and DBA's page carries each listing's sort timestamp. The actor checks on every run that the listings really are in that order (runs where they aren't make no position check). A listing seen on the last run is judged gone only if the listings that arrived above it since could not have pushed it out of the window, with a safety margin at the bottom edge. Re-published listings (which jump back to the top with a fresh timestamp) count as arrivals, and DBA's "Betalt placering" paid slot, which sits outside the organic list, counts as present and takes no position. Errors land on "still for sale".

There is no gone check on a first run, on an empty or blocked fetch, when a page fails mid-way, or when you changed the search filters under the same hand-named watchlist. Each gone listing is reported once; if it comes back (a sale that fell through, a paused ad), it is taken off the gone count and not re-alerted.

Limits. Listings that vanish near the bottom edge of what a run reads, or after they already slid out of it, are not reported. A busy nationwide search ("iphone", 7,000+ matches) needs a maxListingsPerSearch comfortably above the number of new listings between two runs; a narrow search that fits inside one run is judged completely. The watchlist forgets listings unseen for 30 days and keeps at most 5,000, so sellThroughRate is lifetime within those bounds. Prices are DKK only.

Sell-through rate. sellThroughRate = listings that went gone / listings ever tracked by this watchlist, from 0 to 1, rounded to 3 decimals. It is counted even when "Alert on gone listings" is off, so the rate is right from the first day a watchlist exists. medianPrice is the median DKK price of every listing the run read for that search.

Billing. Gone rows are ordinary listing rows, billed like any other returned listing. There is no separate event, and a gone row never makes a request (it is rebuilt from what the watchlist remembered at the last sighting).

Example input (a daily sold-items tracker for one model; only gone rows come back):

{
  "searches": ["iphone 15 pro"],
  "maxListingsPerSearch": 300,
  "deltaMode": true,
  "deltaName": "iphone15pro-sell-through",
  "alertOnNew": false,
  "alertOnPriceDrop": false,
  "alertOnGone": true
}

The first run only seeds the watchlist (no gone check is possible yet); gone rows start from the second scheduled run. Use the dataset's "Sold signals" view to see them next to the sell-through rate and median price.

How to use

1. In the Apify Console. Open the actor page and click Start — the searches field is already pre-filled with a working example. Results land in the run's dataset as soon as each item is found.
2. Via the API. Call it directly with a POST request — no Console needed once you have an API token:

   curl "https://api.apify.com/v2/acts/accountable_eel~dba-listing-lookup/run-sync-get-dataset-items?token=<YOUR_TOKEN>" \
     -X POST \
     -H "Content-Type: application/json" \
     -d '{"searches":["iphone"]}'
   

3. On a schedule. Save this actor as an Apify Task with the input you want, then add a Schedule (hourly, daily, weekly) so it runs on its own — no server of your own required.

4. Turn on "Only return results that are new, or dropped in price" under Monitoring once you want a watchlist instead of a one-off dump — see "Monitoring / only new results" above.

Input

{
  "searches": [
    "iphone"
  ]
}

One per line — plain keywords (e.g. "iphone", "sofa") or a full search-results URL copied from dba.dk, including any filters it already has. No login required. Accepted formats: iphone, sofa, https://www.dba.dk/recommerce/forsale/search?q=cykel&price_to=2000.

Sample output

query	found	status	searchQuery	listingCount	totalAvailable	truncated	newCount	priceDropCount	goneCount	detailsCount	monitorStatus	sellThroughRate	medianPrice	listings	listingId	title	price	currency	condition	conditionRaw	conditionBucket	listingStatus	brand	imageUrl	description	city	category	changeType	isNew	previousPrice	priceDropPct	firstSeenAt	lastSeenAt	daysListed	lastPrice	gonePresumedSold	url	scrapedAt
iphone	true	OK	iphone	100	7691	true	0	0		0				<all listings found (full list)>	25593920	Apple iPhone 12 Pro 128 GB grå	1300	DKK	Used	https://schema.org/UsedCondition	good	available	Apple	https://images.dbastatic.dk/dynamic/default/item/25593920/7b14b009-671c-4da4-afdc-36b7918ca230	Apple iPhone 12 Pro 128 GB grå												https://www.dba.dk/recommerce/forsale/item/25593920	2026-10-07T06:20:31.038Z

A search that finds nothing comes back as a row with "found": false and is never charged. A quiet monitoring run comes back as a row with "found": true, "monitorStatus": "NO_NEW_ROWS" and is also never charged.

How to filter

Every filter below is optional and off by default — leave it out and you get exactly the same rows as before. minPrice/maxPrice/includeKeywords/excludeKeywords already existed (the latter two match anywhere in the row — title, description, brand, condition, etc. — not just the title). This section adds conditions, availability, cities and brands. A filtered-out listing is dropped before it's billed — you never pay for a row you didn't get.

1. Only "Used"-condition iPhones under 2,000 DKK:

{
  "searches": ["iphone"],
  "maxPrice": 2000,
  "conditions": ["good"]
}

conditions matches DBA's own itemCondition (New → new, Refurbished → like_new, Used → good, Damaged → fair; no grade at all → unknown) — never guessed from the title. DBA has no "new with tags" concept, so new_with_tags never matches anything here.

2. A named brand only:

{
  "searches": ["forstærker"],
  "brands": ["nad"]
}

brands is a case-insensitive "contains" match against the listing's own brand field — no extra request, and this actor keeps paging (up to its normal page budget) until it has enough matching rows, not just enough rows looked at.

3. Exclude a keyword from the title, with per-listing city (needs details on):

{
  "searches": ["cykel"],
  "excludeKeywords": ["defekt", "reservedele"],
  "cities": ["københavn"],
  "includeDetails": true
}

cities matches the listing's city as a case-insensitive "contains" — but city is only known once includeDetails fetches the item page, so cities narrows nothing when that's off (documented in the input's own help text). availability exists too, but DBA's search feed never shows a sold listing (a sold one is simply delisted — see "Sold signal" above), so it's a no-op kept only for input-shape consistency with the rest of the portfolio.

Use it from Clay, n8n, Make, or an AI agent

This actor runs synchronously over plain HTTP — call it directly from a script, a workflow tool, or an AI agent, no Apify Console needed once you have an API token.

curl "https://api.apify.com/v2/acts/accountable_eel~dba-listing-lookup/run-sync-get-dataset-items?token=<YOUR_TOKEN>" \
  -X POST \
  -H "Content-Type: application/json" \
  -d '{"searches":["iphone"]}'

n8n. Add an HTTP Request node: Method POST, URL https://api.apify.com/v2/acts/accountable_eel~dba-listing-lookup/run-sync-get-dataset-items?token=<YOUR_TOKEN>, Body Content Type JSON, JSON Body {"searches":["iphone"]} (swap in an expression from an earlier node for a real value).

Clay. Add an "HTTP API" column: Method POST, URL https://api.apify.com/v2/acts/accountable_eel~dba-listing-lookup/run-sync-get-dataset-items?token=<YOUR_TOKEN>, Body {"searches":["{{search}}"]}, mapping the row's search into the searches array.

MCP. In Claude, Cursor, or any MCP client with the Apify MCP server, ask for "DBA Scraper | Apify" — the agent will find and run this actor.

For monitoring, point the Task's webhook (Console → your Task → Webhooks) at a Discord or Slack incoming-webhook URL, a Google Sheets automation (via Zapier/Make/n8n listening for the webhook), or an n8n/Make workflow's own webhook trigger, so new listings and price drops land wherever you actually work instead of a dataset you have to remember to check.

Tips

• Start monitoring with "Seed silently" on for any watchlist you're setting up on a schedule — it avoids a wall of "new" alerts for your entire current search result on day one.
• Keep a monitored search reasonably narrow (a model, a tight category, a price band) — see "Price- drop window limitation" above for why a huge catch-all search misses price drops on listings that fall outside what gets fetched.
• Turn on "Fetch extra details per listing" only when you actually need city or category — it adds one request (and one listing-details charge) per delivered listing.
• A BLOCKED status means DBA returned a 403, 429, or 503 — lowering "Max concurrency" or trying again shortly usually clears it; this actor doesn't need a proxy by default.
• Paste your own already-filtered DBA search URL instead of rebuilding filters in this actor's Input tab — anything the URL already filters on carries straight through.

vs. hand-rolling your own DBA bot, and vs. blackfalcondata

Multiple Danes have published their own DBA-monitoring scripts on GitHub rather than pay for one — a real, visible demand signal that nothing convenient exists yet. This actor is that same job as a hosted, schedulable API: no server to run, no scraper to maintain against DBA's markup changing under you.

The current Apify Store leader for DBA, blackfalcondata, has 25 users and zero reviews — a one-off dump, no monitoring mode. This actor works across any DBA category, adds new-listing and price-drop monitoring as a first-class mode, and bills only for listings actually delivered.

	What it costs	What you get	Trade-off
This actor	Pay per listing returned; nothing for a miss or a quiet monitoring run	Any DBA category, monitoring mode (new + price drop + gone), a sold signal with sell-through rate and median price, optional city/category details	Posted-time and seller identity aren't available (DBA doesn't expose them statically); "gone" means delisted, not a confirmed sale
A hand-rolled GitHub script	Free to run, costly to build and maintain	Whatever the author built it to do	Breaks silently when DBA's markup changes; no hosting, scheduling, or webhook built in
Doing it yourself	Your time to build and maintain a scraper, a seen-listing store, and a price-tracking layer	The same data	This actor's JSON-LD parsing, monitoring state, and redaction are the maintenance burden it absorbs

Data & privacy

This actor reads public search results that anyone can see without logging in. It doesn't log in, solve CAPTCHAs, or reveal hidden contact details. Seller identity is off by default: DBA doesn't expose a seller id, name, or type anywhere in the pages this actor reads, so none is output. Contact-info-like strings (emails, phone numbers) inside a listing's description are redacted to [redacted]. Turning on seller info in a future version of this actor would make you responsible for having a lawful reason to process it. Not affiliated with DBA / Schibsted.

Is this allowed? It collects the same public listing data your browser shows, for the searches you choose. It's built for monitoring a search, not for copying the marketplace. Check that your use fits DBA's terms and your local law.

FAQ

Why is a row empty, or why does found say false? Either the input line was blank (status: BAD_FORMAT), DBA returned a 403/429/503 (status: BLOCKED), or the search matched nothing on DBA right now (status: NOT_FOUND). Check the message column for the specific reason. None of these are billed.

Am I charged for a quiet monitoring run? No. When monitoring finds nothing new and no price drops, you get one row with monitorStatus: NO_NEW_ROWS and listingCount: 0 — Actor.charge() only fires when listingCount is greater than zero.

Does it find sold listings? What is the sell-through rate? With monitoring and "Alert on gone listings" on, yes, as far as DBA lets anyone tell: a remembered listing that has left DBA comes back once as a gone row, and gonePresumedSold marks the ones seen on 2+ runs in a row at or under the run's median price. DBA never confirms a sale, so "gone" means delisted (most often sold, sometimes withdrawn or edited out of your search). sellThroughRate is the watchlist's lifetime share of tracked listings that went gone. Gone rows are billed as ordinary listing rows. See "Sold signal and sell-through" above.

Does this return posted-time or seller name? No. DBA's search and item pages don't serve an original posted time or a seller statically (confirmed directly against the live site) — this actor doesn't fabricate them. You get listing details (title, price, condition, brand, image, description) plus, with "Fetch extra details per listing" on, city and full category path from the item page.

What does "Fetch extra details per listing" actually fetch? One extra request per delivered listing to that listing's own DBA item page, for its city, full category path, and full (redacted) description. It's billed as a separate listing-details event and is automatically skipped, at no charge, whenever the proxy configuration is set to RESIDENTIAL.

Can I paste my own filtered DBA search instead of using this actor's price filters? Yes. Paste the full search-results URL from your browser's address bar as a search line — any filters already in it (price range, category) carry through unchanged.

Do I need to configure proxies? No. This actor uses Apify's default proxy configuration (no proxy group) out of the box — DBA doesn't block plain requests from Apify's datacenter IPs. RESIDENTIAL is available as a fallback if that ever changes.

Can an AI agent call this directly? Yes. It's registered on the Apify MCP server — an agent in Claude, Cursor, or another MCP client can find and run it by name ("DBA Scraper | Apify"), or you can call the REST endpoint shown above from any script or workflow tool.

Related actors

• Blocket Scraper — the same search and monitoring shape for Blocket.se, Sweden's largest classifieds marketplace (same Schibsted/Vend platform family).