Kijiji Scraper: New Listing & Price Drop Alerts

You give this actor a search (a keyword, or a pasted kijiji.ca search URL) and it returns one row per listing: title, price, location, condition, seller hash, image and link, read straight from Kijiji's own server-rendered search page — no login, no headless browser, no anti-bot proxy needed for this target. Turn on monitoring and the same search becomes a Kijiji alert for new listings and price drops: only what's genuinely new or cheaper since your last run, charged only for what's delivered and free on a quiet day, with everything exportable to Google Sheets, n8n, Make or Clay.

Who it's for

Someone hunting a used bike, a couch or a specific vehicle on Kijiji who doesn't want to refresh the search page by hand ten times a day. A reseller watching a category for underpriced finds to flip. A developer wiring Kijiji results into a Discord bot, an n8n workflow or a Google Sheet without maintaining their own scraper against Canada's largest classifieds site, which has no official public API. Anyone who has looked at a paid Kijiji-alert app and would rather pay per listing than a monthly subscription for one search.

Why this one

• Reads Kijiji's own server-rendered data, not a headless-browser render. The listings are already in the page's __NEXT_DATA__ payload on the very first response — no JavaScript execution, no CAPTCHA, no proxy needed for this target (confirmed live).
• Monitoring is built in, not bolted on. Turn on "Only return results that are new or dropped in price" and a search becomes a watchlist: new listings and genuine price drops since your last run, tracked in a named store so it survives between scheduled runs.
• Promoted (Top Ad) listings are labelled, not hidden or double-counted. Kijiji sells top placement separately from the ranked organic feed; this actor tags isPromoted and keeps paid placements out of the new/price-drop bookkeeping so a re-bumped ad never fakes a "new" alert.
• Never charged for a miss, and never charged for an already-seen listing. A search that turns up nothing costs only the run fee. On a monitored search, a listing you've already been shown once is dropped before billing, not after.
• Seller identity stays anonymous. You get an anonymous seller hash, rating and verified flag; Kijiji's own search results never expose a seller's name at all (see the FAQ), so this actor never invents one.

What you get

Every row starts with the same five fields, then the parent search summary, then one row per listing (turn off "One row per listing" in the Input tab to get one grouped row per search instead of one row per listing; you're charged the same either way).

Field	Type	Description
query	text	The search you submitted, echoed back
found	boolean	true if the search returned at least one listing
status	text	OK on a match; NOT_FOUND, BLOCKED, BAD_FORMAT on a miss
message	text	Plain-English reason for a miss, only present when found is false
scrapedAt	ISO 8601 datetime	When the search was run
searchQuery	text	The keyword actually searched
listingCount	number	Listings returned for this search (what you're billed on)
promotedCount	number	How many of those are promoted (Top Ad) listings
totalAvailable	number	How many listings Kijiji reports in total for this search
truncated	boolean	true if more organic listings existed than were fetched
newCount	number	New listings this run (monitoring only)
priceDropCount	number	Genuine price drops this run (monitoring only)
listingId	text	Kijiji's own listing ID
url	link	The listing's page on kijiji.ca
title	text	Listing title
description	text	Listing description, with phone numbers and email addresses replaced by [redacted]
price	number	Price in CAD, when the listing has a fixed amount
priceEur	number	Price converted to EUR at the ECB daily reference rate, when a rate is available
priceType	text	FIXED, CONTACT (observed live), or Kijiji's other documented filter values (GIVE_AWAY, SWAP_TRADE)
currency	text	Always CAD
location	text	Seller's listed location name
latitude / longitude	number	Listing coordinates, when Kijiji provides them
postedAt	date	Kijiji's sortingDate — the listing's most recent bump/republish
firstListedAt	date	Kijiji's activationDate — when the listing was originally posted
imageUrl	link	First listing photo
imageCount	number	How many photos the listing has
condition	text	Condition, when Kijiji's category-specific attributes include one
brand	text	Make/brand, when the category has a matching attribute (e.g. vehicles) — often null on general classifieds
size	text	Size, when the category has a matching attribute (e.g. bike frame size) — often null
category	text	Kijiji's numeric category ID
sellerId	text	Anonymous hash of the seller's poster ID, always present
sellerName	text	Always null — Kijiji's search results never expose a seller display name (see FAQ)
sellerRating	number	Seller rating, when Kijiji reports one
sellerVerified	boolean	Kijiji's own seller-verified flag
isPromoted	boolean	true for a paid Top Ad placement
isNew	boolean	true on a newly-seen listing (monitoring only, else null)
changeType	text	new, price-drop, or seen (monitoring only, else null)
previousPrice	number	The last known price, only on a price-drop row
priceDropPct	number	Percent drop, only on a price-drop row
firstSeenAt	date	When this listing first appeared on a run (monitoring only, else null)

Sold items: not tracked, and why (read this before you rely on "gone")

This actor does not detect sold or delisted items, and does not offer a sell-through rate. That is a deliberate, tested decision, not an oversight.

Kijiji's own search UI offers a "Most recent" (date-descending) sort next to its default "Best match" order. Detecting a sold item reliably (the approach this portfolio's Vinted and Mercado Libre actors use — see vinted-listing-lookup and mercadolibre-listing-lookup) needs a feed order you can trust run to run, so a listing that's missing can be judged "fell off the back of a trustworthy order" rather than just "the ranking reshuffled." Before writing any of that logic, this build tested the "Most recent" sort directly: appending Kijiji's own documented query-string shape (?sort=DATE&order=DESC) to a real search URL twice, live, and in both cases Kijiji silently kept serving its default "Best match" order — isSelected stayed on "Best match" and the returned listings' dates did not descend. A separate check confirmed the default order isn't stable either: two plain fetches of the identical default search a few minutes apart returned the same 40 listings but in a different order. With no confirmed way to get a trustworthy order, and no stable default to fall back on, this build stops short of gone/sell-through detection rather than fake it.

What does still work, and works well, without a trusted order: new-listing and price-drop detection. Both are keyed by Kijiji's own listing ID, not by position, so an unstable "best match" order doesn't corrupt them — a listing's ID either has been seen before or it hasn't, regardless of where it lands in the results. The one caveat: if maxListingsPerSearch is set well below totalAvailable, the exact set of IDs returned can shift slightly between runs on relevance grounds alone, which can occasionally look like movement unrelated to a real listing change. Set maxListingsPerSearch generously relative to a search's total match count if you plan to monitor it.

Price

Pay-per-event. A flat per-run fee covers session/proxy warmup; you're billed per item only when data is actually found and returned — see .actor/pay_per_event.json for exact prices. A miss is never charged.

1,000 listings through this actor costs a small fraction of a cent in our own running cost (no proxy needed for this target), priced below the category leader's $0.95 per 1,000 results — see this build's handoff doc for the exact proposed schedule. A search that returns nothing costs only the run-start fee.

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~kijiji-listing-lookup/run-sync-get-dataset-items?token=<YOUR_TOKEN>" \
     -X POST \
     -H "Content-Type: application/json" \
     -d '{"searches":["bike"]}'
   

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.

Input

{
  "searches": [
    "bike"
  ]
}

One search per line — keywords like "iphone", or paste a kijiji.ca search URL (e.g. https://www.kijiji.ca/b-ontario/bike/k0l9004). Accepted formats: bike, iphone 13, https://www.kijiji.ca/b-canada/bike/k0l0.

Sample output

query	found	status	searchQuery	listingCount	promotedCount	totalAvailable	truncated	newCount	priceDropCount	listings	listingId	url	title	description	price	priceEur	priceType	currency	location	latitude	longitude	postedAt	firstListedAt	imageUrl	imageCount	condition	brand	size	category	sellerId	sellerName	sellerRating	sellerVerified	isPromoted	isNew	changeType	previousPrice	priceDropPct	firstSeenAt	scrapedAt
bike	true	OK			<promoted (top ad) listings among them>					<all listings found (full list)>				<description>	<price (cad)>	<price (eur, ecb rate)>						<posted / last bumped (date)>	<first listed (date)>				<brand / make>			<seller (anonymous hash)>	<seller name (never available — see readme)>			<promoted (top ad) listing>	<is this listing new?>		<previous price (cad)>	<price drop %>		1970-01-01T00:00:00.000Z

A search with nothing new comes back as a found: true row with listingCount: 0 and an empty listings array, and is never charged. A search that finds nothing at all comes back as found: false, also never charged.

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~kijiji-listing-lookup/run-sync-get-dataset-items?token=<YOUR_TOKEN>" \
  -X POST \
  -H "Content-Type: application/json" \
  -d '{"searches":["bike"]}'

n8n. Add an HTTP Request node: Method POST, URL https://api.apify.com/v2/acts/accountable_eel~kijiji-listing-lookup/run-sync-get-dataset-items?token=<YOUR_TOKEN>, Body Content Type JSON, JSON Body {"searches":["bike"]} (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~kijiji-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 "Kijiji Scraper | Apify" — the agent will find and run this actor.

Monitoring: new listings and price drops

Schedule this actor (Apify Schedule, hourly or daily) with deltaMode: true and a deltaName so the watchlist survives between runs. Each run compares this search's listing IDs against a named key-value store (kijiji-listing-lookup-delta), reports only listings that are genuinely new or whose FIXED price has dropped since the last run, and stamps a NO_NEW_ROWS heartbeat, uncharged, on a quiet day. Point the schedule's webhook at Discord, Slack, Google Sheets or n8n to get a live feed instead of a page you refresh by hand.

Turn on "Skip alerts on the first run" if you're seeding a brand-new watchlist and don't want a flood of "new" alerts for listings that were already there before you started watching.

Promoted listings and monitoring. A Top Ad placement is the same underlying listing a seller paid to bump, which means it can resurface at will without being a new item. This actor never lets a promoted listing register as new or a price-drop: with monitoring off, a promoted row is returned with its delta fields null (never tracked); with monitoring on and alerts wanted, a promoted row is dropped entirely from that run's alerts (it is never billed as a false monitoring signal). It is still returned and billed normally in a plain, non-monitoring search.

Limitation, stated plainly. See "Sold items: not tracked" above for why there's no gone/sold signal. For new/price-drop specifically: keep maxListingsPerSearch generously above a search's totalAvailable (or at least a healthy fraction of it) if you want monitoring to be reliable — Kijiji's default "best match" order isn't stable run to run, so a search fetched close to its true size sees the least noise.

Tips

• Leave "Most listings to return per search" at the default (100) for a search this size; raising it pages Kijiji automatically, at a proportional cost.
• Narrow to one province with "Region" if you only care about listings in your area — confirmed live to change the actual returned listings, not just a label.
• priceType/minPrice/maxPrice filtering happens after fetching, so a very narrow filter combined with a low maxListingsPerSearch can return fewer rows than you'd expect; raise the max if so.
• Name your watchlist (deltaName) if you plan to change filters later and want to keep the same monitoring memory across that change; leave it blank and a filter change starts a fresh one automatically.
• A BLOCKED status is unusual on this target (no anti-bot posture was observed), but if you see one, lower "Max concurrency" and try again shortly.

vs. alternatives

	What it costs	What you get	Trade-off
This actor (kijiji-listing-lookup)	Priced under $1 per 1,000 listings returned, nothing for a miss — see this build's handoff doc for the exact schedule	One row per listing from a Kijiji search: price, location, condition, seller hash, image, plus optional monitoring for new listings and price drops	No sold/gone detection (see the dedicated section above) — Kijiji's "Most recent" sort didn't take effect under live testing, so this actor doesn't fake a position-based signal. No server-side price-range API was found, so min/max price filtering happens after fetching.
memo23/kijiji-scraper (category leader, 26 users/30 days)	$0.95 per 1,000 results	Search plus optional detail-page fetch	Priced per raw result rather than a keyword-search-shaped billing unit; no documented monitoring/delta mode.
Doing it yourself	Your time, plus tracking Kijiji's Apollo-cache field names and handling pagination, redaction and a seen-set for monitoring	The same data	This actor's field mapping, price-drop logic, redaction and delta-store handling are the maintenance burden it absorbs.

Prices for third-party tools are approximate, as researched September 2026, and are not tracked here — check the vendor before relying on the comparison.

FAQ

Why is a row empty, or why does found say false? Either the search returned no listings at all (status: NOT_FOUND), the input wasn't a usable search or URL (status: BAD_FORMAT), or Kijiji returned an unexpected response (status: BLOCKED). Check the message column. None of these are billed.

Does it find sold items or give me a sell-through rate? No — see "Sold items: not tracked, and why" above. This is a deliberate, tested decision: Kijiji's documented "Most recent" sort didn't take effect under two live tests, and the default order isn't stable run to run either, so there's no trustworthy feed order to judge a missing listing against. New-listing and price-drop detection are unaffected — both are ID-keyed, not position-keyed.

Am I charged for a listing I've already seen? No, not with monitoring on. Actor.charge() only fires for listings marked new or a genuine price-drop once delta mode is dropping already-seen rows. Without monitoring, every listing returned is billed once, same as any other run.

Is this a live check or a stored database? Live. Every run fetches Kijiji's own search page fresh; nothing is cached or stored on our side beyond the small watchlist state (listing IDs and last-known prices) that monitoring needs to work, which lives in your own actor's key-value store.

Why is sellerName always empty? Kijiji's search-results data genuinely doesn't include one — confirmed by inspecting every field Kijiji returns for a listing at search time across multiple live fixtures: only a numeric poster ID, a rating and a verified flag are present. A display name only shows on the individual listing's own page, which this actor doesn't fetch (that would turn one request per search into one-plus-N). This actor doesn't invent a name to fill the gap.

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 ("Kijiji Scraper | Apify"), or call the REST endpoint shown above from any script or workflow tool.

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 anonymous by default and always (there is no raw-seller-info opt-in on this actor, because Kijiji's search results carry no seller name to expose in the first place). Phone numbers and email addresses embedded in a description are redacted — confirmed necessary live: a real fixture's description contained a raw phone number and email address in plain text. Not affiliated with Kijiji or eBay Classifieds Group.

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 Kijiji's terms and your local law.

Related actors

• Marktplaats Scraper — the same keyword-search-plus-monitoring shape, for the Netherlands' largest classifieds marketplace.
• Vinted Scraper — the same shape, with full sold-item / sell-through detection where the underlying feed order is trustworthy enough to support it.