Fiverr Scraper

Pulls gig listings off Fiverr search and category pages. One row per gig, with the title, the seller, what the gig is rated, what it starts at, how fast it delivers, and the address of the listing itself.

Start with the awkward part, because it changes what you get back. A Fiverr results page holds 48 slots, and about 20 of them are paid placements. The same gig is frequently served twice: once in an ad slot, once organically. Across nine pages of one keyword, 432 slots held 289 distinct gigs. This actor deduplicates on the gig id before anything is written, so you are never handed the same gig twice and never billed for it twice. It also means a page yields roughly 30 new gigs, not 48. Plan your page count around 30.

No account, no cookie, no login. It reads the same page a signed-out visitor sees.

What you get

One row per gig:

Field	What it is
gigId, gigSlug, gigUrl	Fiverr's own id, the URL slug, and the full link
title	the gig title as written by the seller
gigImage	the listing thumbnail
sellerName, sellerDisplayName, sellerUrl, sellerImage	username, the name they display, profile link, profile photo
sellerLevel	new_seller, level_one_seller, level_two_seller or top_rated_seller
sellerCountry, sellerLanguages, sellerOnline	country code, language codes, online right now
sellerIsPro, sellerIsAgency	vetted Pro seller; listing belongs to an agency
sellerRating, sellerRatingCount	the seller's score across everything they sell
rating, reviewsCount	this gig's score and review count, which is the one people usually want
startingPrice, currency	the cheapest package, in the currency Fiverr rendered
startingPriceUsd	the same figure converted to US dollars
hourlyRateUsd	filled in only where the seller offers hourly work
deliveryDays, extraFastAvailable, packagesCount	delivery on the cheapest package, whether a rush option exists, how many packages
offersConsultation, hasSubscription	consultation offered; the gig can be bought as a subscription
tags	Fiverr's own facet values for the category
categoryId, subCategoryId	Fiverr's numeric category ids
isPromoted, isFiverrChoice, isFeatured	paid slot, Fiverr's Choice badge, featured
searchLabel, searchUrl, page, position, scrapedAt	where the row came from

A real row, from a run on 20 September:

{
  "gigId": "156202007",
  "title": "create a professional minimalist logo design",
  "gigUrl": "https://www.fiverr.com/kairachel451/create-a-professional-logo-design",
  "gigImage": "https://fiverr-res.cloudinary.com/t_main1,q_auto,f_auto/gigs/156202007/original/6986058e637c412b45483f0f5f51f2b57e2de93c.jpg",
  "sellerName": "kairachel451",
  "sellerDisplayName": "Rachel",
  "sellerUrl": "https://www.fiverr.com/kairachel451",
  "sellerLevel": "top_rated_seller",
  "sellerCountry": "PK",
  "sellerIsPro": false,
  "sellerLanguages": ["en"],
  "sellerRating": 4.93,
  "sellerRatingCount": 14304,
  "rating": 4.9,
  "reviewsCount": 14182,
  "startingPrice": 25,
  "currency": "USD",
  "startingPriceUsd": 25,
  "deliveryDays": 3,
  "packagesCount": 3,
  "tags": ["3d", "monogram", "wordmark", "pictorial", "ai", "jpg", "pdf", "png", "psd", "eps", "svg"],
  "categoryId": 3,
  "subCategoryId": 49,
  "isPromoted": false,
  "page": 1,
  "position": 8
}

Note the title and the URL slug disagree. Sellers rename gigs and Fiverr keeps the original slug, so read title for what the gig is called today and gigUrl for where it lives.

Input

Give it keywords, or paste listing addresses, or both.

{
  "searchQueries": ["logo design", "wordpress developer"],
  "startUrls": [
    { "url": "https://www.fiverr.com/categories/graphics-design/creative-logo-design" }
  ],
  "maxItems": 200,
  "maxPagesPerSearch": 5,
  "sellerLevels": ["level_two_seller", "top_rated_seller"],
  "maxStartingPriceUsd": 100,
  "maxDeliveryDays": 3
}

maxItems is the ceiling on rows returned and therefore the ceiling on what the run can cost. Duplicates, filtered-out gigs and failed pages are all free, so nothing else can push past it.

The filters (sellerLevels, proOnly, minRating, minReviews, maxStartingPriceUsd, maxDeliveryDays, excludePromoted) are applied to rows after the page is read. A gig dropped by a filter is not written and not charged. Verified on real runs: asking for top-rated sellers only returned 19 rows, all of them top rated; asking for a $30 ceiling and 2-day delivery returned rows topping out at exactly $25 and 2 days.

Leave the proxy settings alone unless you have your own. The run already handles Fiverr's page limits on its own.

What it does not do

This is the part worth reading before you buy.

• Listing pages only. It does not open individual gig pages, so there is no package comparison table, no full gig description, no gig FAQ and no buyer reviews. What you get is what a results page carries.
• No seller profiles. sellerRating and sellerRatingCount come off the listing card. It does not visit the seller's page for their portfolio, description or full review history.
• Fiverr stops at page 21. Page 22 answers 404. That caps one keyword at 1,008 slots, which after deduplication is roughly 600 gigs. If you need more than that from one topic, split it into narrower keywords or work through category pages.
• tags are Fiverr's category facets, not seller tags. For logo design you get style, logo type and file formats. A different category returns entirely different keys, and about 6% of rows carry none at all. That was 853 of 909 rows on one run and 2,020 of 2,155 on another.
• No sort control, on purpose. Fiverr ignores sort_by on the page it serves to a signed-out visitor. Three different sorts were measured returning the same gigs in the same order, so offering a sort here would be a button that does nothing. Same for Fiverr's own seller-level and Pro parameters. Use this actor's filters instead, which do work.
• Fiverr never really returns nothing. A deliberately nonsensical keyword still came back with 40 gigs, because the search falls back to loose matching. If your keyword is too narrow you get approximate results rather than an empty set, and those are real rows and are charged. Use specific keywords you would actually type into the site.
• Prices depend on where the request came from. Fiverr converts the page into a currency it picks from the requesting address, so startingPrice may arrive in EUR or ZAR. startingPriceUsd carries the US-dollar figure using the conversion rate Fiverr itself reported on that page. If you only care about one number, use that one.
• startingPrice is the cheapest package. It is not the average order value, and it is not what most buyers of that gig actually pay.
• Promoted gigs are kept by default. They are real gigs sitting in paid slots, flagged as isPromoted. Turn on excludePromoted if you only want organic results.
• Anything behind a Fiverr login is out of scope.

Billing

Pay per gig. One charge per gig actually written to the dataset. The current rate is on the Pricing tab next to the Input tab.

What is free:

• duplicate slots, where Fiverr served the same gig twice
• gigs dropped by your filters
• searches that come back empty or blocked, which write an explanatory row instead
• the sample row you get from an empty input
• a run that fails outright

So the most a run can cost you is maxItems gigs plus the run fee, and usually less.

Every failure writes a row saying what happened, with charged: false on it, rather than throwing. A failed run would still bill the run fee, and nobody should pay to be told they left a field empty.

Reliability, measured

Numbers from real runs on 19–20 September 2026:

• 20 varied keywords in one run (logo design, wordpress developer, seo audit, voice over, unity game development and fifteen others) returned 909 distinct gigs with no failed searches.
• A deep run of 5 keywords × 21 pages returned 2,155 gigs from 104 of 105 pages. The one page that came back without data wrote an uncharged row saying so.
• Price, rating, review count and delivery time were present on 100% of those rows.

When Fiverr refuses a page, the actor tries again before giving up. It also catches a specific thing Fiverr does that is easy to miss: answering HTTP 200 with a short page holding eight slots, all of them advertising, where the same query reported 187,000 matches a second earlier. That is not an empty result and it is not an error code, so it is detected by the absence of organic listings and fetched again.

FAQ

Do I need a Fiverr account or cookies? No. It reads the pages Fiverr serves to a signed-out visitor.

Can I scrape a whole category? Yes. Paste the category address into startUrls, for example https://www.fiverr.com/categories/graphics-design/creative-logo-design, and set maxPagesPerSearch. The 21-page ceiling applies to categories as well.

Why did I get fewer gigs than pages × 48? Because about 20 of the 48 slots on each page are advertising, and the same gig often appears in both an ad slot and an organic one. Those repeats are removed and not charged. Around 30 new gigs per page is normal.

How do I get only organic results? Set excludePromoted to true. Skipped ad slots are not charged.

Why is startingPrice in euros? Fiverr localises the page by the requesting address. Read startingPriceUsd instead. It is converted using the rate Fiverr reported on that same page.

Can I filter by seller level or Pro status? Yes, through sellerLevels and proOnly on this actor. Do not put Fiverr's own filter parameters into a pasted URL and expect them to work. Fiverr ignores them on this page, which was measured rather than assumed.

Can I sort by price or by newest? No, and no actor reading this page honestly can. Fiverr ignores the sort parameter here. Pull the rows and sort them yourself.

What happens if Fiverr blocks the run? You get a row with errorCode: BLOCKED explaining it, and you are not charged for that search. Re-running a few minutes later usually works, since the limit is applied per address.

Does it get gig descriptions, packages or reviews? No. This reads listing pages. Individual gig pages are a different job.

Can I use my own proxies? Yes. Put them in proxyConfiguration.proxyUrls under Advanced and they are used exactly as given.