Web Scraping Services: Pricing and When You Don't Need One

Nine of the ten companies below publish entry prices on their websites. One wants a sales call first. If you’re shopping for a web scraping service, that’s the good news: you can comparison-shop with real numbers, which was impossible five years ago. The bad news is that the published numbers hide three different meters, and picking by headline price is how teams end up paying $49 a month for what a cron job would do for free.

Every price here was live on the vendor’s own pricing page the day this was written, October 2026. At the end there’s a five-question test for when you don’t need to pay anyone, because a rising share of people searching for a webscraping service or a web scrape service actually need neither.

The best web scraping services at a glance

Ten companies, three business models, one table. Ranking logic: I grouped by what they actually sell, then ordered by published entry price within each group.

Company	What it sells	Published entry price	Scales to	The catch
ScrapingBee	Scraping API, credit-metered	$49/mo, 250,000 credits (1,000-credit free trial)	Higher tiers with stealth proxies	Stealth-proxy pages burn credits 75× faster
Oxylabs	Scraper APIs + proxy networks	Starter $30/mo, plus $5 welcome credit	Enterprise from $2,500/mo	Bills per 1K results; big catalogs add up
Apify	Marketplace of scrapers + compute	Free plan with $5 of credits; Starter $19/mo	Business $999/mo	Most store actors bill per result on top of compute
Zyte API	Metered extraction API	$5 free credit; pay-as-you-go from $0.13 per 1K pages	$500/mo commitment drops rates to $0.06 per 1K	Browser rendering raises the rate to $16.08 per 1K
Bright Data	Unblocker API + ready-made datasets	Web Unlocker free 5K requests/mo; then $1.50 per 1K	Scale plan $499/mo (383K requests included)	Best rates need monthly commitments
Browse AI	No-code page monitors	Free 50 credits; $19/mo billed annually	Higher tiers with more monitors	1 credit = 10 rows; premium sites cost 2–10 credits per run
Kadoa	AI adaptive extraction	No published prices, free trial	Custom quote	Quote-gated; you can’t do the math before the call
Datahut	Managed data feed	$99/mo per website, up to 25K records	Business $249/mo; Enterprise custom	Per-website billing adds up across many sources
Grepsr	Managed, record-based	One-time projects from $100	Growth and Enterprise packs, custom	Recurring packs require a quote
ScrapeHero	Managed, full service	$550 one-shot per site; $199/mo per website	Enterprise Premium from $8,000/mo	Setup fees on subscriptions; redirects count as pages

Three scraping vendors changed prices this year alone (ParseHub $149 to $189, Octoparse $83 to $69, Apify rebuilt its plan structure entirely), so treat any table like this one as a snapshot, not a constant.

What a web scraping service actually sells

Strip the marketing and every vendor here sells one or more of three things: unblocking infrastructure (proxies, CAPTCHA handling, headless browsers), scheduling and delivery, or human labor. The price differences make sense once you know which one you’re buying.

A site scraping service in the metered tier is mostly a proxy bill with an API on top. Residential bandwidth runs $6 to $8 per gigabyte at Oxylabs and Apify’s published rates, and that’s their cost center, not their margin. A website scraping service in the managed tier is mostly payroll: an engineer builds the scraper, QA checks the output, and someone keeps it alive when the site redesigns. That’s why managed pricing starts where self-serve pricing ends.

Metered APIs: you pay per request

The meter counts requests, credits, or results. ScrapingBee’s 250,000 credits sound enormous until their multiplier table applies: a simple static page costs 1 credit, but a page that needs the stealth proxy costs 75. Zyte and Oxylabs meter per 1,000 pages or results instead, which is at least divisible in your head. This tier suits data scraping services that feed an app or dashboard continuously, where per-unit costs stay predictable.

Subscriptions: you pay to belong

Apify and Browse AI charge a monthly fee that includes some usage. The fee buys the platform; the usage buys the data. Apify’s FAQ states plainly that unused credits expire at the end of the billing cycle, and Browse AI’s 1 credit per 10 rows means a 5,000-row extract eats 500 credits. Subscriptions favor steady, repeated jobs where the fee amortizes.

Managed: you pay for people

You send a spec; humans build and run the pipeline and deliver CSV, JSON, or an API feed. Grepsr prices by records, Datahut by website per month, ScrapeHero by site and refresh frequency. This is the tier where a data scraping companies search actually lands, and where the price covers maintenance you’d otherwise do yourself.

Self-serve APIs: four price lists, checked this week

ScrapingBee starts at $49 a month for 250,000 credits, with 1,000 free trial credits and no card required. The credits are the trap: their help center publishes the multipliers, and any target behind serious bot protection bills at 75 credits per page. Do that division and the Freelance plan covers roughly 3,300 protected pages a month. Fine for a known-light workload; expensive as a surprise. (We compared it line by line in our ScrapingBee alternatives page.)

Zyte is the company behind Scrapy, and its API prices like an engineer’s tool. Pay-as-you-go HTTP requests run $0.13 to $1.27 per 1,000 responses depending on site complexity, with a $5 free credit and no subscription. The honest headline is the browser column: rendered pages run $1.01 to $16.08 per 1,000, which is the real price of JavaScript at scale. Commit $500 a month and the simple tier drops to $0.06 per 1,000.

Oxylabs reorganized everything into one subscription: Starter $30, Basic $100, Flexible $500, Enterprise from $2,500 a month, all products included. Its result-metered APIs bill per 1,000 results, from $0.50 on the Starter tier down to $0.25 at ten million results. A $5 welcome credit lets you test before committing. The result-based meter is fair for product catalogs where each row is a parsed entity, less fair when your “results” are thin pages.

Bright Data is the enterprise heavyweight: 50,000+ customers, and a free Web Unlocker tier of 5,000 requests a month that costs nothing to try. Pay-as-you-go runs $1.50 per 1,000 successful requests, and they only charge for success. The interesting product for non-developers is the dataset marketplace: ready-made extracts (Amazon products, LinkedIn profiles, Google Maps places) from $2.50 per 1,000 records, with refresh discounts up to 80% for monthly subscriptions. Buying a dataset instead of commissioning a scrape is the shortest path to data when your target is popular.

Platforms: the middle shelf

An online scraping service sounds like something you do in a browser tab, and on this shelf it is. You point, click, and schedule without writing code, and the platform runs it in their cloud. If your mental model of an online web scraping service is “set up a monitor, get emails when it changes,” this is your aisle. Most online web scraping services work this way; the three below are the ones worth a tab.

Apify is less a tool than a marketplace: developers publish actors (scrapers) that anyone can run, most billing per result on top of your plan’s compute, and the flagship web-scraper actor alone counts 125,000+ users. The free plan includes $5 of usage, Starter is $19 a month, and Business is $999. Read the actor’s own pricing page before running anything; the platform fee is rarely the whole bill. Our Apify alternatives comparison does the actor math on a real job.

Browse AI monitors pages and emails you the diff. The free tier gives 50 credits (500 rows), and $19 a month billed annually buys 12,000 credits a year. It’s the cheapest way to watch 10 competitor pages for price changes, and the wrong tool for bulk extraction. Full breakdown in our Browse AI review.

Kadoa is the AI-native entry: describe the schema, and it figures out the selectors, adapting when sites change. It’s also the one company here with no public prices at all. The free evaluation exists, but consumption pricing comes as a quote. That’s not disqualifying; it just means you negotiate without a benchmark.

Custom web scraping: when people do it for you

Managed is the tier most searchers mean by web data scraping services: you never touch a scraper, and someone answers the phone when the data looks wrong. Most web scraping companies refuse to tell you what this tier costs, which is exactly why the three below, all with published entry prices, top any honest ranking.

Datahut starts at $99 a month per website for up to 25,000 records, delivered as CSV or JSON. Business, at $249, raises the cap to 100,000 records and adds API delivery. Per-website billing is clean for one competitor feed and punishing for a twenty-source portfolio. Their pitch: “no hidden fees.” Their site claims six of the ten largest global retailers as customers.

Grepsr prices one-time extractions from $100, based on records rather than requests, which is the friendliest way possible to buy a single dataset. Among the website data scraping services in this tier, records beat pages as a billing unit whenever targets paginate. Recurring work moves to custom-quoted Growth and Enterprise packs. Fourteen years in business, and the concierge model shows: you get an account manager, not a dashboard.

ScrapeHero publishes the most complete price ladder of the top web scraping companies, and deserves credit for it: $550 one-shot per site (1,000 to 5,000 pages), $199 monthly per website, Enterprise Basic from $1,500, Enterprise Premium from $8,000 a month. Two details matter when you compare. Setup fees are separate on subscription plans. And their own billing FAQ says a “page” is an HTTP request, so redirects and XHR calls can count as pages. That’s not sneaky; it’s disclosed. But it means your 1,000-product catalog might bill as 3,000 pages if the site chains redirects. The best web scraping companies for enterprise work are the ones that answer this question before you ask.

What web scraping solutions charge for, line by line

Compare an Oxylabs quote against your own AWS bill and the margin looks enormous. It isn’t, because the published proxy price is not the product. Data scraping solutions in the API tier bundle: rotating residential IPs that survive a day before burning, browser fingerprints current enough to pass, CAPTCHA budgets, and the 3 a.m. work of noticing a target changed its markup. Buying that stack piecewise costs engineers you haven’t budgeted.

The managed tier adds the second expensive thing: accountability. When ScrapeHero charges $8,000 a month, part of that is SLA, compliance paperwork, and a person who explains why yesterday’s feed was 4% short. You can replicate the scraping. Replicating the accountability is harder. That’s the honest case for the data scraping solutions at the top of the market, and no amount of local tooling replaces it at enterprise scale.

When you don’t need a service at all

Five tests, in order. If you pass any of them, close the pricing tabs.

The site already offers the data

An export button, an official API, an RSS feed, or a bulk-download portal beats any vendor at any price. Companies pay scraping APIs to fetch pages that the same site’s API serves free and structured. Check the footer, the developer portal, and robots.txt before you shop.

Small, public, and static

I timed this today: the Hacker News front page, 30 rows, parsed by a 20-line Python script in 0.8 seconds; a local scraper tool did the same page cleanly in 2.2. A thousand pages a month of straightforward HTML is a lunch-hour script, not a $49 monthly subscription. The free tier roundups at our free scraping APIs list cover the cases in between, and the open-source tools list covers the DIY end.

You need it once

A one-time analysis of five product pages doesn’t need infrastructure; paste the HTML into an LLM and ask for JSON. For a one-time 4,000 pages, Bright Data’s free Web Unlocker tier covers it within a month at zero cost, and Grepsr’s $100 one-time pack is cheaper than configuring anything.

The data can’t leave your machine

Compliance teams kill vendor quotes for a reason: customer data, licensed content, or internal tools behind SSO often can’t be shipped to a third party’s cloud. Local tools sidestep the question entirely. One factual aside, since it’s the category we build in: Magpie is a free open-source CLI (AGPL-3.0-or-later, or a commercial license) that scrapes, crawls, and monitors on your own hardware; the Desktop app adds a GUI with 5 free runs, then $49 a month. It doesn’t defeat CAPTCHAs or bot walls, by design, which is exactly why it passes the compliance test the cloud vendors fail. The trade-off is cloud versus local in one page.

Recurring but tiny

A page checked once an hour, four fields, one email on change. That’s a cron job and 30 lines of code, or a local watch command. Honest caveat: your hour is not free. If maintenance eats more than two hours a month, the $99 managed minimum is cheaper than you are. Below that line, you’re subscribing to avoid a script you’d enjoy writing.

When paying is the right call

Volume past ~50,000 pages a month with any anti-bot resistance is where DIY dies quietly: proxies get burned, headers go stale, and the failure rate becomes your problem. Targets behind real walls (social platforms, marketplaces with login gates) are what Bright Data’s datasets and Oxylabs’ enterprise tiers exist for. And no engineering team is the strongest argument of all: a managed feed at $99 a month beats a half-built internal scraper that one departing contractor understood. Buying is also the legally safer default for anyone uncertain about the rules; start with our legality explainer, then confirm with a lawyer who knows your jurisdiction.

Red flags before you sign

• Multiplier tables. Ask what your target class actually costs per page, not the 1-credit brochure number. ScrapingBee publishes theirs (1/5/10/25/75); make every vendor match that disclosure.
• Page definitions. Get the redirect and XHR billing rule in writing. ScrapeHero discloses theirs in an FAQ; hold others to the same standard.
• Credit expiry. Apify’s credits lapse monthly. If you’re spike-shaped (one big run a quarter), expiring credits triple your real cost. If you need proxy help for a DIY build instead, the proxy guide covers the raw market.
• Scraper ownership. On managed plans, ask who owns the code when you leave. “Access to the Grepsr Platform” is not the same as a pipeline you can self-host.
• Quote-only pricing. Not a scam, but demand the baseline table ScrapeHero publishes. Vendors who won’t anchor a number are pricing on what they think you’ll pay.

Web scraping services FAQ

How much does a web scraping service cost?

Self-serve entry points run $19 to $49 a month (Apify Starter, ScrapingBee Freelance, Oxylabs Starter at $30). Managed plans start at $99 a month per website (Datahut) or $550 as a one-time job (ScrapeHero). Enterprise managed contracts run from $1,500 to $8,000+ monthly. All figures from vendor pricing pages, October 2026.

Is there a free web scraping service?

Bright Data’s Web Unlocker includes 5,000 requests a month free, Apify grants $5 of credits, and a local tool costs nothing to run. “Free” ends where volume, JavaScript, or bot walls begin. For the full free tier comparison, see our scraping API roundup.

What’s the difference between a scraping API and a managed service?

An API sells infrastructure: proxies, browsers, CAPTCHA handling. You keep the target logic and the maintenance. A managed service sells the data itself: they build, run, and repair the scraper. Developer teams with shifting targets fit APIs; everyone else, or anyone with a stable recurring feed, fits managed.

Is it legal to buy scraped data?

Buying doesn’t launder the source. Public, non-personal data is the broadly safer zone; personal data and login-walled content are where buyers and providers both get burned, and rules differ by jurisdiction. Read the legal explainer for the framework, then run anything people-shaped past a lawyer.

Should I build or buy?

Buy when scraping is adjacent to your business and the feed is stable; a $99 managed plan is cheaper than engineer hours. Build when collection is core product infrastructure at real scale, when data can’t leave your infrastructure, or when the job is small enough for a script.

The short version

Under a thousand public pages with no walls: script it or run a local tool, spend $0. Recurring feeds and no developer: Datahut or Grepsr, from $99 to $100. Walled targets or API-scale volume: Oxylabs or Bright Data, from $1.50 per thousand requests. Enterprise: ScrapeHero publishes the honest ladder, $550 to $8,000 a month. Start at the free tiers either way; nine of these ten companies will let you test for nothing, which is more than the eleventh, the one selling the quote, will ever say out loud.