Web scraping tools comparison: cloud platforms vs self-hosted scripts
Cloud scraping platforms rent you infrastructure. You pay a subscription plus metered usage for every run that executes on their servers, and the scrapers, proxies, and collected data live in their account. Self-hosted script templates flip the model: one purchase, the code runs on your machine, and no meter counts your runs. This page compares the two on pricing, control, and maintenance, then ends with when a cloud platform is the right choice, because for some jobs it plainly is.
How cloud platforms work and price
Apify is the reference point, and it is a good product, so this is a description rather than a complaint. You build or rent an Actor, a packaged scraper. Runs execute on Apify's servers. Billing has two layers: a subscription by plan tier and platform usage credits measured in compute units. One compute unit is 1 GB of memory held for one hour, so a run's memory footprint and duration feed directly into the bill. Actor authors set their own rental prices on top, monthly or per result.
Other platforms shape the same idea differently. ScrapingBee meters API calls, Browserless meters browser sessions, and both sit in front of headless infrastructure you never see. The shared properties of the cloud model:
- Every run consumes metered units on someone else's account, so cost scales with usage and never drops to zero.
- Scrapers and results live in platform storage under the retention limits of your plan until you export them.
- Scaling means upgrading a tier or burning more credits, which you can do without touching infrastructure.
That trade, no ops work in exchange for variable billing and custodial data, is exactly what many teams want. The question is whether it is what you want.
Side by side
| Dimension | Cloud platform (Apify and similar) | Self-hosted templates (this pack) |
|---|---|---|
| Cost model | Subscription tier plus metered compute units or API calls; spend varies with run volume | One-time purchase, then free per run |
| Where code runs | Platform servers, under platform limits and logs | Your machine or your own server |
| Where data lives | Platform storage, plan-based retention, export to get it out | Local files: CSV, JSON, JSONL you write where you point the script |
| Source access | Actors you rent are black boxes; your own actors are editable but run only on the platform | Full JavaScript source, readable and editable, no runtime dependency on the seller |
| Setup | Account, plan, Actor config or no-code editor | Node 18+, one npm install, one chromium download |
| Scaling up | Upgrade tier, add credits; heavy concurrency handled for you | You manage concurrency, proxies, and scheduling yourself |
| Lock-in | Actors, schedules, and stored results are tied to the platform account | None; plain Node.js files that run anywhere Node runs |
| Good fit | Volume jobs, no-code teams, managed anti-bot infrastructure | Developers who want fixed cost, control, and code they can change |
The self-hosted side: what the pack buys
The Web Scraping Scripts Pack is ten self-contained Node.js scripts built on Playwright: paginated tables to CSV, infinite-scroll collection, login sessions with saved storage state, price and stock monitors with webhook alerts, bulk form submission, job board scraping with cross-run dedupe, a robots-aware sitemap crawler, screenshot diffing, API polling to JSONL, and a retry wrapper with backoff and user-agent rotation. The manifest is a free download if you want the full inventory first.
You run every script yourself. That is the whole point: no account, no API key, no per-run meter, and no upload of anything you collect. Each file opens with a how-to-run comment and marks the selectors to edit with an EDIT constant. If a site changes its markup, you fix the selector in your own copy instead of waiting for a third-party Actor's maintainer to notice.
The honest cost of that control is maintenance. You patch selectors when sites change, you supply proxies if a target blocks your IP, and you schedule runs with cron or your own scheduler. The pack hands you the patterns; operating them stays your job.
When a cloud platform is the right choice
For these situations, pay for the platform and skip this pack:
- You scrape heavy JavaScript sites at real volume and want managed browsers, proxies, and anti-bot handling that someone else keeps alive.
- Your team is not made of developers, and a no-code editor plus hosted schedules beats editing JavaScript files.
- You need hundreds of concurrent runs and billing that flexes with usage instead of hardware you own.
- You rely on a specific marketplace Actor that already does the job well, and renting it costs less than building and maintaining your own.
At steady heavy volume, a platform's unit pricing can also land below the engineering time of self-hosting. If your monthly bill is USD 50 and replacing it would eat ten hours of work, the platform is winning. That is a legitimate outcome, and the subscription is buying real operations.
When self-hosted scripts make more sense
Choose the pack when most of these hold:
- Your targets are modest: a few sites, a few thousand pages, daily or hourly refreshes rather than millions of requests.
- You write code already, and you want scrapers you can read, diff in git, and change without asking a platform for permission.
- Predictable cost matters more than elasticity. Paying once beats a subscription that rises with every successful run.
- The data should not pass through a third party: client work, internal monitoring, or anything where you would rather keep collection local.
- You have been burned by an Actor breaking or a plan tier changing under you, and you want the failure modes in your own hands.
One line sums the trade: the platform sells operations, the pack sells the code, and only you know which one you are short of.
Common questions
Is this pack an Apify alternative?
For developers running their own scrapers at small to medium scale, yes: it replaces the per-run metering and custodial storage with local files and a one-time price. It does not replace Apify's managed proxies, no-code editor, or marketplace of ready Actors. If those are what you are paying for, keep paying for them.
What does it cost to run after buying?
Nothing per run. The scripts execute on your machine against your own Node.js install. Costs beyond the one-time purchase are whatever you already pay for hardware and, if a target requires them, proxies.
Can I modify the scripts?
Yes. You get the full source, personal and commercial use, and no restriction on editing for your own projects. The one limit is redistribution of the files as-is, stated on the tooling page.
Checkout through Getly. Full source included; you run everything on your own machine.