"$49/mo" buys you raw HTML or a credential. It doesn't write your parser, host your scheduler, watch your runs, or rebuild the pipeline the week the site changes. On a hard target it also doesn't succeed - it just charges you for the failed attempts. Here's the real cost, and when a flat retainer is cheaper.
The headline price on a scraping API is the price of an attempt, not a result. Between that attempt and a row in your database sits everything the plan does not include: the parser you write, the scheduler you host, the runs you babysit, the rebuild you do when the markup shifts or a new anti-bot layer switches on, and the proxy add-ons you bolt on when the default pool starts getting blocked.
None of that shows up on the pricing page. All of it shows up in your total cost. The honest comparison is not "$49/mo vs a retainer" - it is everything below vs a flat fee that prices the outcome.
Line items, not a fabricated "$X saved" number. Judge it against your own targets.
| Cost line | DIY + scraping API | Data Works retainer |
|---|---|---|
| Tool / credits | $49-$999/mo (multiplies 5-75x on hard sites) | Included |
| Engineer time (build) | Days-weeks of a dev | Included |
| Engineer time (maintain) | Ongoing, spikes on every site change | Included |
| Proxies for anti-bot | Extra add-on, often still fails | Dedicated carrier IPs, included |
| Failed-request waste | You pay for 200-OK-but-blocked pages | We eat the retries |
| Who's accountable when it breaks | You, at 2am | The engineer who built it |
| Predictability | Credit bill swings with target difficulty | Flat monthly |
Scraping-API plan prices and credit multipliers are as of mid-2026, from vendors' public pricing - verify against their current pages. Verified July 2026.
For one or two easy sites and an engineer who wants to own the scraper, DIY plus a cheap API is genuinely the right, cheaper call. The retainer wins when the targets are hard, the breakage is constant, or the engineer-hours cost more than the flat fee.
If you are in the left column, we will send you away with a clear conscience.
Some vendors advertise "you only pay for successful requests." True and fair - but a 200-OK page that came back blocked still bills as success, and it never refunds the engineer-hours when your parser breaks. A retainer prices the outcome (rows delivered), which is the only metric that pays your rent.
On easy sites, yes - genuinely. On hard ones, the credit multipliers, the engineer time to build and maintain the parser, and the failed-request waste usually add up to more than a flat retainer. The TCO table above walks the line items one by one.
You can, and on hard targets it still often fails on IP reputation - anti-bot systems score the exit IP before any challenge runs. Dedicated carrier IPs are the durable fix; our anti-bot page explains the mechanism.
Then DIY is genuinely right - and we will tell you so on the scoping call instead of selling you a retainer you do not need.
Tell us your targets and we will do the honest math with you - including the cases where a cheap API really is the better buy.