String vs Scrapingdog (2026): 100 sites compared
Which is better, String or Scrapingdog?
String returned verified content on 97.0% of requests in the open Web Data Frontier Benchmark. Scrapingdog returned 45.6%. The September 16, 2026 run covered 100 real sites with five attempts each.
- Choose String if the dataset spans many protected domains. String returned 97.0% of requests and Scrapingdog 45.6%.
- Choose Scrapingdog if price decides it, at $0.09 per 1,000 plain requests, or one of its 40+ dedicated APIs saves you parsing work.
- Do not switch if your targets are covered by Scrapingdog's dedicated endpoints and already return complete data.
Scrapingdog is cheaper. Its $90 Standard plan includes enough credits for 100,000 premium-proxy requests, while String Growth is $99 a month, all of it usage. String's advantage is coverage: 485 successful responses against 228, with 2 target shutouts against 32. Pick Scrapingdog for low-cost volume on known targets and its catalog of dedicated APIs. Pick String when a varied protected target list has to come back.
TL;DR
- String passed 485 of 500 requests, or 97.0%. Scrapingdog passed 228, or 45.6%, down 8.7 points on the August run.
- Scrapingdog returned nothing on 32 of 100 sites. String passed 29 of those 32 on every attempt.
- Scrapingdog beat String on one target, Foot Locker, 5 of 5 against 4, and passed all five attempts on 22 sites.
- String's failure-aware latency score was 7.06 seconds, the fastest of 16. Scrapingdog's was 26.55 seconds, the slowest; in August it was third-best at 12.51.
- Scrapingdog Standard is $90 per month for 1,000,000 prepaid credits and 50 concurrent requests.
- Scrapingdog's plain request costs 1 credit, JavaScript rendering 5, and premium proxies 10. Failed requests are refunded.
- Scrapingdog includes 40-plus dedicated APIs, official Node.js and Python SDKs, and an MCP server on every plan.
- Every number comes from a public benchmark repository that accepts your own API keys.
| String | Scrapingdog | |
|---|---|---|
| Product type | Unblocker: fetch, search, sitemap, remote browser | Web Scraping API plus 40-plus dedicated search, retail, social, and AI APIs |
| Benchmark success | 97.0% (485/500), 1st of 16 | 45.6% (228/500), 13th of 16 |
| Latency score | 7.06s, 1st of 16 | 26.55s, 16th of 16 |
| Sites with no successful attempt | 2 of 100 | 32 of 100 |
| Billing near $100/mo | Growth: $99, all of it usage at $0.16 to $3.25 per 1,000 successful requests | Standard: $90 prepaid, 1,000,000 credits |
| Failure billing | Blocked requests are free | Failed, blocked, and timed-out requests are refunded |
| Tested configuration | Automatic fetch-path and proxy selection | Premium residential proxy, raw HTML, JavaScript rendering off |
| MCP and SDKs | Hosted MCP registry entry | MCP plus Node.js and Python SDKs |
| Where it wins | Coverage, fewer shutouts, verified-content billing | Price, dedicated API breadth, explicit controls |
How we tested, and our conflict of interest
We built the benchmark and sell the product that ranks first. The target definitions, methodology, provider adapters, and September results are public so buyers can inspect the test and rerun it.
Each provider received the same 100 URLs five times. A response passed only when it contained a marker from the real page. A captcha or block page that returned HTTP 200 failed.
The Scrapingdog adapter set premium: true and dynamic: false. That routed through Scrapingdog's premium residential proxies and returned raw HTML without JavaScript rendering. Premium proxies are its strongest published anti-bot proxy path. Turning on dynamic could improve JavaScript-only pages, but this run does not prove how the combined configuration performs.
What each product is
Scrapingdog's generic Web Scraping API handles rotating proxies, captcha bypass, optional JavaScript rendering, country targeting, sessions, headers, and raw HTML. It also has a Screenshot API and a rotating proxy endpoint that routes recognized domains into dedicated APIs.
The larger differentiator is its API catalog. One key unlocks more than 40 endpoints for Google Search, Maps, News, Shopping, Trends, and AI Mode; Amazon products, search, reviews, and offers; LinkedIn profiles and jobs; YouTube; TikTok; X; Walmart; eBay; Zillow; Yelp; and more. Those endpoints return structured JSON rather than generic page HTML.
String is narrower. Send a URL to one fetch endpoint and receive markdown, HTML, or JSON. Proxy selection, fingerprints, captcha handling, rendering, and retries happen behind the request. String also has /search, asynchronous sitemap discovery, a remote browser over CDP, and MCP tools. It does not offer Scrapingdog's large catalog of domain-specific parsers.
The same request in Scrapingdog and String
One job, done both ways: fetch a JavaScript-heavy product page and get its content back. Each request follows the vendor's own docs for that job, checked against those docs on October 15, 2026. Each one reads its key from an environment variable, such as $STRING_API_KEY.
curl "https://api.scrapingdog.com/scrape?api_key=$SCRAPINGDOG_API_KEY&url=https%3A%2F%2Fwww.example.com%2Fproduct%2F123&dynamic=true&premium=true"
curl https://request.usestring.ai/v1/fetch \
-H "Authorization: Bearer $STRING_API_KEY" \
-H "Content-Type: application/json" \
-d '{"url": "https://www.example.com/product/123", "format": "markdown"}'
Scrapingdog takes the key in the query string and returns raw HTML; markdown needs formats=markdown. dynamic=true renders JavaScript and premium=true moves to residential proxies, 25 credits together against 1 for a plain call. Scrapingdog's own pages disagree on whether dynamic is on by default (on versus opt-in), so set it explicitly. String takes the key in a header, picks the fetch path and proxy class for each request, and bills the path it used. Add "executeJS": true if you want to force a browser.
Success rate: String 97.0%, Scrapingdog 45.6%
String passed 485 of 500 requests. Scrapingdog passed 228, a gap of 257 successful responses across the same target list.
| Per-target outcome | String | Scrapingdog |
|---|---|---|
| Passed all 5 attempts | 93 | 22 |
| Passed 1 to 4 attempts | 5 | 46 |
| Passed no attempts | 2 | 32 |
Scrapingdog's 22 perfect targets are Canada Goose, Kayak, Wall Street Journal, Google, Zoopla, Foot Locker, Facebook, Asda, eBay, GOAT, Tripadvisor, Vinted, Airbnb, Macy's, Alibaba, Booking.com, CoinMarketCap, TradingView, Nike, Delta, GitHub and Bing. For workloads concentrated on those sites, its lower price is compelling.
The 32 shutouts include Temu, Mouser, Crunchbase, Lowe's, Realtor.com, Skyscanner, Expedia, Louis Vuitton, Hyatt, Neiman Marcus, Saks Fifth Avenue, Idealista, StockX, Glassdoor, Ralph Lauren, Etsy, Yahoo Finance, Allegro, Indeed, Monster and LinkedIn. String passed 29 of the 32 on every attempt. Temu and Idealista defeated both products, and String passed four of five on Allegro.
The bigger change is the middle column. Scrapingdog now returns content on some attempts and nothing on others for 46 of the 100 sites, against 24 in August and String's 5. Almost half its target list produces a different dataset on every run.
Scrapingdog beat String on one target. On Foot Locker it returned all five attempts to String's four.
Anti-bot coverage
Scrapingdog's premium proxy result was strongest on Kasada and on sites without a named commercial anti-bot vendor. PerimeterX is its weakest multi-site category at 26.7%, and it returned nothing at all on the single Fastly target, which it cleared in August.
| Anti-bot system | Sites | String | Scrapingdog |
|---|---|---|---|
| Cloudflare | 15 | 100% | 40.0% |
| DataDome | 19 | 92.6% | 37.9% |
| Akamai | 20 | 98.0% | 49.0% |
| PerimeterX | 9 | 100% | 26.7% |
| AWS WAF | 6 | 100% | 43.3% |
| Kasada | 3 | 93.3% | 66.7% |
| Fastly | 1 | 100% | 0% |
| No named vendor | 27 | 96.3% | 57.8% |
String led in every category. Across the seven multi-site categories, the gap ran from 27 to 73 points. JavaScript rendering could change individual misses, so buyers should repeat this table against their targets with dynamic=true where the content requires it.
Speed: String leads on both measures
String's failure-aware latency score was 7.06 seconds, the fastest of 16. Scrapingdog's was 26.55 seconds, the slowest of 16, having placed third in August at 12.51. On the 21 sites both products returned on all five attempts, the median per-site time was 1.70 seconds for String and 21.00 for Scrapingdog. The score uses the per-target p75 of successful attempts and assigns a penalty when no page returns, so a fast block does not count as speed.
String also leads on raw mean attempt time, which includes failures: 5.31 seconds against 20.32 for Scrapingdog.
Pricing: Scrapingdog is cheaper
Scrapingdog Standard is $90 per month for 1,000,000 prepaid credits and 50 concurrent requests. A rotating-proxy request costs 1 credit, JavaScript rendering costs 5, and a premium-proxy request costs 10. At full use:
| Scrapingdog configuration | Credits/request | Pages included | Price per 1,000 |
|---|---|---|---|
| Rotating proxy | 1 | 1,000,000 | $0.09 |
| JavaScript rendering | 5 | 200,000 | $0.45 |
| Premium proxy | 10 | 100,000 | $0.90 |
String Growth's $99 monthly fee is usage credit. Usage per 1,000 successful requests is $0.16 for a request path on a standard proxy, $0.90 on a premium proxy, $0.80 for a browser on a standard proxy, and $3.25 for a browser on a premium proxy.
Scrapingdog wins the price comparison. One hundred thousand premium-proxy successes fit in Standard for $90. The same count on String Growth's premium request path is $90.00 of usage, so the bill is the $99 fee. String's plain path is $16.00 of usage at 100,000 successes, also billed at $99 and still above Scrapingdog's $90 rotating-proxy bundle. Any of the fee you do not use rolls over.
Both products avoid charging for a block. Scrapingdog says failed, blocked, and timed-out requests are credited back. String bills only when verified page content returns. Scrapingdog's monthly credits reset on the billing date, while String credits roll over.
Price does not fill a target gap. At the benchmark's 45.6% result, Scrapingdog returned at least one page from 68 of 100 targets and nothing from 32. If your fixed list includes those misses, the cheap request is not a substitute for delivery. If your list sits inside its 22 perfect targets, Scrapingdog is the cheaper option.
Where Scrapingdog wins
- The cheapest published plain-fetch rate in the tracked field: $0.09 per 1,000 at Standard.
- Premium residential requests at $0.90 per 1,000 when the full Standard allowance is used.
- More than 40 domain-specific APIs with structured output, all available on every plan.
- Node.js and Python SDKs plus an MCP server that covers the full API catalog.
- A long public scale ladder from 200 free credits to more than one billion monthly credits.
Where String wins
- 97.0% measured success against 45.6%.
- 93 perfect targets against 22, and 2 shutouts against 32.
- No need to decide between plain, dynamic, or premium flags before each request.
- Verified-content billing and credits that roll over.
- A smaller gap between the documented product and the benchmark configuration: Scrapingdog's optional JavaScript renderer was off in this run.
Which should you pick?
Pick String when the dataset spans many protected domains and missing pages cost more than request volume. Pick Scrapingdog when your sites are in its coverage, price is the main constraint, or a dedicated Google, retail, social, or search endpoint saves parsing work. Test Scrapingdog with premium proxies and JavaScript together on pages that need rendering, and compare the body content rather than accepting HTTP 200 as success.
Switching from Scrapingdog to String
What carries over. A GET to api.scrapingdog.com/scrape becomes a POST to String's /v1/fetch with the key in a header. Keep the URL list and the code that reads HTML. formats=markdown maps to "format": "markdown", dynamic=true to "executeJS": true, and country to countryCode. premium=true has no String equivalent, because String picks the proxy class itself.
What you rebuild. The 40-plus dedicated APIs that return structured JSON for Google, Amazon, LinkedIn, Zillow, and the rest, plus webhook delivery. String's /search endpoint covers live Google results, jsonSchema replaces AI extraction for the fields you name, and /v1/fetch takes screenshots. Every other dedicated endpoint becomes a fetch plus your own parsing.
Integrations to reconnect. Scrapingdog publishes Node.js and Python SDKs, a local MCP package, an n8n node, and a Google Sheets add-on. String ships no language SDK, so an SDK call becomes one HTTP POST. Its hosted MCP server runs at mcp.usestring.ai, n8n calls String through the HTTP Request node, and Zapier, LangChain, and the Vercel AI SDK have setup guides in String's docs.
Test on your own failing URLs first. Test Scrapingdog with dynamic=true and premium=true together on the pages that need rendering, and compare body content rather than accepting an HTTP 200 as success. In this run Scrapingdog returned nothing on 32 of the 100 sites, and String returned 29 of those on every attempt. Move only the targets that improve. Leave custom headers off for the test: String does not bill a blocked request unless that request sent custom headers. The first 20,000 String standard fetches are free with no card. If you are weighing more than two tools, the ranking of 16 web scraping APIs uses the same benchmark.
FAQ
Is String better than Scrapingdog?
For measured coverage, yes: String passed 97.0% of requests against Scrapingdog's 45.6%, with 2 total shutouts against 32. Scrapingdog is cheaper and has more dedicated APIs, but it is no longer the faster product: its latency score fell from third of fifteen to last of sixteen between runs.
Is Scrapingdog cheaper than String?
Yes. Scrapingdog Standard is $90 for 1,000,000 credits, including 100,000 premium-proxy requests. String Growth is $99 a month, all of it usage, with premium requests at $0.90 per 1,000 successes. At 100,000 premium successes, that is $90 for Scrapingdog and $99 for String.
Was Scrapingdog tested with premium proxies?
The September run sends every provider through its strongest published configuration. The adapter routes through premium residential proxies. The 45.6% result is that path, not every possible combination of Scrapingdog's parameters.
Does Scrapingdog charge for failed requests?
No. Its pricing page says failed, blocked, and timed-out requests are refunded to the credit balance. String also does not bill a blocked request and requires verified page content.
Does Scrapingdog have an MCP server?
Yes. Scrapingdog publishes an MCP server plus official Node.js and Python SDKs. The MCP tools cover its generic scraper and dedicated API catalog. String also hosts an MCP server with fetch, search, and sitemap tools.
Which sites worked perfectly on Scrapingdog?
Twenty-two of the 100 targets passed all five attempts, among them Google, Kayak, Facebook, eBay, Airbnb, Booking.com, Nike, Delta and GitHub. String passed 21 of those 22 on every attempt and four of five on Foot Locker.
What are Scrapingdog's dedicated APIs?
They are structured endpoints for domains and data types such as Google Search, Maps, News, Shopping and AI Mode; Amazon products and reviews; LinkedIn profiles and jobs; YouTube; TikTok; X; Walmart; eBay; Zillow; and Yelp. Every plan includes the full catalog.
How can I verify these numbers?
Run the public benchmark repository with your own keys. Its Scrapingdog adapter is short and explicit. Try the published premium configuration first, then add JavaScript rendering for targets whose real content requires it.