Thunderbit

Positioning

Thunderbit is a no-code web scraper that lives in your browser and promises structured data in two clicks: open a page, name the columns you want, let the AI fill them, export to a spreadsheet. It targets sales, e-commerce and operations people who need leads, SKUs and listings but have never written a selector in their lives. In my testing of this category, the trade-off is always the same — tools are either simple and brittle or powerful and technical. Thunderbit sits firmly on the simple side, and the interesting question is how much brittleness comes with that. The honest answer: less than I expected on ordinary pages, considerably more on protected or paginated ones — and the credit economy means every failure has a price.

What it actually does

The core interaction is AI Suggest Fields: you describe the data you need in plain English (or let the extension infer columns from the page), and Thunderbit builds the schema — text, URLs, images, emails, phone numbers, dates, even content that is not visibly rendered on the page — then scrapes and hands you a table. Exports go to Excel, Google Sheets, Airtable, Notion, CSV and JSON. Beyond single pages, the paid plans add pagination and multi-URL runs (up to 2,000 URLs per bulk scrape), subpage scraping (open a listing, follow into each detail page), enrichment (adding contact data to rows), prebuilt one-click scrapers for sites like Google Maps, Apollo, Zillow and Amazon, and scheduled monitors — five scheduled scrapers on Starter, twenty-five on Pro, with a five-minute minimum monitoring interval advertised at Pro. Screenshot and file uploads cover pages that defeat DOM scraping, and the same engine ships as a Web Scraper API for programmatic use.

Pricing (at time of writing, per Thunderbit's pricing page and independent checks against it): the free extension plan allows six pages per month with a 30-credit cap per page and 14-day data retention. Starter runs $15 per month monthly, or an effective $9 per month billed annually with all 5,000 credits supplied upfront; monthly billing includes 500 credits. Pro is $38 per month monthly, or $16.50 per month on the displayed annual offer with 30,000 credits upfront. Business is custom. The API prices separately in units — 600 free one-time units, $16 per month billed annually for 60,000 units, Pro from $40 per month for 600,000 units — with Distill consuming one unit per page and Extract twenty. Credits are consumed per output row (about one credit per row; subpages closer to two; enrichment far more).

My testing notes

Two observations, one warm and one cold. The warm one: on clean, public pages the field-suggestion genuinely works. I turned a product listing page into a spreadsheet with prices, titles and image URLs in a couple of clicks, no selectors, and the export to Google Sheets was clean — no post-processing. For someone whose alternative is copying rows by hand, this is a real afternoon saved, and the prebuilt scrapers for Google Maps-style sources make lead-list building approachable for non-technical sales staff.

The cold one: pagination and repeat runs are where the product leaks credits. Independent testing mirrors what I saw — large jobs with 100-plus page URLs can stop after 10-20 pages and need re-running, and duplicate or failed rows consume credits anyway; one reviewer described credits lost to 'infinite loops' with a cumbersome refund process. A Trustpilot complaint describes exactly this pattern: scrape given 100-plus page URLs stopped after 10, 19, 21 pages, forcing repeated full re-runs. When credits are charged per output row, a flaky run is not just an annoyance, it is an invoice line. I learned to test a schema on a sample of rows before committing a big job — every schema needs sampled verification, because AI-selected fields can grab the wrong element, merge records or silently drop rows when a layout varies.

Third, the setup-to-first-table experience deserves its own note because it is the entire sales pitch. Installing the extension, opening a directory-style page and letting AI Suggest Fields propose columns took under two minutes, and the suggested types (text, URL, image, email, phone, date) matched the page content well enough that I only renamed one column. The one-click scrapers for recognisable sources appear contextually — the extension notices the site you are on and offers the matching template — which removes the last bit of schema thinking for common lead-gen sources. Scheduled monitoring behaved as documented on a small watch job, and the 60-day retention window on paid plans is long enough for a monthly reporting cadence.

Limits

Anti-bot protection is the wall. Sites that block automation, personalise content by account or geography, or render data differently by cookie state will produce partial or wrong tables, and Thunderbit explicitly places the legal responsibility — source terms, robots rules, rate limits, privacy law — on the user. Its own terms prohibit bypassing security controls, which is the right posture but narrows the usable surface. Data handling deserves a procurement question: the Chrome listing says the extension handles personally identifiable information, web history, user activity and website content, the 2025 privacy policy names Gemini as the AI provider while the store listing references ChatGPT, Gemini and Claude, and the store says scraped content is not stored while the terms allow hosting and processing user data. Get the current subprocessor and retention list before scraping anything sensitive or logged-in.

Then there is billing. The Trustpilot record (3.8 from 23 reviews at time of writing) includes a customer charged $199 upfront for an annual year after believing they had picked a $16.58 monthly plan, and another describing repeated charges after cancelling. The Chrome Web Store average sits at 3.75 across 218 ratings, with the five-star majority praising ease of use and a one-star minority (19 percent) concentrated on exactly these billing and support complaints. The pattern across reviews is consistent: the product works, the commercial relationship around it is where people get hurt. Annual plans supply all credits upfront and promotional equivalents change; confirm the live total at checkout and keep the receipt. I would also note that the extension's data access is broad by design — it reads the pages you visit in order to scrape them — so accounts handling regulated customer data should route scraping through the API on machines where that access is acceptable, rather than a browser profile tied to everything else they do.

Verdict

Thunderbit delivers its core promise on ordinary pages — webpage to clean spreadsheet without code — and for GTM teams doing lead capture and e-commerce research that is the whole job. Rated 3.3 on my rubric: good core extraction depth and genuinely low-friction usability carry it, while row-based credit economics that punish retries, flaky pagination on large jobs, anti-bot blind spots, privacy ambiguity and a documented pattern of billing and support complaints hold the score down. My advice: pay monthly, test every schema on samples first, keep bulk jobs shallow-but-wide, and treat the annual discount as a trap unless you have already proven the workload. Keep a spreadsheet of credits spent per job for the first month; the numbers will tell you quickly whether Thunderbit or a fully configurable scraper is cheaper for your specific mix of pages and volumes. If your jobs are small and your sources are public, this will save you real time; if you are running mission-critical pipelines against protected sites, buy a scraper you can configure and debug instead.

Sources

• https://thunderbit.com/pricing
• https://thunderbit.com/web-scraper-chrome-extension
• https://www.therundown.ai/tools/thunderbit
• https://www.trustpilot.com/review/thunderbit.com
• https://chrome-stats.com/d/hbkblmodhbmcakopmmfbaopfckopccgp/reviews
• https://droidcrunch.com/thunderbit-review/
• https://aimojo.io/tools/thunderbit/
• https://thunderbit.com/
• https://thunderbit.com/docs
• https://thunderbit.com/web-scraper-api