PDFTables sells generic table extraction in prepaid credit packs: $50 for 1,000 credits, one credit per page on the Standard extractor and two on the AI extractors, and unused credits expire after a year. If the PDFs you convert are bank and credit card statements, BankXLSX reads them as transactions instead of grid cells, with dates, descriptions, debits, credits and running balances in the right columns. Start free, no credit card.
Last updated August 2026
Upload your bank statement
Drop file here or click to upload
PDF, JPG, PNG, BMP, HEIC, TIFF, MT940
Uploading...
PDFTables.com is a UK cloud service from Cantabular Ltd that converts any PDF into XLSX, CSV, XML or HTML, sold as prepaid credit packs: $50 for 1,000 credits, $150 for 5,000 and $250 for 10,000. One credit covers one page on the Standard extractor and two credits per page on either AI extractor, and unused credits expire a year after purchase. It is a strong general table extractor and a weak bank statement reader, because it returns whatever grid it detects rather than a transaction schema. The main alternative for financial documents is a purpose-built converter. BankXLSX turns PDF and scanned statements into date, description, debit, credit and balance columns tuned to US bank layouts, gives you 15 pages free with no card, and meters pages monthly instead of selling credits that expire.
PDFTables does general table extraction well and has been doing it for years. The reasons buyers shop around are usually about shape rather than quality: how the credits are priced, and whether a grid extractor is the right tool for a financial document at all.
Any credits you buy are valid for one year from the purchase date, and whatever is left over is removed from your balance when that year is up. If your conversion volume is lumpy, you either overbuy and lose the remainder or underbuy and stop mid-project to buy another pack.
The Standard extractor is documented as unsuitable for documents containing only images, so a scanned statement has to go through an AI extractor at two credits per page. The page ceiling also drops from 4,000 to 250 and the upload limit from 500 MB to 50 MB on those extractors.
PDFTables reads the spacing between items to guess rows and columns. That works on a clean report table. A bank statement has wrapped descriptions, repeating page headers, and debit and credit conventions that vary by bank, none of which a spacing algorithm can reason about.
You get XLSX, CSV, XML or HTML. There is no QBO or other native accounting file, so anything headed for QuickBooks needs a second tool or a manual mapping step after the conversion.
Technical support is available by email or phone within normal UK working hours, 9am to 5pm GMT Monday to Friday. For a US team that is roughly 1am to 9am Pacific, so an afternoon problem waits until the following morning.
There is a free account and a free API key, but no free conversion allowance on the pricing page. The smallest way in is a $50 pack, which makes evaluating the output on your own documents a paid decision rather than a trial.
BankXLSX does one job and is priced for it: turning bank and credit card statements into accurate, reconciliation-ready spreadsheets, with a transaction schema instead of a raw grid.
Output arrives as date, description, debit, credit and balance columns, with wrapped descriptions folded back into their row and repeated page headers dropped.
Scanned PDFs and phone photos convert in the same pass, in PDF, JPG, PNG, BMP, HEIC and TIFF, without stepping up to a more expensive extractor.
A monthly page allowance that resets, instead of a prepaid balance that expires twelve months after you bought it.
Templates built for the date formats, debit and credit columns and running balances that US banks and card issuers actually print.
The Plus plan writes a QBO file directly, so statement data reaches QuickBooks without a mapping step in between.
Queue a whole client folder at once on the Plus plan rather than posting one document per API call.
No credit pack to buy first, no extractor parameter to choose.
Drop in a PDF or scanned bank statement. Multi-page and multi-month files work.
Tip: Password-protected PDFs are supported.
The AI reads every transaction into structured date, description, debit, credit and balance columns.
Tip: Scanned pages go through visual OCR automatically.
Export to Excel or CSV, then open it, model it, or import it into your accounting system.
Tip: Columns come out reconciliation-ready.
An honest side-by-side, including the documents PDFTables handles that a statement converter will not touch.
Close a month from PDF statements without cleaning up a misread grid first.
Build workpapers from multi-month statements where the balance has to tie.
Spread a borrower deposit history into consistent columns across several banks.
Get statement data into a model without a per-page credit budget to defend.
PDFTables.com is a cloud conversion service made by Cantabular Ltd, a UK data science company that also builds anonymization software used by the Office for National Statistics. You upload a PDF through the website or post it to a single HTTP endpoint, and it hands back XLSX, CSV, XML or HTML. There is no desktop app to install and nothing to configure. The company describes the method plainly: an algorithm examines the structures in the PDF and uses the spacing between items to work out where rows and columns start and stop, much the way your eye does when you scan a page.
That description is worth reading twice, because it explains both the strength and the ceiling. PDFTables is genuinely good at pulling a table off a page and handing you the same grid in a spreadsheet. It is used by hedge funds, private equity firms and logistics companies for exactly that. What it does not do is understand what the numbers mean. A bank statement is not one tidy table. It is a header block, an opening balance, a run of transactions that may wrap across pages, sometimes a separate deposits and withdrawals section, interest and fee lines, and a closing balance that has to tie. Hand that to a generic grid extractor and you get a grid, which is not the same thing as a transaction ledger.
PDFTables sells prepaid credit packs rather than a monthly subscription. There are three listed packs and a custom option. These figures were read directly off the PDFTables pricing page in August 2026.
| Pack | Price | Cost per credit | Cost per page, Standard extractor | Cost per page, AI extractor |
|---|---|---|---|---|
| 1,000 credits | $50 | $0.050 | $0.050 | $0.100 |
| 5,000 credits | $150 | $0.030 | $0.030 | $0.060 |
| 10,000 credits | $250 | $0.025 | $0.025 | $0.050 |
| Custom | Quoted on the pricing page | Varies | Varies | Varies |
Custom packs start at 20,000 credits, and PDFTables asks you to contact the team above 100,000. Payment is by Visa, MasterCard or American Express through Stripe, and invoicing is available on request. Every pack includes the same feature set: automatic table recognition, export to XLSX, CSV, XML and HTML, and full API access at no extra charge.
The per-page columns above are arithmetic on the published pack prices, not a separate rate card. They matter because the sticker price hides the real number. A firm converting 300 statement pages a month on the AI extractor is spending $30 at the 5,000 pack rate, or $600 a year, and burning 7,200 of the 5,000 credits it bought, which means buying again mid-year.
Not in any practical sense. PDFTables lets you create an account at no cost and issues an API key with it, and the API itself carries no separate charge, but conversions consume credits and credits are bought in packs starting at $50. There is no free monthly allowance advertised on the pricing page and no free tier alongside the paid packs. The company frames the API as free to use so long as you have an account and credits, which is a different thing from a free plan.
By page, multiplied by the credits-per-page rate of whichever extractor you choose. PDFTables compares it to photocopying. A ten-page PDF run through the Standard extractor costs ten credits. The same file through either AI extractor costs twenty. If a PDF has been laid out with several document pages printed onto one PDF page, you are charged for the PDF page, not the logical pages, which works in your favor. The three extractors differ in more than price.
| Extractor | API parameter | Credits per page | Max pages per PDF | Max upload size | What it is for |
|---|---|---|---|---|---|
| Standard | standard | 1 | 4,000 | 500 MB | Fastest and original. Not suitable for documents that contain only images, meaning scans with no text layer. |
| Intelligent #1 | ai-1 | 2 | 250 | 50 MB | Well structured documents and cleaner scans. |
| Intelligent #2 | ai-2 | 2 | 250 | 50 MB | Poor scans, confusing layouts and handwriting. |
Notice what happens to the ceilings when you step up to an AI extractor. The page cap drops from 4,000 to 250 and the file size cap drops from 500 MB to 50 MB. Both AI extractors can also return paragraphs alongside tables if you set the extract parameter, which the Standard extractor cannot.
Yes, through the AI extractors, and this is the point most comparison articles get wrong in one direction or the other. The Standard extractor is documented as unsuitable for documents containing only images, so a photographed or scanned statement with no text layer will not work there. Intelligent #1 handles cleaner scans and Intelligent #2 is described as handling poor scans and even handwriting. So scanned work is supported, at double the credit rate and inside the tighter 250-page and 50 MB limits. Budget accordingly if your clients still mail paper.
It is a legitimate commercial service with a named UK company behind it, and its published security posture is reasonable. Transfers are encrypted, the service runs on Amazon Web Services, the team will sign non-disclosure agreements on request, and uploaded data is deleted after 72 hours. That 72-hour window is also the redownload window, so if you close the browser tab you can return to the same URL and fetch the file again until it expires.
The considerations for a US buyer are practical rather than alarming. Your documents are processed by a UK-based provider, which is a question your compliance team may want to answer before bank statements for clients start leaving the country. Support is by email or phone within UK working hours, 9am to 5pm GMT Monday to Friday, which lands roughly between 1am and 9am Pacific. If something breaks at 3pm in Chicago you are waiting until the next morning in Cambridge.
These are not the same category of product, and pretending otherwise helps nobody. PDFTables is a general purpose table extractor with an excellent API. BankXLSX is a financial document converter. The table below is the quickest way to see which one your problem belongs to.
| What matters to you | BankXLSX | PDFTables |
|---|---|---|
| Core job | Read bank and credit card statements into a transaction schema | Extract any table from any PDF into a spreadsheet grid |
| Output columns | Date, description, debit, credit, balance, normalized per bank | Whatever grid the algorithm detects on the page |
| Multi-page statements and running balances | Stitched into one continuous ledger | Pages separated by blank rows or one sheet per page |
| Pricing model | Monthly subscription from $49, or $24 a month billed yearly | Prepaid credit packs from $50 |
| Do unused units expire | Monthly allowance resets each month | Credits expire one year after purchase |
| Free evaluation | 15 pages, up to 5 pages per extraction, no card | Free account and API key, but conversions need paid credits |
| Scanned pages | Visual OCR at the same page rate | AI extractors only, at 2 credits per page |
| Output formats | Excel XLSX and CSV, plus QBO on the Plus plan | XLSX, CSV, XML and HTML |
| Merchant cleanup and categorization | Included on the Plus plan | Not part of the product |
| Non-financial PDFs | Not the target, this is a statement converter | Yes, any document with tables |
| Support hours | US business hours | 9am to 5pm GMT, Monday to Friday |
| Best for | Bookkeepers, accountants and lenders working with statements | Developers and analysts extracting arbitrary tables at volume |
Almost always because the extractor found a different grid than the one you can see. PDFs carry no table structure, only characters at coordinates, so every tool has to guess where columns begin from the whitespace. Three things break that guess on bank statements in particular: a description long enough to wrap onto a second line, which the extractor reads as a new row with a blank amount; a debit and credit pair printed in one column with a minus sign or a trailing CR, which lands as text rather than a number; and page headers and footers repeating mid-file, which insert junk rows into the middle of the ledger.
A generic table extractor cannot fix any of those, because fixing them requires knowing that this is a bank statement and that every row must have a date and an amount. That is what a purpose-built converter adds. It applies a transaction schema, folds wrapped descriptions back into the row above, splits or signs the amount column consistently, and drops repeated headers. If your grid extraction keeps needing 20 minutes of spreadsheet cleanup per statement, that gap is the reason.
Be honest about this before you switch anything. If the PDFs you convert are not financial statements, PDFTables is the better tool and we are not a substitute. Research reports, published statistics, price lists, logistics manifests, scientific tables, government data releases: PDFTables handles all of them and we do not. Its API is also genuinely good, with documented examples in Python, PowerShell, PHP, C#, VBA, Java, R, C, Go and Node.js, a one-line curl call, and a balance endpoint so you can monitor credits programmatically. If you want to convert a PDF straight from an Excel macro, the VBA examples alone may decide it.
Volume shape matters too. Prepaid credits that last a year suit sporadic work better than a monthly subscription does. If you convert 400 pages in March, nothing in July, and 900 pages in November, a $50 pack sitting in your account is cheaper than eight months of a plan you are not using. Buy the credits, use them when the work arrives.
The converter wins when the documents are statements and the output has to be trusted by someone other than you. Bookkeepers closing a month, accountants building workpapers, lenders spreading borrower deposit history, and business owners reconstructing a year of activity all need the same thing: every transaction, correctly dated, with the amount on the right side of the ledger and a balance that ties. That is a schema problem, not a grid problem.
BankXLSX is built around US bank and card issuer layouts, so the date formats, debit and credit conventions and running balance styles that trip generic extractors are already handled. Scanned statements and phone photos go through visual OCR in the same pass rather than costing double. On the Plus plan you can queue up to 50 files at once, define your own extraction templates, get merchant names normalized and transactions categorized, and export a QBO file for QuickBooks. And the free demo gives you 15 pages with no credit card, so you can test the layout your own bank prints before spending anything. Start on the convert bank statement to Excel page for the Excel path, or the PDF bank statement to CSV converter if CSV is what your downstream system wants.
Plenty of finance teams end up with one of each, and that is a reasonable outcome rather than a failure to choose. Keep PDFTables wired into the pipeline that scrapes tables out of vendor reports and published data, where its API and its 4,000-page Standard extractor earn their keep. Send bank and card statements to a converter that knows what a transaction is. The two never compete for the same document, and neither one has to be stretched into a job it was not designed for.
Buyers evaluating PDFTables for statement work usually have several tabs open. If the others are statement-specific tools, the DocuClipper alternative comparison and the DocuClipper pricing breakdown cover the closest cloud rival, while the MoneyThumb alternative page covers the converter lenders and accountants weigh most often. For desktop software that runs entirely on your own machine with no upload at all, see the ProperConvert alternative and ProperSoft alternative pages. If you are comparing API-first document extraction rather than a web tool, the Amazon Textract alternative and LlamaParse alternative pages cover that end of the market, and the SaasAnt alternative page covers bulk import into QuickBooks. For the whole field ranked in one place, our roundup of the best bank statement converter software weighs accuracy, price and output formats side by side, and if you are also considering the free open source route, the Tabula alternative write-up explains why a 2018 release with no OCR usually costs more than it saves.
PDFTables sells prepaid credit packs rather than a subscription. As listed in August 2026, 1,000 credits cost $50, 5,000 credits cost $150 and 10,000 credits cost $250, with custom packs available from 20,000 credits. That works out to between $0.025 and $0.05 per page on the Standard extractor, and double that on either AI extractor.
No. You can create an account and get an API key at no cost, and the API carries no separate charge, but every conversion consumes credits and credits are only sold in packs starting at $50. There is no free tier or free monthly page allowance advertised on the pricing page, so evaluating the output on your own files is a paid step.
By PDF page, multiplied by the credits-per-page rate of the extractor you pick. The Standard extractor costs 1 credit per page, and both Intelligent extractors cost 2. If a PDF prints several document pages onto one PDF page, you are charged for the PDF page only. Unused credits expire one year after purchase.
Yes, but only through the AI extractors. The Standard extractor is documented as unsuitable for documents containing only images, meaning scans with no text layer. Intelligent #1 handles cleaner scans and Intelligent #2 handles poor scans and handwriting, both at 2 credits per page with a 250-page and 50 MB limit per file.
It is a legitimate service from Cantabular Ltd, a UK data science company. Transfers are encrypted, processing runs on Amazon Web Services, the team will sign NDAs, and uploaded files are deleted after 72 hours. The practical caveats for US buyers are that documents are processed in the UK and support only covers 9am to 5pm GMT.
Yes, and it is the strongest part of the product. It is a single multipart HTTP POST to one endpoint with your API key, so it works from any language, and PDFTables publishes examples for Python, PowerShell, PHP, C#, VBA, Java, R, C, Go and Node.js. A separate endpoint returns your remaining credit balance as an integer.
Four: Excel XLSX, CSV, XML and HTML. XLSX can be requested as a single sheet containing all pages or as one sheet per PDF page. CSV puts a blank row between pages. There is no QBO, QFX or other native accounting format, so QuickBooks-bound data needs a second step.
Because PDFs store characters at coordinates, not tables, so every extractor has to infer columns from whitespace. On bank statements the usual culprits are descriptions that wrap onto a second line, debit and credit values sharing one column, and page headers repeating mid-document. A converter that applies a transaction schema fixes all three; a generic grid extractor cannot.
It depends what is in the PDFs. For arbitrary tables in reports, statistics or price lists, PDFTables is hard to beat and a statement converter is not a substitute. For bank and credit card statements, a purpose-built converter such as BankXLSX fits better: it outputs a transaction schema rather than a grid, handles scans without a rate increase, and gives you 15 free pages to test your own bank first.
Turn PDF statements into clean Excel files.
Turn PDF statements into clean CSV files.
Compare BankXLSX with DocuClipper.
Bulk import into QuickBooks compared.
Cancel anytime from your account settings · refund policy
Get started converting bank statements to spreadsheets.
USD
per month
per month
$288 charged today
Choose speed vs accuracy when extracting
| Base AI Faster | 2,500 pages |
| Pro AI Best accuracy | 500 pages |
Scale statement conversion across your team with automation.
USD
per month
per month
$888 charged today
Choose speed vs accuracy when extracting
| Base AI Faster | 10,000 pages |
| Pro AI Best accuracy | 2,000 pages |
Enterprise-grade bank statement conversion and controls.
USD
per month
per month
$ charged today
Choose speed vs accuracy when extracting
| Base AI Faster | pages |
| Pro AI Best accuracy | pages |