Convert Tax Form PDFs to Google Sheets

Why tax season still breaks people

Every year, the same thing happens. A mid-size tax prep firm I know gets buried in October — not because they don't have enough staff, but because two people are sitting there typing K-1 box values into a spreadsheet for nine hours a day. K-1s from partnerships. 1099s from every bank and brokerage imaginable. W-2s, 1098s, the whole pile. Almost all of it arrives as PDFs. And before any actual tax work can happen, someone has to get that data out of the PDF and into something usable.

That's the bottleneck. Not the tax law. Not the filing. The data entry.

And honestly, it's a hard problem — because the formats are all over the place. A K-1 from one partnership looks nothing like a K-1 from another. Some PDFs are clean and machine-generated. Others are scanned at a gas station in February, slightly crooked, ink smudged on box 14. A 1099-B from Schwab won't look anything like one from Vanguard. Template-based tools collapse under that kind of variation. You end up maintaining a template library that's perpetually six forms behind.

How the extraction actually works

Lido takes a different approach. Instead of matching against a template, the AI reads the form the way a human would — it finds Box 1, Box 2, Box 14, and every other field based on context and layout, not a pre-built map. Drop in a K-1 prepared by a small regional accounting firm you've never seen before, and it still pulls the right values. That's the part that surprised me when I first tested it.

Extracted data goes straight into a Google Sheet. No copy-paste. No intermediate file.

For batch work, you upload everything at once. Lido processes each form independently and consolidates it all into one Sheet — one row per document, standardized columns across the board. That Sheet becomes your working file for the engagement. You're reviewing, not re-entering.

Which tax forms actually work well

K-1 (Form 1065 / 1120-S). These are the worst ones to deal with manually, which makes them the best use case here. Every preparer has their own layout. Lido's extraction engine doesn't care — it finds the boxes regardless. I've seen it handle K-1s from tiny LLCs and massive private equity funds in the same batch without issue.

1099s — all of them. 1099-INT, 1099-DIV, 1099-B, 1099-MISC, 1099-NEC, 1099-R. The variation across issuers is wild. Batch these and you get a single consolidated Sheet with everything normalized. It's kind of a relief, honestly.

W-2. Clean digital W-2s come out nearly perfect. Scanned ones — the kind clients photograph on their kitchen table — are a little less reliable on the small-font box values, but still dramatically faster than typing them by hand. Worth it.

1098 (mortgage interest). Simple form, few fields. Lender, interest paid, points, property tax. It just works.

How firms are actually using this

The workflow that seems to click for most people goes something like this.

Intake: Clients drop documents into a shared Google Drive folder, or email them to a dedicated address. Lido picks them up and starts processing automatically. No one has to manually kick anything off.

Review: Extracted data shows up in a client-specific Google Sheet. The preparer goes row by row, spot-checking values against the original PDF. Confidence scores surface the fields that the vision model wasn't certain about — so you're not reviewing everything, just the stuff that actually needs eyes on it.

Transfer: Export to CSV and import into Lacerte, ProConnect, Drake, UltraTax — whichever you're using. Or hit the API if you want to push data directly without the manual export step.

The time savings vary, but 60–80% reduction in document processing time is a reasonable estimate based on what firms have reported. On a 200-K-1 engagement, that's the difference between a week of someone's life and an afternoon of focused review. Those aren't abstract numbers — I've heard almost exactly that from a partner at a firm that processes private equity K-1s at scale.