BasisFile guide

Fuzzy Match Customer Names Across Spreadsheets — A 30-Second Guide

You're trying to combine two lists. One has 'Microsoft Inc.', the other has 'Microsoft, Inc'. To Excel, those are different companies. To you, they're obviously the same. Doing this manually for 500 rows is a half-day of work — and you'll still miss some.

How BasisFile fixes this in 30 seconds

  1. 1

    Upload your two files

    They share a name, email, or company column you want to match on. CSV or Excel both work.

  2. 2

    AI suggests the join key

    It picks the column that looks like the natural match key, then matches values like 'IBM', 'I.B.M.', and 'International Business Machines' as the same entity.

  3. 3

    Review, then download

    Each match shows a confidence score. Accept everything, or drill into edge cases. Download the merged file as CSV or Excel.

Before vs after

Three rows in each list — actually one company.

Two lists

ListCompanyRevenue
CRM exportAcme Corp£42,000
Finance exportAcme Corporation£42,000
Marketing exportACME CORP.£42,000
CRM exportIBM£128,000
Finance exportI.B.M.£128,000

One matched entity

Company (canonical)SourcesRevenue
Acme CorporationCRM, Finance, Marketing£42,000
IBMCRM, Finance£128,000

Most users go straight to Pro

The free tier is enough for occasional cleanup. If you do this weekly, Pro pays for itself the first Monday.

Most Popular

Pro

£15/mo
  • Unlimited file cleans
  • Up to 100,000 rows per file
  • No watermarks on output
  • Save merge & cleaning templates
  • Cancel anytime
Try free — no signup requiredSee all plans →

Frequently asked questions

How does the AI decide what's a match vs a different entity?

It combines string similarity (edit distance, token overlap) with semantic understanding from a language model. So 'Microsoft Inc.' and 'Microsoft, Inc' are matched on string similarity; 'IBM' and 'International Business Machines' are matched semantically.

Can I tune the strictness of the fuzzy match?

Yes. Pick strict (only obvious matches), balanced (default), or loose (suggests more, you accept individually). Confidence scores are shown for every match so you can review borderline cases.

Does it work for emails, addresses, and phone numbers too?

Yes. Emails normalise case and dotted Gmail aliases. Addresses match across formatting variations. Phone numbers match across country code and punctuation differences.

What languages does it support?

Any. The matching uses a multilingual model, so Japanese, German, Arabic, and right-to-left scripts all work the same way.

Will my data leave my browser?

Files are uploaded encrypted, processed in memory, and deleted within 24 hours. We never train models on your data and never share it with third parties.

Stop wasting Mondays on data hygiene

Drop your file, get a clean one back in 30 seconds. No signup needed for the free tier.

Try free — no signup required