8 min read
How to anonymize CSV files for ChatGPT
Anonymize CSV for ChatGPT in three steps: pseudonymize PII locally, paste the safe data, reverse the AI’s output with your mapping file.
Read: How to anonymize CSV files for ChatGPTBlog
Notes on comparing, splitting and cleaning messy spreadsheet data. No hype, no oversell. Just what we have learned shipping MessyMatch.
- 6 min read
Why Excel breaks at 1 million rows
Excel caps spreadsheets at 1,048,576 rows, and starts hurting long before that. Here is what that limit comes from and what to do once your data is past it.
Read: Why Excel breaks at 1 million rows - 7 min read
VLOOKUP vs XLOOKUP vs just diffing two files
VLOOKUP and XLOOKUP are great in a live spreadsheet. They are the wrong tool when you need to compare two files once and act on the result. Here is how to tell which approach you need.
Read: VLOOKUP vs XLOOKUP vs just diffing two files - 7 min read
Fuzzy matching explained for messy data
What fuzzy matching is, what Jaro-Winkler actually measures, when to use it and. More importantly. When not to. A practical guide for messy real-world data.
Read: Fuzzy matching explained for messy data
What you will find here
Short, practical posts about the work behind clean data. Most are written after we hit a problem ourselves and want a clear write-up that we wish had existed when we started. No clickbait, no thought leadership, no sponsored content.
- Practical guides on comparing, splitting, merging and anonymising data without code.
- Honest comparisons between spreadsheet workflows (VLOOKUP, XLOOKUP, Power Query) and dedicated tools.
- Notes on fuzzy matching, normalisation, encoding, leading zeros and other things that ruin diffs silently.
- Privacy-first walkthroughs: how to prepare a CSV for ChatGPT without leaking customer data.
- Each post lists the exact tier of MessyMatch that does the job, including the free path when there is one.
Topics we keep coming back to
If you read three posts here you will probably see one of the topics below. They are the questions our users keep asking and the ones we keep refining.
Comparing two messy files
Why naive byte diffs fail, how cleaning rules change the result, when fuzzy matching helps and when it lies.
Splitting and merging large data
How Excel breaks past a million rows, why SQL dump imports cap at 50 MB on shared hosting and what to do about both.
Anonymising data for ChatGPT and other LLMs
The minimum you must remove (names, emails, IDs, IBANs, DNIs), why reversible mapping matters and how GDPR pseudonymisation differs from anonymisation.
Fuzzy matching honestly
What Jaro-Winkler actually measures, when blocking saves you and when string distance gives you false confidence.
Spreadsheets in 2026
Where VLOOKUP still belongs, where XLOOKUP is genuinely better, and where neither belongs and a real diff is the answer.
Blog FAQ
How often do you publish?+
Roughly once or twice a month. We publish when we have something we would have wanted to read. No editorial calendar.
Is the blog written by humans?+
Yes. Every post is drafted, edited and proofed by a person. AI is sometimes used to check tone or catch typos, never to autogenerate the text.
Can I suggest a topic?+
Yes. Email hello@messymatch.com with the workflow that gave you the idea. We do not promise to cover it, but we keep a list and the best suggestions move up.
Do you accept guest posts?+
Not currently. We keep the voice and the technical depth consistent by writing everything in-house.
Is there an RSS feed?+
Not yet. For now, bookmark the blog index. The post list is short enough to scan.
Where do I report a mistake in a post?+
hello@messymatch.com. We update posts in place and add a short note at the bottom when the correction is material.