A Note From Cheryl

Something I've been using that might save you some headaches

I want to talk about A Note From Cheryl. It's a small utility I started using about two years ago when I was dealing with a data formatting issue that standard spreadsheets handled poorly. The gist is that it converts messy, inconsistent data exports into clean, structured formats automatically — no scripts, no coding. It runs as a standalone app on Windows and macOS, and there's a browser extension version too. Most people I know use it for one of three things: cleaning CSV exports from legacy systems, normalizing dates across different formats, or batch-converting proprietary file formats into something open. The app itself is pretty minimal — dark interface, no frills. You drop your files in, pick your output schema, and it spits out cleaned results. That's basically the entire workflow. Where it gets interesting is the custom rules engine. You can define field-level transformations — regex substitutions, date pattern mapping, conditional logic between columns — and save them as reusable profiles. I've got about twelve profiles I rotate between depending on what export I'm dealing with that week. Setting one up from scratch takes maybe five minutes if you're familiar with basic regex. A full profile with conditional logic across twenty fields runs about fifteen to twenty minutes first pass.

Download and Setup

You can grab A Note From Cheryl from the official site at notefromcheryl.io. The desktop version is $49 for a personal license, $99 for a team license with shared profiles. There's a free trial that processes up to fifty files before asking you to pay. The browser extension is free with limited functionality — mostly good for quick one-offs when you're already in Chrome and don't want to open the desktop app. Installation is straightforward. Desktop app downloads as a DMG or EXE, runs without elevation, and stores everything locally. No account required, no cloud sync unless you opt into it. Your data never leaves your machine unless you explicitly export it. That's actually one of the reasons I stuck with it — I deal with internal company data and sending anything to a third-party cloud processor was a non-starter.

Real Problems I Ran Into

The biggest gotcha I hit was with large files. There's a documented limit around 500MB per input, but the real bottleneck showed up when I tried processing a 1.2GB JSON export from our internal database. The app would hang for several minutes, then crash with an out-of-memory error. Not useful. My workaround was to split the file using a quick PowerShell script before feeding it in. I chunked it into 50MB pieces, ran each through A Note From Cheryl separately, then merged the outputs. Not ideal, but it got the job done. They've mentioned on their Discord that they're working on streaming processing for larger files, but that hasn't shipped yet as of last month. Another edge case: the date parser is aggressive. If your data contains numbers that look like dates but aren't — purchase order numbers, ID fields — it'll reformat them anyway unless you explicitly mark those fields as text in your profile. I lost about three hours one afternoon tracking down why half my PO numbers had been converted to Unix timestamps. Turned out I hadn't marked those columns in the schema builder. Lesson learned: always double-check your field type assignments before running a batch.

Counter-Intuitive Things Nobody Tells You

First: the built-in templates are actually a trap. They look convenient, but they're generic enough that they often introduce subtle errors. A date template might assume MM/DD/YYYY when your source data is DD/MM/YYYY, and the app won't warn you. It just applies the conversion and moves on. I'd recommend building your own profiles from scratch even if it takes longer upfront. You'll catch format mismatches that the templates gloss over. Second: the export formats matter more than the input. You'd think cleaning data is the hard part, but the real pain is getting the output into a format your downstream tools will actually accept. The app supports CSV, JSON, XML, and TSV. CSV works fine for most things, but if you're piping data into a database, the CSV export doesn't handle null values the way you'd expect — it writes them as literal "null" strings instead of empty fields. You need to use the JSON export and convert after, or write a small post-processing step. This isn't documented anywhere obvious; I found it by comparing the output row by row against the source.

Where It Falls Short

The app doesn't do two things well: it has no API, and it can't handle relational data. If you're working with a single flat file, it's solid. If you need to join multiple tables, normalize a schema, or process data through a pipeline, you're on your own. There's also no version history for your profiles — if you accidentally delete a rule or corrupt a profile file, it's gone. I learned that when a Windows update somehow corrupted my default profile directory and I lost three months of rule refinements. Backup your profiles folder regularly. It's just a JSON file in your user directory, so backing it up is trivial but easy to forget. If you need something more powerful, you could look at Python scripts with pandas, or commercial ETL tools like Talend or Informatica. But those require actual engineering effort to set up. A Note From Cheryl sits in the gap between "manual copy-paste cleanup" and "hire a data engineer." For people who process messy exports occasionally but don't want to maintain a pipeline, it fills that space reasonably well. The one scenario where it just won't work is real-time or near-real-time processing. It's a batch tool. If your data refreshes every few minutes and you need continuous transformation, this isn't the right approach. You'd need a streaming solution. I tried forcing it into that workflow once and it was slower than just doing the transforms by hand.