If you're still cleaning and standardising spreadsheet data by hand, you're wasting time that AI-native tools can now reclaim. Normalisation and stemming, two techniques long buried in natural language processing, have finally become practical for everyday spreadsheet users, and that changes what "data preparation" actually means.
Here's what normalisation does in plain terms: it takes messy, inconsistent text and makes it uniform. Think of a column filled with "NYC," "New York City," "new york," and "ny city." A normalisation step recognises they all refer to the same place and standardises them to one form. Stemming goes further, stripping words down to their root so that "running," "runner," and "ran" all collapse to "run." Together, these two processes eliminate the grunt work of manual cleanup, no more find-and-replace marathons, no more agonising over whether "Jan" should match "January." The spreadsheet learns to interpret your data the way a human would, only faster and without the errors.
For most teams, this isn't a minor convenience; it's a fundamental shift in how much time they can spend on analysis versus preparation. Industry estimates consistently show that data professionals spend 60 to 80 percent of their time cleaning and organising data. Normalisation and stemming automate that portion directly. The practical result is that a user who once spent Monday morning reconciling name variants can instead spend that time asking questions of the data, spotting trends, building reports, or testing hypotheses. The tool stops being a bottleneck and becomes a collaborator.
That said, this capability only works if the implementation is transparent. Users shouldn't need to know the difference between a stemmer and a lemmatiser, or understand how tokenisation works under the hood. The best AI-native spreadsheets hide that complexity entirely. You type "normalise this column," and the system figures out the rest. That's the bar we hold for any genuine advance in this space: the technology should disappear into the outcome. If a product requires you to configure stemming rules or choose between algorithms, it's still asking you to be a programmer. The goal is to let you stay a data analyst.
Our take is straightforward: normalisation and stemming should be table stakes for any spreadsheet tool that claims to be AI-native. The technology is mature, the use case is proven, and the time savings are too large to ignore. If your current tool still forces you to clean data by hand, it's not you who is slow, it's the tool. The solution is to find one that treats data preparation as something the machine does, so you can focus on what the data actually says.