More and more fashion commentary accounts are incorporating search data and growth metrics into their content — which is fun to see!! But as this type of content becomes more common, so do the mistakes. So: I wanted to share a few quicky rules (or just thoughts) to help people interpret fashion data more accurately and avoid some common mistakes/misinterpretations.
Data is subjective. If you let a completely “objective” algorithm determine what is on trend and what is not — the results would be extremely basic and boring. The highest search volume items are always something like “white tee shirt” or “Levis jeans” (and who is going to report that). The interesting data points that get reported are selected and filtered through the interests and experiences of the selector.
Seasonality exists. Seasonal items will always show strong increases and decreases as the seasons change. This doesn’t mean that they are “trending” or “not trending”. You will need to look at the year-on-year growth (not weekly/monthly/quarterly growth!) to determine if the item will be more or less popular for the upcoming season.
10 to 100 is a 900% increase. It’s best to be wary of growth metrics if they are not reported or supported with absolute volume (if there are no sinners on this one than I am dead). For example, if a clothing item goes from 10 to 100 searches in a week, that’s technically a 900% increase — but the volume is still too low to indicate a meaningful trend or justify any real decisions based on that data point.
Cherry picking can prove anything. When analysing a specific trend, the number of possible search term variations is surprisingly high, and depending on which terms and associated metric you choose, you can (most of the time) choose a data point that either proves or disproves the trend you are researching.
Data can easily be misinterpreted. I find it difficult to explain this one without an example. I was recently looking into the western/country/americana aesthetic and noticed that most search terms were on the decline. However: terms which used specific pieces from this aesthetic were still on the rise (cowboy hat outfit, cowboy boots, fringe coat, etc). I came to realise that the decline was not because there was decreased interest in the aesthetic itself, but instead a decrease in people using the term “aesthetic” in their searches.
Sources change the meaning. The source of a datapoint can change how you interpret it. For example, I find that Pinterest is where people go to style items, where TikTok is where people go to be entertained. In line with this, skinny jeans have been trending on TikTok (because this content is “controversial” and entertaining) but not trending on Pinterest — indicating to me that people aren’t really wearing or styling them just yet, despite it trending on TikTok.
Data tells us what is already happening. By the time a search term or data point reaches significant volume, the trend is likely to be quite popular. Therefore, it is great at telling us what is already happening, and (most times) what will continue to happen — but it is not great for predicting or finding new trends.
Final thoughts: Search volume data from platforms like Pinterest or Google or TikTok can be incredibly useful when understanding the cultural/fashion landscape. That said, it should be treated more as a tool for observation rather than for prediction, and be considered one signal among many, alongside cultural shifts, runway trends, and (of course) what people are actually wearing in real life.
Note: I am not calling a specific person or account or whatever platform out and I have made some of these mistakes myself (if these were real sins, I’d be the first one in hell). Honestly, this article is as much a reminder to myself to be a good, honest data gal as it is anything else.
And that’s it for now! In Part 2, I’ll get into more detailed examples and offer practical solutions/methods to help you avoid these pitfalls. So subscribe if you’re interested in seeing that <3
I think this could also be titled "7 sins of data" and be just as valid!
I LOVE this post -- re: #2, whenever some major publication is like "plaid is in for autumn 2025!" I have a "florals for spring?" moment