No analysis is perfect. Here are a few situations where the thematic approach needs adjustment.

• Small sample sizes: If a product has only a handful of reviews, clusters won't form. Treat them as anecdotal and order a small test batch instead.
• Fake or incentivized reviews: Look for patterns like repetitive language across many reviews or an unusually high number of five-star reviews with no text. Those are red flags.
• Seasonal shifts: Flavor preferences change with the weather. A "summer fruit" liquid might get rave reviews in July but flop in December. Keep that in mind when you order.
• Device trends: As new devices hit the market, review patterns shift. A liquid that was a dud on old devices might shine on a new one, or vice versa.

When these edge cases pop up, fall back on your own testing. Order a sample, vape it yourself, and get a feel for how it behaves across devices. That's the final layer of verification.