Sure, data’s the map. First layer: grab sales numbers for every bean type, roast level, and seasonal offering. Then run a trend filter—look for spikes that recur each quarter, see what flavors bump up when a particular weather pattern hits. Combine that with social chatter: scrape coffee blogs, Instagram captions, and sentiment scores for words like “creamy,” “bold,” or “refreshing.” Once you have the frequency and mood curves, feed them into a simple predictive model—maybe a moving‑average or a quick machine‑learning cluster. That tells you which flavor profiles are gaining traction and which latte art styles are resonating. Then test a pilot: a limited run of the predicted combo, measure response, adjust. Repeat, and you’ll stay a step ahead of the next craze.

Blueberry‑honey on dark roast could work, but it’s all about timing. Test a small batch, capture feedback, and compare the sugar‑cream balance to the current heat‑tuned profiles. If it spikes in the data, scale up; if not, tweak the ratio. Let me know when you have the samples ready, I’ll crunch the numbers.

Great, let me know the exact proportions you use. I'll set up the tracking sheet and we can evaluate the flavor profile against current trends as soon as the samples arrive.