How to Spot a Trend Before It's Obvious

Most trend-spotting advice is useless because it's post-hoc. The good books were written about trends that already won. The bad ones "predict the future" with no falsifiable mechanism. Neither helps you make a real bet on what to build today.

Real trend-spotting is mechanical. You read specific signals from specific sources at a specific cadence, and over time you can tell when a category is going from early-but-niche to about-to-explode. Below is the system we use to populate our trends database — and how you can use the same one yourself.

The four signals that actually predict adoption

Trends become trends when these signals fire together. Any one alone is noise. The combination is signal.

Signal 1: Search interest growing 3+ months in a row

Google Trends is free and underused. For any concept, pull the 5-year graph. The pattern you want to see: a flat baseline for years, then 3+ consecutive months of upward movement, with the latest month at least 2x the trailing 12-month average. That's a category waking up.

What you don't want: a single spike (probably PR or news cycle, not durable adoption) or a smooth gradual rise that's been happening for years (mature category, late). The sweet spot is recent acceleration over a sleepy baseline.

Signal 2: New tools shipping into the category in the last 90 days

Open Product Hunt. Filter by category. Sort by votes. Count how many tools related to your theme launched in the last 90 days. If that count has doubled vs. the prior 90, the category is heating up.

Twitter/X also works: search the category keyword and filter by recent. Count how many founders are publicly building in it. A category with 5+ new founders publicly building in the last 90 days is moving from "weird idea" to "land grab."

Signal 3: Funding announcements from credible investors

VCs front-run trends. They're wrong often but their bets are signal because they're followed by other VCs, which creates the funding wave that gives early companies fuel.

Watch Crunchbase, TechCrunch, and Twitter announcements for seed/Series A deals in your category. When 3+ companies in the same niche raise within 6 months, the category is about to get crowded — and validated.

Signal 4: Real users complaining about real workflows

The most underrated signal. Read Reddit, Discord, and niche subreddits in your theme. Look for posts that say "the existing tools for X are terrible" or "I wish there was a way to Y." If multiple unrelated people complain about the same gap within 30 days, the category has real demand that the existing tools aren't serving.

This signal is the one that separates a real trend from a hype trend. If founders are building but users aren't complaining, the category may have no actual demand yet — just supply.

The 12 sources we monitor weekly

1. Google Trends — search interest over time
2. Product Hunt — what shipped this week, by category
3. X/Twitter search — recent posts in target keywords
4. Crunchbase — seed/Series A funding in target categories
5. Reddit — top posts in niche subreddits + complaint threads
6. Indie Hackers — what founders are building publicly
7. GitHub trending — what dev tools are growing
8. Y Combinator request for startups — annual + interim updates
9. Hacker News — Show HN posts + comment threads
10. App Store + Play Store new releases — for mobile categories
11. Niche industry newsletters — 2–3 per category you care about
12. Conference programs — what topics get track headlines

You don't need to read all 12 every week. Pick 4–5 that cover your candidate category and read them on a fixed day. Treat it like a hobby. After 90 days, you'll have intuition no shortcut produces.

The pattern that signals "build now"

Specifically, you want this combination of signals firing together:

• Google Trends: recent 3-month upward trend, 2x+ over baseline
• Product Hunt: 3–5 launches in the category in the last 90 days (not yet 15+)
• Funding: 1–3 seed rounds announced in 6 months (not yet Series B+)
• Reddit/Discord: at least 5 unique user complaints in 30 days about existing options

When all four fire, the window is open. You have roughly 12 months to ship before the category gets crowded. That's your build window.

How to act on a trend you've spotted

Spotting is the easy part. Acting is harder. Three patterns work:

Pattern A: Build the on-ramp

New technology trends usually need a consumer-friendly wrapper before they go mainstream. Wrap the hard technology in a simple UI, sell to the people who want the outcome but don't want to learn the technology. AI image generation needed Photo AI. Web scraping needed ScrapingBee. LLMs needed TypingMind.

Pattern B: Build the picks and shovels

When everyone's building in a category, sell tools to the builders. The AI app boom created an explosion of demand for AI infrastructure, observability, evaluation, and ops tools. Builders are well-funded, sophisticated buyers — a great early market.

Pattern C: Build for a vertical the platform ignores

Horizontal platforms always leave vertical needs underserved. When a platform trend explodes, you can almost always build a vertical-specific version: AI for legal, AI for restaurants, AI for dentists. Smaller market but easier to win, and the customer knows exactly what they need.

Anti-patterns to avoid

Don't chase trends you don't personally care about

Trends are 18–36 month bets. If you don't already find the space interesting, you won't survive the boring middle. Conviction matters more than spotting.

Don't bet on trends without distribution

A great trend with no distribution channel is a great trend you can't monetize. Make sure your distribution hypothesis (SEO, community, integrations) is plausible before committing.

Don't bet on trends that require infrastructure that doesn't exist yet

Sometimes a trend looks ready but the underlying infrastructure is still 1–2 years out. Edge AI for consumer devices, for instance, has been "about to happen" for years. The model performance, hardware, and tooling all need to align. If your idea requires three of those to mature simultaneously, you're early.

The trend database we maintain at /trends tracks 45 candidate themes with growth rates, opportunity scores, and time-to-mainstream estimates — the output of running this system continuously. The best use of it isn't to read it once. It's to revisit it every month, watch which themes are accelerating, and let your idea hypotheses cluster around the strongest signals.