❓ Problem

Google Trends is the closest thing the internet has to a shared referee. Someone claims a thing is blowing up, someone else pulls the chart, argument over. It serves around 5.9m visits a month doing exactly that job.

The catch is that Google Trends measures search, and search is downstream of discovery. People search for a term afterthey have heard it somewhere. Increasingly that somewhere is X, which still runs around 500m posts a day across 611m monthly users and remains the default venue for tech, finance, politics and culture forming an opinion in public. If you want to know when "vibe coding" or "GLP-1" or a new ticker actually entered the conversation, X is where it happened first and Google is where it showed up second.

And yet X gives you nothing. The trending panel is a live feed with no memory: hot at 2pm, gone by dinner, no chart, no history, no side-by-side, no export. It is a weather vane, not a barometer.

The data to fix this exists, it is just priced for Unilever. In February 2026 X moved new developers onto pay-per-use billing at roughly $0.005 per post read with a 2m read monthly ceiling, and the self-serve Pro tier that used to include historical search was closed to new signups. Full-archive search now sits behind an Enterprise contract from $42,000 a month. Everyone else gets a rolling seven-day window, which is to say they get vibes. So a Stripe product lead ends up asking on X whether Google Trends for Twitter exists, and nobody has a link to send him.

✅ Solution

A search box that turns any phrase, cashtag or handle into a mention-volume chart on X, going back years.

• The core view is deliberately Google Trends. Type a term, get a line. Add up to five for a side-by-side, set the date range, share the URL. Zero learning curve because the mental model already exists in everyone's head.
• Share of conversation, not raw counts. Absolute mention volume rises and falls with platform activity, which makes every chart lie a bit. Plotting each term as a share of total daily posts strips out platform drift and shows real growth.
• Semantic clustering instead of exact match. "Vibe coding", "vibecoding" and "I just let Claude write the whole thing" are one behaviour, not three keywords. Embedding-based clustering rolls them into a single line, which is the thing keyword tools have never done well.
• Slice it to check whether it is real. Filter by follower tier, sentiment and geography to see whether a spike is 400 founders in a group chat or a genuine population-level shift. This is the difference between a chart and an answer.
• Patient Zero is the hook. Surface the first accounts to use a term before it broke. It is the most shareable object in the whole product, it flatters the people who were early, and those people have large followings.
• Then breakout alerts and embeds. Alert me when any term in my watchlist crosses a growth threshold. Embed any chart anywhere, watermarked, because Google Trends screenshots are internet currency and every embed drags the logo into someone else's blog post.

The wedge is one chart for one term. The business underneath is a structured, queryable index of what X talked about and when.

📊 Key Numbers

Market size

• The social media listening market sits at roughly $12.15bn in 2026 and is forecast to hit $22.61bn by 2030, and 88% of agencies now call listening critical to their business. That is the TAM, and it is mostly enterprise seats.
• The realistic SAM is narrower and better: the prosumer trend-research slice. Exploding Topics proved people will pay $39 a month for trend charts, and it was profitable enough that Semrush bought it.