Catch a topic while it is still cheap
By the time a term has search volume worth reporting, the cost of ranking for it has already gone up. Dorqa scores topics on how fast they are moving, and shows the five weighted parts the score is made of so you can see which one is carrying it.
Every number on this page came from a real run.
Every score on this page is shown with its parts. Browse the public trend pages to see the scoring on real topics without signing up.
The score, and the five parts it is made of
Most trend products publish a number. Dorqa publishes the number, the five components underneath it, and the weight applied to each, so you can see whether a score is being carried by developer activity, by research, or by search demand.
- Five components: developer, community, research, market, search
- Weights shown on the page: 35 / 30 / 20 / 10 / 5
- The weighted sum reproduces the headline score
- Rescored daily, because velocity is a rate of change
AI Agents scores 4.5. Research 5.4 and community 5.1 are pulling it up; developer 3.7 and market 3.2 are holding it down. 3.7×.35 + 5.1×.30 + 5.4×.20 + 3.2×.10 + 5.3×.05 = 4.49.
You can check that arithmetic yourself. That is the reason for showing it.
94
+2% on the prior seven daysFour of six sources, and the other two are named
A score built on four sources and a score built on six are different claims. Dorqa shows which of the sources relevant to a topic actually returned signals, and marks the ones that returned nothing instead of leaving them out.
- GitHub, Hacker News, arXiv and Product Hunt returned signals for AI Agents
- HuggingFace and Search Trends returned none, and are shown as empty
- 94 signals in the last seven days against 92 the week before
- The signal breakdown is per source, not a single blended total
Cost follows attention, with a lag
The window between a topic starting to move and the competition arriving is the only period where content is cheap to rank. Velocity scoring exists to find that window while it is still open.
- Publish into a rising topic rather than a saturated one
- Brief writers on direction, not just on volume
- Spot decay early enough to stop spending on it
- Feeds straight into the AI Visibility and SEO layers
- Trend EngineEvery scored topic, filtered and sorted
- Trend ExplorerThe full detail view for one topic
- Topic LookupScore any topic on demand
- Trending ProductsAmazon purchase demand, by market
- Trending StartupsOpen-role counts as an early indicator
- Meta TrendsClusters of related topics
- Tracked TrendsWatch a topic and see what changed
Purchase demand, and the part it cannot see
Demand comes from Amazon. The seller count, price band and retailer names come from Google Shopping, because Amazon only ever lists itself. Every card badges which source produced which figure.
- 800 products on this run, 124 with high demand and low competition
- Amazon reports demand in bands, so 100K means 100,000 or more, not exactly 100,000
- Google Shopping matched 622 of the 800 and names the retailers listing each one
- The other 178 show no seller data rather than a zero
The Ordinary Glycolic Acid 7% Exfoliating Toner sits at 100K bought/mo on Amazon at $14. Google Shopping puts the same product across 22 sellers in a $17 to $25 band, at Shein, Stylevana and Ulta Beauty.
Amazon is under the band the rest of the market is holding. Neither source shows that alone.
1 of 3 readings
14 open roles, no trend claimed yetHiring velocity, once there is enough of it to mean anything
Startups come from Y Combinator plus India accelerators and funding trackers. Open roles come from Google Jobs, read weekly. A growth reading appears only after three readings, and until then the card prints how many it has.
- Scaling means open roles rose 10 per cent or more against the previous week
- Cooling means they fell 10 per cent or more; steady is roughly flat
- 21 of 136 startups were scaling on this run
- The date the hiring data was read is printed, so you know its age
What this actually means
Trend velocity is the rate at which interest in a topic is changing, measured across several independent sources at once and rescored every day. It is a derivative, not a level: a topic with modest absolute volume that is accelerating hard scores higher than a large topic sitting flat. The published score is a weighted average of five separately scored components, and the weights are shown alongside it.
It is not search volume
Volume tells you how big a topic is today, which is also what every competitor bidding on it can see. Velocity tells you which direction it is moving, which is the part that decides whether you are early or late.
It is not a single number
The score is a weighted average of five components, and all five are shown with their weights. AI Agents scores 4.5 because research and community are strong while developer and market activity are not. A tool that hands you 4.5 and stops has told you less.
Coverage is not confidence
Six sources are relevant to a technology topic and four of them returned signals for AI Agents. HuggingFace and Search Trends returned nothing, and the page says so rather than quietly scoring across four and calling it complete.
One of six layers
They share one project, one subscription and one set of credits. See what Dorqa replaces.
Before you start
Six sources are treated as relevant to a technology topic: GitHub, Hacker News, arXiv, HuggingFace, Product Hunt and search trends. The page shows how many of them actually returned signals for the topic you are looking at, and names the ones that did not.
Daily. Velocity is a rate of change, so a score computed once and left alone stops meaning anything within a week.
It is a reason to look. Velocity says a topic is moving; it does not say the topic fits your client, or that you can win it. Read the score alongside its components and your own judgement about relevance.
Google Trends reports relative search interest from one source. Dorqa combines five separately scored components into one number, prints the weight applied to each, and rebuilds the score daily.
Amazon, for units bought in the last month. Amazon publishes that in broad bands rather than exact counts, so 100K means 100,000 or more, not exactly 100,000. The seller count, price band and retailer names come from Google Shopping instead, because Amazon only ever lists itself. On the run shown here Google Shopping matched 622 of 800 products; the other 178 show no seller data rather than a zero.
After three weekly readings. Open roles are read from Google Jobs once a week, and until a startup reaches three readings its card says how many it has rather than showing a trend. Scaling means open roles rose 10 per cent or more against the previous week, cooling means they fell 10 per cent or more, and steady is roughly flat.
Be first. Not just found.
Free tier, no card. Run your first check in a few minutes.