Any research tool can hand you a list. The harder part is knowing which items on it are real, current and worth your time. That's why every question in OverHeard shows its provenance: where it came from, how often it recurs and when it was last seen.
Think of these three fields as the evidence behind the question. This guide covers how to read them.
Source: where was it asked?
OverHeard combines two kinds of source: public community discussions, like Reddit, and search-suggest behaviour. They tell you different things.
- Community discussions show questions people ask each other. They tend to be longer, more specific and more emotional: "our puppy won't stop biting when we pick him up, is this normal at 12 weeks?" These are gold for understanding the worry and the language.
- Search suggestions show what people type on their own. They're shorter and more direct: "puppy biting 12 weeks". They're a strong hint that the question has search demand.
A question that shows up in both is especially strong. People are discussing it and searching for it.
Recurrence: is it a pattern or a one-off?
One person asking something once is an anecdote. Recurrence counts how often the same question, including close rewordings, appears across sources. AI groups those rewordings together, so you see the underlying question once with its full count instead of five scattered versions.
High recurrence is the best single sign that a post will find readers. It also tells you the existing answers aren't working: if they were, people would stop asking.
Last seen: is it still being asked?
Questions have a life cycle. Some are evergreen: people will be asking how to stop a puppy biting for as long as there are puppies. Others spike around a change, like a new rule, a price rise or a product launch, then fade.
Last-seen tells you where a question is in that cycle. A question with high recurrence and a recent last-seen date is live. The same recurrence with a last-seen date from long ago may be a question that's already been answered well somewhere.
A quick way to decide
- Recent and recurring? Strong candidate. Write it.
- Recurring but not recent? Check whether it's evergreen. If yes, still worth writing. If it was tied to a moment that's passed, skip it.
- Recent but only seen once? Keep an eye on it. If you monitor the topic, you'll see if it starts recurring.
- Neither? Leave it. There are better questions on the board.
Read the original
The numbers are a filter, not a substitute for reading. Before you write, open the source and read the thread. You'll find the details that make a post genuinely useful: the follow-up questions, the answers people rejected and why, the words they use to describe the problem. That's what separates a post that answers the question from one that just mentions it.
Treat every question as evidence. Check where it came from before you build on it.
What OverHeard won't pretend
We show questions as-is, with their provenance, because we'd rather you judge the evidence than trust a black box. When measured search volume and difficulty aren't available, the numbers are clearly labelled as estimates. And if our live sources are ever unavailable, the page shows sample data that's labelled as illustrative, not measured. You should never have to wonder whether what you're looking at is real.