Every founder I talk to has a version of the same story. A deal goes quiet. Weeks later they find out the buyer went with a company they had never heard of. Not the big name they were worried about. Someone small, focused, and already selling to the exact people they wanted.
The problem was not that the information was hidden. It was public. A job post, a case study, a changelog, a talk at a regional conference. The problem was that nobody was looking for it, because looking for it is boring, and nobody had the time.
The dashboard nobody opens
The usual answer is a competitor dashboard. You list ten companies. You track their pricing pages, their blogs, their LinkedIn. You set up alerts.
I built one of these once, back when Waveloom was a workflow builder. Competitor price tracking, a check every hour. It worked. It also produced a stream of updates nobody wanted. A price moved by two percent. A product went out of stock and back. Every change was real, and almost none of it mattered.
That is the pattern with every alert product I have used. Week one, you read everything. Week two, you skim. Week three, you mute the channel. The alerts keep firing into a room with nobody in it. Then the one that mattered lands in the same pile and you miss it.
Alert floods fail because they answer the wrong question. They tell you what changed. You want to know whether the answer to a question you care about has changed.
Follow a question, not a company
Here is the shift. Instead of "track these ten competitors", ask something specific:
- Who is selling to compliance teams in DACH?
- Is anyone shipping approval-gated AI agents?
- Which agencies have started offering Linear migrations?
- Has anyone raised money to do what we do, but for hospitals?
A question has a bar built in. Most updates from most companies do not change the answer. A new hire at a competitor does not. A new customer logo in your exact segment does. A blog post about their culture does not. A pricing page that now lists a compliance tier does.
When you follow a question, most of what happens in the world is irrelevant by definition. That is the whole point.
What a good watch looks like
A watch, in my experience, needs three things.
A question. Not a keyword. "AI agents" is a keyword and it will bury you. "Is anyone shipping approval-gated AI agents for operators?" is a question, and a careful reader can tell whether a given article answers it.
A clear search focus. Name the market, the product or the category. The question is what gets searched, so "compliance software for DACH mid-market firms" finds more than "stuff in our space." One question per watch. A watch that tries to cover five topics matches none of them well.
A bar for what counts as a change. This is the part most tools skip. A watch should have an explicit idea of what would make it speak. A new company entering the segment. A named customer in your target market. A product that does the thing you thought only you did. Everything below that bar is silence.
If you write these three things down, you already have a better system than a dashboard. You could run it by hand every Friday: search, skim, decide whether anything moved. Most founders do not, because Fridays fill up. So I built it to run in the background.
How watches work in Waveloom
You describe the watch in plain language. On a schedule, Waveloom turns it into a public-news search, reads the handful of results that come back, and asks one question of each: does this directly and materially move the answer? Most runs, nothing does, and the run ends there.
When a result does qualify, it does not jump straight into your feed. It goes through the same path as an email or a Slack thread: the confidence bar, the hard cap on cards per batch and notifications per day, and the rule that a topic with a card gets a thread, not a second card. Weak or repeated findings are dropped, not queued for later.
What reaches you is one discovery card. It has a one-line "why this matters" written against your question and the evidence link, so you can check the claim before you act on it. You can open a thread on the card and ask for more, or dismiss it. Dismissals teach it where your bar really is.
It is worth being clear about what a watch is not. It searches public news. It does not subscribe to a list of pages, diff a competitor's pricing page, or catch every edit to a changelog. If there is a source you cannot afford to miss, like a vendor's status page or a regulator's mailing list, keep your direct subscription. A watch is for the questions you would otherwise never get around to asking.
A watch that says nothing is working
This is the part I want to be honest about. A well-formed watch can go a week without saying anything.
That feels wrong at first. You set something up, it stays quiet, and you wonder if it is running.
Each watch has a status, so you can tell a quiet run from a paused or failed one. If it ran and found nothing that cleared the bar, that is correct behaviour. The alternative is the dashboard you muted in week three. If the question is "who is selling to compliance teams in DACH?" and nobody new showed up in the news this week, the right number of cards is zero.
Quiet is not proof that nothing happened, though. Public-news coverage is incomplete, and a small company can go a long time without anyone writing about it. Treat a quiet week as "nothing cleared the bar," not "nothing exists."
Why this matters for a founder
You do not lose deals to competitors because you failed to read enough alerts. You lose them because the market moved and you found out last.
A watch is a cheap way to find out earlier. One question, a clear focus, an explicit bar, and a system that only speaks when the answer changes. The first time a discovery card shows you a company you had never heard of, selling to your exact buyer, you will understand why the quiet weeks were worth it.
Watches are the part I am tuning most actively with the beta cohort, especially the bar for what counts as a change. If there is a question you would like followed, tell me what it is when you join the waitlist.