How to Cut Noise From Buying Signals
The visits and events that look like intent and are not, how to filter them out, and how to tell whether your alerts are earning their place.
5 min read
A signals program can fail without missing a single buyer. It sends more alerts than the team can act on, the team learns that most of them lead nowhere, and soon nobody opens them. The fix is to send fewer.
The lists below are our judgment, not the result of a study. Use them as a starting point and check them against your own results.
Signals That Look Like Intent and Are Not
Your own customers. People who already pay you visit the pricing page, read the docs, and log in every day. Unless you are looking for expansion, exclude them.
Your own team. Employees, contractors, and your agency. Filter out your own networks and logged-in staff.
Job candidates. A burst of visits to the homepage, the about page, and the careers page from one company is someone preparing for an interview.
Competitors. They read your pricing and changelog too.
Students and researchers. University networks, and anyone whose visit is one long read of a blog post.
Bots and security scanners. Automated tools open links in emails and crawl pages. A click on every link in an email within seconds of delivery is a scanner, not a reader.
Shared networks. A co-working space, a hotel, or a large internet provider can show up as one very active company.
A single page view. One visit to one page, with no return, is the weakest signal there is.
Broad topic surges. A report that a large company is reading more about a common subject. At a large company, someone is usually reading about most subjects.
Old news. A funding round from last quarter. Other sellers saw the same news when it broke.
Five Filters That Cut the Noise
- Limit alerts to a list. Decide which accounts you care about and only alert on those. Start here.
- Set a floor. Require a minimum amount of activity before something counts. A jump from very little activity to a little more is not a spike.
- One alert per account per day. A busy afternoon should produce one message.
- Require two signals. A pricing visit plus a return visit, or a fit match plus a usage jump. One of anything is a maybe.
- Add a cooldown. Once an account has been flagged and worked, do not flag it again for the same reason the next day.
Negative Signals Count Too
Some events should lower an account's priority. A trial that went quiet. A champion who left the company. A visit to the cancellation page. An unsubscribe. It is easy to build alerts that only add reasons to reach out. Keep a short list of reasons to stop.
Check Whether Your Alerts Are Working
Once a month, count three numbers: how many alerts were sent, how many someone acted on, and how many led to a real conversation. If most alerts are never acted on, you are sending too many, and the answer is a tighter filter, not a reminder to the team. There is no standard threshold. Pick one for your team and watch whether the ratio improves.
Then read the ignored alerts as a group. They usually share a cause: one noisy signal type, one segment that never buys, one page that is not as high-intent as it looked. Turn that one off.
How Leapio Fits
Leapio's signals are built around these filters. Most fire only for target accounts. Each signal is limited to one per account per day. A usage spike compares an account with its own recent activity and needs a minimum level before it fires. Each built-in signal can be paused, custom rules have their own cooldown, and Preview Matches shows which companies a rule would catch before you turn it on. See Signals and Target Accounts.