A buying signal is an observable event that raises the odds a company is considering a purchase right now. It does not say someone will buy. It says this is a good week to get in touch, instead of writing to the same company on a random Tuesday.
The difference between a team that works on signals and a team that works on sequences shows up in reply rates. The same message sent a week after a department head changes, and sent on a random Tuesday, is not the same message.
Your own signals. Events on your channels: website visits, especially pricing and product pages, email opens and clicks, LinkedIn engagement, downloads, demos watched, abandoned forms. These are the most precise of the three, because you know exactly who and when. They have one limit: they only cover companies that already found you.
Company signals. Publicly visible events: a new job opening for a role that deals with your problem, a change in the technology stack, a funding round, entering a new market, opening an office, a merger, a new CEO or department head. These need no tooling, only the time to watch for them.
Bought signals. Intent data from outside vendors, telling you someone at a given company is reading about your category online. The least precise of the three, because they usually tell you the company is interested, not who. Covered separately.
Most signal lists online are catalogues of everything measurable. That is useless, because a signal with no relationship to buying is noise that eats a rep’s attention.
Checking takes a day and works like this. Take your last twenty won deals and go back three months before each one was created. Write down what was happening at those companies. Then do the same for twenty companies that did not buy. The events that show up in the first group and not the second are your signals.
The result usually surprises people. In one case the best signal was not any on-site activity at all, but a job advert describing exactly the process the product automates. Nobody would have guessed that at a whiteboard.
Every signal also has a window. A new role is worth attention for a few weeks, not a year. If you do not know how long a signal lives, you do not know when to stop reacting to it.
| Signal | Source | Window |
|---|---|---|
| Hiring for a role that owns your problem | Job boards, careers page | 4 to 8 weeks |
| New head of the department you sell to | First 90 days in role | |
| Change in the technology stack | Technographic data, their site | 1 to 3 months |
| Third pricing-page visit in a week | Website visitor identification | A few days |
| Funding round | Trade press, registries | 1 to 2 quarters |
| Champion changed jobs | Immediately, both ways |
The last row is underrated because it cuts both ways. A champion who left takes knowledge of your product to a company that does not have it, and is the easiest conversation you will have all year. The company they left just lost the person pushing the topic internally, so the deal needs rescuing or dropping on purpose.
A spreadsheet and one hour a week. Seriously.
Columns: company, what happened, date, how I know, who responds, what they did. After two months you have data you cannot buy: you know which signals actually precede a conversation at your company and which are noise. Only then do you know what is worth automating.
Doing it the other way round, buying a tool first, ends with the tool producing hundreds of alerts, nobody knowing which to act on, and everyone muting it within a quarter. I have seen this several times and it always plays out the same way.
The first tool genuinely worth adding is website visitor identification, because it turns traffic you already pay for into company names. The rest can wait.
A signal is not a reason to send an offer. It is a reason to get in touch in a way that makes sense given what just happened.
The rule that rescues the most messages: write about the event, not about yourself. A new department head does not want to read about your product in week one. They want to know what companies in their situation do in the first three months.
Second rule: one signal, one message, not a five-step sequence. A sequence assumes silence means no interest. With signals, silence usually means you hit the wrong week.
Does a signal replace an ICP? No. A signal from a company outside your profile is a distraction. Fit first, signal second, never the other way round.
How many signals should I track? Three to five at the start. Each extra one lowers the quality of response, because nobody reviews thirty alerts a week.
Can I do this without visitor identification? Yes, on public signals. Hiring and leadership changes need no tooling, only regularity.
Who should respond, marketing or sales? Whoever holds the relationship. What matters more is that it is agreed in advance, because a signal both sides are waiting on expires.
Signals feed the heat score described in account scoring in ABM. Bought signals are covered in B2B intent data. The full model is in the guide to Account-Based Marketing for B2B SaaS.
Signals run in the other direction too. I checked what an AI agent sent to find a price sees across 26 Polish B2B software vendors, and where it gets stopped: the report on readiness for AI agents.
Daniel Tupczyński runs MOAT Marketing, an Account-Based Marketing partner for B2B SaaS companies. MOAT builds and runs ABM systems alongside the client’s own sales and marketing team, from market research and prospecting into Tier 1 accounts through to insights for product and strategy, so the know-how stays in the company. He also runs marketing at a B2B SaaS company.