Intent data infers interest from research behaviour, such as content engagement and topic surges. Software signals record detectable changes in an account's technology environment, such as a new vendor footprint or a tool disappearing. Both start from observations. What separates them is how closely the observed activity relates to the buying decision you care about, and how much still has to be inferred.
Buying processes can leave observable software evidence: a new vendor footprint, a tool disappearing, or a change in the products an account uses. Those events can give sales teams a more specific starting point than a general increase in research activity. They still do not tell you what the buyer intends to do next.
Picture two accounts marked "high intent" in your CRM. People at the first have read three articles about support software. At the second, a competing support tool was first detected nine days ago.
Both observations can be dated. Neither proves a purchase is coming. But they support different conversations. One suggests interest in a topic. The other suggests a specific tool may be under evaluation or already being deployed.
This guide separates what was observed, what it reasonably suggests, what remains unknown and what to do next. It also explains how freshness and detection confidence change the reading. If the wider idea is new to you, start with Timing Intelligence 101.
Separate what you observed from what you inferred.
Every signal has two parts: the observation and the inference built on it. Mixing them up is how a reasonable signal turns into a wrong message.
Behavioural intent also starts from observations, such as research activity at an account. The inference is what that activity means for buying. One practical difference is that you can inspect first-party engagement on your own site, while a third-party topic surge is usually built from activity elsewhere that you cannot see. A software signal starts from a different observation, a detectable change in the account's technology environment. The inference there is what that change means for the decision.
Behavioural signals show research or engagement. Software signals show detectable changes in an account's technology environment. Neither category has a fixed level of buying relevance. A single article view and repeated pricing-page visits from known stakeholders are different signals. So are a newly detected free widget and a possible evaluation of a competing product. Judge each signal by what was observed, how closely it relates to your offer, and what remains unknown.
For the definitions in full, see Purchase Intent vs Intent Data. The practical question here is different: given a signal, what can you responsibly conclude?
What each signal can and cannot tell you
Every signal supports a limited conclusion. The skill is knowing where it stops. The table below sets out six common signal types. The first column separates what was observed from what was inferred or derived.
| Signal and basis | What it reasonably suggests | What it cannot establish | Appropriate next action |
|---|---|---|---|
| Behavioural intent Observed: topic surge or anonymous visits at the account. Inferred: that the topic relates to a buying decision. | Someone at the account is interested in the topic | Budget, a buying window, who is researching, or whether it is a buyer, a competitor or a student | Do not treat topic interest alone as proof of buying readiness. Assess the activity alongside account context, and use it to build a watchlist or nurture audience. |
| Repeat first-party engagement Observed: known stakeholders returning to your pricing or implementation pages. Inferred: that they are assessing your offer. | People at the account are looking closely at how your product would be bought or rolled out | Whether they are evaluating, comparing, validating a choice already made or researching for someone else, and how far along the decision is | Review promptly. If fit and context support it, reach out about the question they seem to be working on, without quoting the visits. |
| New software installation / possible trial Observed: a new vendor installation on the account's domain in the past 30 days. Inferred: a possible trial. | A relevant tool may be under evaluation or being deployed | That it is a contractual trial, who owns it, how it is going, or that it is being tested against you | Prioritise by ICP fit, then reach out within days with a problem-led message. |
| Software change Observed: a new vendor appears and a previous one is removed. Inferred: a replacement or consolidation. | The account may have replaced or consolidated tools | Why they switched, whether they are happy, whether it covers the whole company, or their procurement status | If they left you, investigate the loss and assess whether winback is viable, using the new vendor's estimated renewal as one planning input. If they left a rival, look for the adjacent need. |
| Renewal timing Derived: an estimated window, worked out from typical contract cycles for the vendor. Not observed directly. | A natural decision point may be approaching | The real contract end date, the term, the notice period or how satisfied they are | Share something useful early. Ask about timing in discovery. Never state the date as fact. |
| Churn signal Observed: a previously stable vendor footprint has recently disappeared. Inferred: the tool may have been removed. | A tool may have been switched off, so a need may be open | That it was cancelled rather than replaced or rebuilt, or whether a successor exists | Confirm with a second signal before outreach. For your own customers, start a check-in, not a pitch. |
Two wording choices are deliberate. A new installation is not a confirmed trial, and an estimated renewal window is not a contract date.
Behavioural intent
Behavioural intent describes attention, not action. The reader may be evaluating a purchase, writing a report or learning the topic for a new role. Treat it as a reason to pay attention, not as proof of buying readiness.
Repeat first-party engagement
Repeated visits to your pricing or implementation pages by people you can identify sit closer to your offer than a single article view, and you can inspect them yourself. They can still mislead: the visitor may be checking a decision already made or researching for someone else. Treat a pattern as a prompt to review the account, not a verdict.
A new software installation (possible trial)
A new vendor installation suggests that a tool in your category may be under evaluation or being deployed. Be careful with the word "trial". An installation is what was observed. Whether it is a formal trial, a quick test by one team or a rollout already agreed is something only a conversation will tell you. A newly detected free widget is also not the same as a competing product. For how these signals are found, see how to detect when a company starts a competitor trial.
A software change
A software change combines two events: something new appears and something old goes. That suggests the account may have replaced or consolidated tools, not that anyone made a formal decision. If the account moved away from you, investigate the loss before planning a winback, and treat the new vendor's estimated renewal as one input, as in B2B winback on competitor renewal timing. If it moved away from someone else, an adjacent need may be the better angle.
Renewal timing
Renewal timing is an estimate worked out from typical contract cycles, so it can be wrong in either direction. An account on a multi-year deal may look due when it is not, and a monthly plan has no meaningful window at all. Used well, it tells you when to start being useful, in the stretch before a renewal when switching costs are likely to be lowest. Used badly, it becomes a rep saying "I know your contract ends in March". The warm account play shows how renewal windows and competitor activity work together.
A churn signal
A churn signal suggests a tool may have been removed and a need may now be unmet. It is also the signal most likely to mislead. A footprint can disappear because a site was rebuilt, a tool was replaced, or the detection missed it on the day, so treat it as a prompt to look for a second signal, not a trigger. For your own customers, the right response is a check-in from customer success, not a sales message.
How freshness changes the reading
The same signal means different things depending on how old it is. Three things are easy to mix up: how often detection refreshes, how long a signal takes to reach you, and how old the underlying event is.
MarketSizer refreshes detection daily, and company lookups are re-checked live when you ask. New trial signals typically reach your CRM within about 5 days of the real-world event, which is an estimate, not a guarantee. Those are refresh frequency and delivery delay. Event age is the third, and it is the one you have to judge. A fresh lookup confirms the footprint that is detectable today. It does not confirm what the account currently intends to do.
MarketSizer uses a 30-day classification window for newly observed installations. This is a signal classification, not confirmation of the vendor's contractual trial period or the buyer's evaluation deadline. A signal that persists beyond 30 days is re-classified as an active subscription. First detected also means first observed by the detection system. It is not the exact installation date.
| Time since first detected | How to read it | What to do |
|---|---|---|
| Early in the window | The detection is recent. Evaluation or deployment may be under way. | If fit supports it, reach out quickly with a message about the problem, not the product. |
| Mid-window | Evaluation is plausible but not confirmed. The installation could be a test or a rollout. | Check whether the activity persists and whether other evidence supports outreach. |
| Late in the window | The signal is ageing. Reassess rather than assuming urgency. | Check for a second signal before investing time. Keep the message light. |
| Past the window | A persistent signal is re-classified as an active subscription, so it is no longer flagged as a trial. | Read it as a tool now in the account's stack. Look for other evidence before outreach, or move the account back to a watchlist. |
Put an age next to every signal in your CRM, and write down what age is too old for each type. Without that, a stale signal looks as urgent as a fresh one.
How detection confidence changes the reading
A signal is only as useful as your confidence that it is real. Observed evidence can be wrong. A footprint can be missed, misread or attributed to the wrong product, and that is true of every detection method.
MarketSizer grades each vendor signal for confidence, alongside when it was first detected and when it was last seen. Detection confidence measures how strongly the observed footprint is supported. It does not measure the probability that the account will buy from you. A strong grade can mean you are confident the tool is present. It cannot tell you the company is shopping, dissatisfied or ready for your offer.
A stronger grade is a reason to take the signal seriously. A weaker grade is better treated as a prompt to look for a second signal first. Freshness and confidence work together, and a reliable signal still has to earn its place against the account priorities your reps already have.
| Fresh signal | Ageing signal | |
|---|---|---|
| Stronger grade | Prioritise for review. Reach out when relevance and account context support it. | Review, and check whether a second signal confirms the account is still active. |
| Weaker grade | Watch closely. Look for a second signal before outreach. | Hold. Revisit only if something new appears. |
Let the signal shape the outreach. Be careful about quoting it back.
The signal should help decide who you contact, when you contact them and which problem you raise. Avoid presenting inferred or uncertain software activity as something you know about the prospect. Reference an event directly when it is public, relevant and accurately described.
Quoting an uncertain observation back at a prospect rarely lands the way it feels from the sender's side. "I noticed you started a trial last week" reads as being watched. It also commits the rep to a claim that may be wrong, because an installation is not a confirmed trial, and it hands the prospect a reason to wonder how you know. A public product launch, or an evaluation the prospect has told you about, is different. Those you can reference directly.
Compare two openings for the same signal.
Quoting it back: "Hi Sam, I saw you started testing a new tool for support last week. Want to see how we compare?"
Treating it as a hypothesis: "Hi Sam, a lot of support teams we speak to hit the same snag with reporting once ticket volume grows. I do not know if that applies to you, so I would rather ask: is it something you are dealing with, or is something else on your list this quarter?"
The second message makes no claim about what you saw, and it treats the problem as a guess rather than a fact. If you know something specific about the account, from a call, a job post or their own site, use that instead. This is the idea behind From Signal to Trust: a signal is not permission to sell. It is a reason to be useful at a moment when usefulness is more likely to be welcome. It is also why signal volume alone did not fix outreach, as covered in why intent data made B2B outreach worse.
Where behavioural intent still earns its place
None of this makes behavioural intent useless, or software signals a shortcut. Behavioural signals are a good filter and a good way to build audiences. How much weight they deserve depends on what was observed, and a single article view and repeated pricing-page visits are very different.
If a topic surge shows interest from accounts that already fit your ICP, put them in a nurture programme or on a watchlist, and look for account context before sales outreach. When a software signal appears on the same account later, you have both: an account that was warming up and a dated event that says something has changed.
What to do this week
You do not need a new process to start. Three small changes are enough.
- Pick one signal you can describe precisely. Write down what was observed, what you infer and what remains unknown, then route it to a named owner.
- Set a freshness rule. Decide how old a signal can be before it stops triggering outreach, and show its age wherever reps see it.
- Rewrite one template. Take one outreach message and check each line that mentions an observation. Keep it only if it is public, relevant and accurately described. Otherwise, express it as a hypothesis about the problem.
When you are ready to rank accounts across all of this, how to prioritise SDR outreach with purchase intent covers the full priority stack.
Frequently Asked Questions
What is the difference between buying signals and intent data?
Intent data infers interest from research behaviour, such as content engagement and topic surges. A buying signal is any observable event that may raise the likelihood of a purchase, such as a new tool appearing, a renewal approaching or repeat pricing-page visits. Intent data is one type of signal. Both start from observations, and both need interpreting.
Is a competitor trial a buying signal or intent data?
It can be both. A confirmed competitor trial provides evidence of evaluation. A newly detected installation is an observation that may suggest a trial, but does not confirm one. MarketSizer classifies a new installation as a possible trial for 30 days, and it still needs interpreting before you act.
Can a buying signal be wrong?
Yes. Detection can miss a tool, misread one or attribute it to the wrong product, and a real event can mean something different from what you assumed. That is why signals carry a date and a confidence grade, and why a weaker signal should be confirmed with a second one before outreach.
How fresh does a buying signal need to be to act on?
The fresher the better, and the right limit depends on the signal. A recent detection may deserve prompt review, but age alone does not establish whether an evaluation is under way. First detected is when the system first observed it, not the installation date. Set a maximum age for each signal type and show it to reps.
Should I mention the signal in my outreach?
Be careful. Use the signal to decide who to contact, when and about which problem, and avoid presenting uncertain software activity as something you know. Reference an event directly only when it is public, relevant and accurately described. Otherwise, write about the problem the signal implies, as a hypothesis.
Are renewal dates a reliable buying signal?
Treat them as estimates. A renewal window is worked out from typical contract cycles, not read from a contract, so it can be wrong. It is useful for deciding when to start being helpful. It is not something to quote to a prospect as fact.
Does a churn signal mean the customer cancelled?
Not necessarily. A churn signal means a previously stable footprint has recently disappeared. The tool may have been cancelled, replaced, rebuilt or simply missed by detection. Confirm with a second signal before acting, and for your own customers, start with a check-in rather than a pitch.
Do I still need intent data if I use buying signals?
It depends on the job. Intent data remains useful as a filter and for building audiences. Software signals can give a more specific starting point for outreach, but neither type is automatically more relevant. Many teams use both, and judge each signal by what was observed and how closely it relates to their offer.
Recommended reading
- Timing Intelligence 101 - the pillar guide to acting on buying windows.
- Purchase Intent vs Intent Data - the definitions and why content engagement is not a buying signal.
- How to Detect When a Company Starts a Competitor Trial - how trial signals are found and why the first weeks matter.
- The Winnable Account Is Warm, Not Hot - combining renewal windows and competitor activity.
- From Signal to Trust - why a signal is not permission to sell.
- Why Intent Data Made B2B Outreach Worse - what happens when inference drives outreach volume.
- How to Prioritise SDR Outreach with Purchase Intent - the priority stack that puts these signals to work.
- Purchase intent - how evaluation signals appear in MarketSizer.
- Our data - what MarketSizer observes and how it is graded.