Most companies that fit your ICP cannot buy from you today. Not because they do not have the problem, and not because they do not have the money — because nothing has happened recently enough to make solving it this quarter’s work rather than next year’s.
The buying window is the stretch of time in which that is not true. It is the thing this whole system is trying to detect, and it is why the answer to “who should we contact?” changes every week even when your ICP does not.
The idea#
A company is capable of buying when four things are simultaneously true: the problem is felt, somebody owns it, budget can be found, and the timing is not blocked. Fit tells you the first is plausible. Only evidence tells you about the other three.
| Condition | What opens it | What the system looks for |
|---|---|---|
| The problem is felt | Scale, load or complexity crossed a threshold | Expansion, volume, headcount growth in the affected function |
| Somebody owns it | A person was hired or promoted into responsibility for it | Leadership appointments, first-in-function hires |
| Budget can be found | Money arrived, or a cost is being defended | Funding, contract awards, restructuring, published cost targets |
| Timing is not blocked | No incumbent contract, no freeze, no competing priority | Renewal dates, procurement notices, absence of distress signals |
How a window opens#
Windows open on events, not on dates. The engine treats a window as open when stacked evidence satisfies at least two of the four conditions above, from at least two signal families, inside an overlapping period.
One condition is not enough. Funding alone tells you money exists, not that anyone intends to spend it on this. A leadership hire alone tells you somebody owns the problem, not that they have anything to spend. It is the overlap that carries the claim.
- Two conditions, two families, one window. The default threshold for calling a window open.
- Three or more raises confidence, not just score. More conditions met means the argument survives one of them being wrong.
- A dated commitment overrides the count. A published deadline the company has to meet is sufficient on its own, because the deadline supplies timing, ownership and budget at once.
How long a window stays open#
Long enough to be worth acting on, short enough that acting late is the same as not acting. Window length is estimated per account from the evidence that opened it, not set as a global constant.
| Opened by | Typical usable length | Why |
|---|---|---|
| A leadership appointment | First 90 days of tenure | New owners change things early; after the first quarter the plan is set |
| A funding event | Roughly months 2–8 after announcement | Deployment lags the announcement; contacting on the day of the news is early |
| An expansion or new site | Around the operational date, either side | Pain is anticipated before it arrives and acute just after |
| A dated commitment or deadline | Until the date, weighted toward it | Urgency increases as the deadline approaches rather than decaying |
| An incumbent contract renewal | 90 days before the renewal date | Earlier is ignored; later is too late to run a process |
These estimates are visible on the account, and they are the reason an account can be qualified today and deliberately not contacted until next month. Being right about the company and wrong about the week is still being wrong.
How a window closes#
Windows close in four ways, and the system distinguishes between them because they lead to different next actions.
- It was filled. The company solved the problem, usually with somebody else. Evidence: a competitor announcement, a hire that implies the decision was made internally. The account is held with a revisit date at the likely renewal.
- It decayed. Nothing new arrived and the supporting evidence aged out. The account returns to the eligible pool and waits for a new trigger. No rejection is recorded, because nothing was learned.
- It was contradicted. New evidence undercuts the reason the window opened — a freeze, a restructuring, a withdrawn plan. The account is held and the contradicting record is attached.
- It was answered. Someone at the company told you. A reply classified as “not now, ask in Q3” closes the current window and schedules the next one, which is the most reliable window estimate available anywhere in the system.
Why this changes the question#
Volume software asks how to reach more of the market. If most of the market cannot buy today, reaching more of it faster mostly produces more people who are annoyed by you before they are ready to hear from you.
Which 20 companies should we contact today?
Twenty is not a magic number — it is the shape of the answer. A specific, dated, defensible set of companies, small enough that each one can be approached like it matters, with the evidence attached so the approach can say something true.
The biggest problem isn't volume. The biggest problem is targeting. Everything downstream of this page — scoring, tiering, campaign design, send pacing — is machinery for acting on that, and none of it improves a list of accounts whose windows are shut.
Next: tiering — how much effort an open window earns.