Evidence that a purchase is coming, months before the notice.
Budget lines, board approvals, hiring, market-research notices and expiring awards are all published, and all of them precede a solicitation. Signals collect that evidence against the buyer it belongs to and say plainly what it is — evidence, never a bid.
- Produces
- Signals — a separate object class
- Carries
- A source, a date and a confidence
- Never carries
- A deadline, because there is not one yet
By the time it is published, the requirement is already written.
A solicitation is the end of a process, not the beginning. Somebody decided there was a problem, found money for it, talked to vendors about how it might be solved, and wrote a requirement that reflects those conversations. The notice is the last four weeks of a two-year story.
A team that arrives at publication is bidding into a requirement shaped by whoever was in the room earlier. That is not a scandal; it is how procurement works. It does mean that the pursuits with the best odds are the ones where you were part of the market research, and those are decided long before anything is posted.
The evidence is public. Budgets are published, boards keep minutes, agencies post job adverts and forecasts, award registers show what is running out. It is scattered across sources nobody has time to read, and none of it looks like an opportunity, which is exactly why it gets ignored.
- Arriving as a strangerThe first conversation with a buyer happening at the pre-bid conference is a conversation the incumbent had a year ago.
- Capture planning without inputsA pipeline built only from published notices cannot plan more than a quarter ahead, because it only knows what is already open.
- Reacting to requirements you could have informedA market-research notice answered late is the one chance to influence a specification, spent.
From a scattered public record to a cluster worth a phone call.
Signals are gathered, attributed, clustered by buyer and category, and given a confidence with the reasoning attached. None of it becomes an opportunity until a solicitation exists.
Collect from sources that are not procurement portals
Published budgets and capital plans, board and council minutes, procurement forecasts, job postings, award registers and news coverage.
Individual signals, each carrying the document it came from and the date it was published.
Attribute to a buyer
A signal is tied to the organization it is about, and to the department where the record names one.
A buyer view that accumulates evidence instead of a stream that scrolls away.
Cluster by category
Signals naming the same buyer and the same category are grouped, and unrelated evidence is left ungrouped rather than forced together.
A cluster with a shape — what is being planned, by whom, and on what apparent trajectory.
Score confidence, and say why
Confidence comes from what kind of evidence it is and how much of it agrees; a published RFI weighs more than a job advert.
A confidence you can argue with, because the reasoning is stored next to it.
Watch for the solicitation
When a matching opportunity is published, the cluster attaches to it rather than being closed and forgotten.
A new opportunity that arrives with months of context already on the record.
Four signals, five months, one shaded window.
A funded budget line, a board decision to replace rather than extend, two job postings and a published request for information — clustered against one buyer. Each one says what it is not, because that is the sentence that keeps a signal honest.
Regional health system — telehealth platform replacement
Four signals from four sources, clustered because they name the same buyer and the same category. No solicitation has been published.
Oct – Dec 2026
Inferred, not announced.
- Budgetconfidence: Moderate
A telehealth platform replacement is funded for the next fiscal year, with an amount attached.
Why this is not an opportunity
A funded line is permission to buy, not a decision to compete. Programmes are re-scoped and money is moved.
- Governanceconfidence: High
The board approved a recommendation to replace rather than extend the current platform.
Why this is not an opportunity
An approval to replace names no vehicle and no date. It can still be satisfied by an existing cooperative award.
- Staffingconfidence: Moderate
Two implementation analyst roles posted, both naming the platform category in the description.
Why this is not an opportunity
Hiring says the work is being staffed. It does not say whether the work will be bought outside.
- Market researchconfidence: High
An RFI asking vendors to describe integration approaches, with a two-week response window.
Why this is not an opportunity
An RFI is market research. Answering it well is how you get to influence the requirement, and it is still not a solicitation.
What the product does with this
The window is inferred from the pattern of the evidence, not announced by the buyer. Nothing here has been published as a solicitation, and the cluster carries no deadline.
Nothing above is a live query or another customer’s pipeline. Every record in it was written for this page, and no buying organization named in it is real.
What keeps this from being intent-data theatre.
The failure mode of every early-warning product is confident nonsense. These are the properties that make the difference.
A separate object class
Signals live in the data model as their own kind of record. They never appear in the opportunity feed and never acquire a deadline they do not have.
The source, always
Every signal links to the document or posting it came from, with its publication date. A signal you cannot open is a claim.
Confidence with reasoning
A confidence level is stored with the sentence that justifies it, so a low one can be argued up and a high one can be challenged.
Clustering by buyer and category
Four weak signals about the same programme are worth more than four strong ones about four different things, and the cluster is what shows that.
Public sources only
Published records: budgets, minutes, forecasts, registers, job postings, news. No purchased contact data, no tracking pixels, no inferred browsing behaviour.
Attachment on publication
When the solicitation appears, the cluster is attached to the opportunity, so the history is on the record from the first day of the response.
Who plans far enough ahead for this to pay.
Signals are for teams whose pipeline has to extend past what is currently open.
A capture team with a long sales cycle
- Wins depend on being known before the requirement is written, and the current pipeline only shows what is already published.
- Clusters watched by category, reviewed monthly, with a named owner for each buyer worth cultivating.
- Capture planning has inputs beyond the open list, and conversations start while the requirement is still moving.
A firm entering a new region
- No relationships, no history, and no idea which buyers in the territory are about to spend on their category.
- Budget and governance signals across the region, filtered to the two categories the firm delivers.
- A short list of buyers with evidence behind it, instead of a cold-call list bought from somebody.
A specialist watching one programme type
- Sells into a category where three or four programmes a year matter and everything else is noise.
- Narrow clusters, high confidence threshold, and alerts only when a market-research notice appears.
- The two moments where influence is possible — the forecast and the RFI — are not the two that get missed.
Signals start at Professional.
Starter is built around what has been published: discovery, matching and response. Buying signals and the pre-solicitation view arrive with Professional, which is the plan for a team doing capture work rather than only responding to what is open.
- Professional and above include buying signals and recompete predictions.
- Starter includes discovery, matching and response, and does not include signals.
- Higher plans raise the listener allowance, which is what governs how many signal sources you can watch.
- Every plan keeps signals and opportunities as separate object classes; that is a property of the data model, not a plan feature.
Plan names and allowances are mirrored from the billing catalog. Amounts are on the pricing page, which reads them from Stripe rather than from a number typed into a marketing page.
What people ask about signals.
Is this the same as intent data?
No. Intent data usually means inferred behaviour bought from a third party — who visited which page, which company looked at what. Signals here are published records: a budget line, a set of board minutes, a procurement forecast, a job advert, an award register entry. Every one of them can be opened and read.
How accurate are the predicted windows?
A window is inferred from the pattern of the evidence and is presented as a range with the reasoning attached. There is no accuracy percentage printed here, because publishing one would require a study that has not been run — and a made-up figure is precisely the thing this feature is designed not to produce.
Can a signal turn into an opportunity automatically?
It cannot become one. When a solicitation matching the buyer and category is published, the cluster is attached to that opportunity, which is a different thing: the opportunity is a real published event, and the signals remain what they were.
Do you contact buyers on our behalf?
No. Nothing in this product contacts a buying organization. What it produces is evidence, an owner and a reason to make a call yourself.
What if the evidence is wrong or the programme is cancelled?
Clusters can be dismissed with a reason, and a dismissed cluster stays dismissed rather than reappearing next month. Programmes are cancelled all the time; a cluster that goes quiet stops being surfaced rather than being kept alive to look busy.
Look at what is already public about your buyers.
Pick three organizations you sell to and see what the published record says they are planning — the budget line, the minutes, the postings and the awards running out.