// Guides · Pipeline

Which deals are at risk this quarter, and how do you know before the forecast call?

By Kevin Bluett, co-founder of Ahoy · Updated September 2, 2026

A deal is at risk when the buyer's behavior has stopped matching the stage it sits in. You know before the forecast call by checking three signal families on every committed deal: activity (still replying, still meeting), relationship (more than one contact, and the decision maker among them), and timing (close date held, stage age inside the norm). A commit deal failing two of the three is not a commit.

What makes a deal "at risk" rather than just slow?

Slow and at risk are different things, and I think most teams run them together. A deal can take twice as long as usual and still be healthy, as long as the buyer keeps doing what a buyer at that stage does, which mostly means replying and agreeing to the next dated step.

Risk is a mismatch between the stage and the buyer. The deal sits in Negotiation, but only one person at the account has ever been on a thread. The close date is next Friday, and it was also next Friday three weeks ago. The rep says the decision maker is on board, but the decision maker has not written back since the second call, and nobody has asked why. A fast-moving deal with one contact and a close date that moved twice is at risk. A slow one with three active contacts and a paper process underway is not, however long it takes.

Forecast categories are where this shows up. I spent about ten years leading CRM engineering at HubSpot, and some of that time went into building forecast categories into the product, so I have watched a lot of teams fill them with hope. Commit is supposed to be a deal the rep will stake the number on: the buyer has said yes, the date is the buyer's date, and what remains is paperwork. Best case will probably close, but something real still has to happen first, a sign-off or a budget release. Pipeline is everything else with a chance. Risk is what moves a deal down a category, and this guide is about making that move before the call rather than during it.

What are the three signal families?

An at-risk deal leaves evidence in one of three places: what the buyer is doing, who at the account is doing it, and when things were supposed to happen. Each family catches something the other two miss, so I would resist the urge to pick a favorite.

Activity signals measure whether the buyer is still moving. Reply latency is the plainest one: a contact who answered inside a day now takes three, twice in a row. A meeting declined without a counter-proposal counts; a reschedule doesn't. Internal forwards matter too, because a deal that keeps gaining readers inside the account is alive and one that stopped gaining them has usually stalled, even when the original contact is still friendly. Activity moves first, and it is invisible if the CRM only records what the rep logged.

Relationship signals measure whether the deal can survive a bad week. Single-threaded means one contact and one point of failure. A decision maker absent from the last two threads means the person who signs has not been on the email or the call in two exchanges, whatever the rep believes about their support. A champion who went quiet has stopped pulling. This family catches what activity misses, because brisk replies from one enthusiastic contact can still be one reorganization away from dead.

Timing signals measure whether the plan is the buyer's plan. A close date that moved twice: one slip happens, two is a pattern. Stage age past the norm, meaning the deal has sat in this stage longer than deals that close from it usually do. No dated next step, where the last call ended with "I'll send it over" rather than a date. Paper process not started, with legal or procurement untouched and the close inside two weeks. A deal can be active and multi-threaded and still have no path to closing this period, and timing is the family that notices.

How do you score risk without a data science team?

You don't need a model. A prospect asked us this year, in those words, for "probability-adjusted estimates on future revenue," and what they needed in practice was a short list of signals, a weight on each, and a rule for what the total means. The weights below are a starting point, not the product of research; tune them against your own closed-lost history after a quarter.

Signal	Family	Weight
Close date moved twice	Timing	3
Single-threaded (one active contact at the account)	Relationship	3
Decision maker silent for 14 days	Relationship	3
No dated next step	Timing	2
Stage age past your norm	Timing	2
Reply latency doubled	Activity	2
Meeting declined and not rescheduled	Activity	2
Paper process not started inside 10 days of close	Timing	2
Buying group shrank	Relationship	1

The rule: a commit deal scoring 5 or more drops to best case, and a commit deal scoring 8 or more drops to pipeline. The thresholds are deliberately blunt. A single 3-point signal is a conversation with the rep rather than a downgrade; two together is a downgrade, because two independent families now disagree with the stage. Score from captured activity, not from the rep's account of it, and score every commit deal, not only the ones that feel shaky. When we go through a team's commit list during a free audit, the deal that surprises the manager is almost never the one they were already worried about.

What do you check on each commit deal before the call?

Five questions, answered from the email, the calendar, and the deal history rather than from the rep.

Has a decision maker replied in the last seven days? Not the champion, not the evaluator: the person whose name will be on the order.
Is there a dated next step that has not passed? A meeting on the calendar, a document due on a day. "Following up next week" is not a dated step.
Has the close date held since the last review? Pull the field history and count the moves this quarter.
Are at least two people at the account active on the thread? Active means they wrote or attended in the last two weeks, not that they were cc'd in June.
Has the paper process started? Legal, security review, procurement, whatever your buyers require. A commit deal closing in ten days with no redlines is a best case deal with an optimistic label.

Any "no" gets written down with the evidence, meaning the date and the thread. "I'm worried about Northwind" becomes "Northwind's decision maker last replied on August 19 and the close date has moved twice," a fact the rep can answer rather than a feeling they can argue with.

Here is a worked example with illustrative numbers. A team goes into the forecast call with ten commit deals worth $420,000, and the manager scores each one the day before.

Six deals score 0 to 3: the decision maker is on a recent thread, there is a dated next step, and the close date has held. They stay in commit at $220,000. Three deals score 6, each single-threaded (3) with a close date that has moved twice (3). Their buyers are still replying, so the reps believe in them, but one contact and a slipping date is a pattern this team has lost to before, and they move to best case at $140,000. One deal scores 8: decision maker silent for 21 days (3), close date moved twice (3), no dated next step (2). The champion is still writing warm emails. It moves to pipeline at $60,000.

Commit goes into the call at $220,000 instead of $420,000, with a written reason on every downgrade. That is uncomfortable for about a minute, and in my experience it is far less uncomfortable than reporting $420,000, closing $230,000, and explaining the gap in October. The smaller number is one the company can actually plan around.

What is deal slippage, and how do you measure it?

Deal slippage is a deal whose close date moves out of the period without the deal being lost. It is the most common way a forecast misses, because the buyer still intends to buy and the number still did not arrive when it was promised. A prospect put the problem to us this year as "deal delays and attrition," which is about as plain a name for it as I have heard.

Measure it as slip rate, the share of deals committed at the start of a period whose close date moved past the end of it. Twenty deals committed and six still open on the first day of the next quarter is a 30 percent slip rate. Track it by rep. A rep with a high slip rate is usually not lying; they are setting close dates from hope rather than from the buyer's plan.

The leading indicator is "moved twice." One move is usually a correction. Two means the date was never anchored to anything and was probably the date the rep needed rather than the date the buyer had, so it carries the highest weight in the table above.

The fix is upstream of the forecast. Set close dates from the buyer's stated compelling event, like the renewal they are replacing or the budget that expires. If the buyer cannot name one, the deal has a guess rather than a close date, and it does not belong in commit. In my experience "what happens on your side if this is not done by month end" gets a better close date than "does the 30th work for you."

How do you present it on the forecast call?

Three lines: commit, best case, and pipeline, each with a dollar figure, the deals in it, and the reason for any change since last week. A downgrade is stated with the signal that caused it, so "Northwind moved to best case; single-threaded, and the close date moved a second time on Monday," with no narration and no "I still feel good about it." If the rep disagrees, they are disagreeing with the evidence, and that is a useful argument to have.

The rest of the call goes to decisions on the at-risk deals, since those are the only ones where the call can change the outcome. Each downgraded deal gets one of three actions, with an owner and a date. Open a second thread, where the champion introduces you to the decision maker or a peer. Ask the closing question: is this still happening this quarter, and if not, when? Or park it outside the forecast and revisit next month. Commit and pipeline need no discussion.

What to look for in a tool

A spreadsheet will run the method above, and I would start there. It will not keep running once the pipeline passes thirty deals, because every signal has to be pulled by hand from a system of record that only knows what the rep typed. For anyone evaluating tools for this, these are the criteria I would use:

Captures buyer activity automatically, from email and calendar, so the signals are real rather than rep-reported.
Shows days since the decision maker's last reply and days in stage on every deal, without a report being built.
Records close-date history, so "moved twice" is visible on the deal rather than buried in a field log.
Flags drift against your own pipeline's norms, not a generic score trained on someone else's data.
Drafts the recovery action, whether a second-thread introduction or the closing question, and holds it for approval with a human in the loop rather than sending on its own.

The last point separates a tool built to remember from one built to act, with guardrails. Ahoy runs this check all week rather than the night before the call: it captures the buyer's activity itself, flags deal drift on each deal as it happens, and drafts the recovery action for approval, so the downgrade list and its evidence exist before anyone opens the forecast. If you would rather see the method run against your own pipeline first, the free CRM audit is a 45-minute walk through your current board, covering deals with no dated next step, deals past their stage age, and how much of the buyer's activity is actually on the record. Nothing to install.

Frequently asked questions

What is an at-risk deal?

An at-risk deal is an open opportunity whose buyer behavior no longer matches its pipeline stage. The usual signs are a decision maker who has stopped replying, a single contact carrying the whole deal, a close date that has moved more than once, or a next step with no date on it. Risk is measured against the stage, not against speed: a slow deal with an engaged buyer is not at risk.

What is deal slippage?

Deal slippage is when a deal's close date moves out of the forecast period without the deal being lost. It usually means the close date was the rep's hope rather than the buyer's plan. Measure it as slip rate, the share of deals committed at the start of a period that were still open at the end of it, and treat a close date that has moved twice as the leading indicator.

How do you calculate forecast accuracy?

Compare the forecast at the start of the period with what actually closed. Divide closed revenue by the committed forecast and express it as a percentage; a $220,000 commit that closes $200,000 is about 91 percent accurate. Track it per period and per rep, and watch the direction rather than chasing a fixed target, because the right number depends on your motion and your deal sizes.

What is a good pipeline coverage ratio?

Three times quota is the common rule of thumb: for every $1 of quota, hold $3 of open pipeline. The rule assumes you win about a third of what you pursue; the underlying math is one divided by your win rate, so a team closing 20 percent needs five times and a team closing 40 percent needs two and a half. It is a rough guide, not a law, and it is meaningless if the deals inside it are at risk. Ten deals of coverage with single-threaded contacts and slipping close dates cover nothing. Check the quality of the deals before trusting the multiple.

How do you identify at-risk deals before they slip?

Check every committed deal weekly against three families of signal: activity (is the buyer replying and meeting), relationship (are two or more people at the account active, including the decision maker), and timing (has the close date held, is the deal inside its stage norm, has the paper process started). Score the signals, downgrade any commit deal that fails two families, and write down the evidence before the forecast call rather than during it.

Book a Demo Get a Free CRM Audit