Section 1
Key takeaways
• A clean pipeline produces forecasts that are 23% more accurate than pipelines with hygiene problems, so hygiene is a forecasting tool, not an administrative chore. • The two signals worth tracking are buyer-verified facts: a dated next step and a close-date slip count. Both survive scrutiny because the buyer, not the rep, put them there. • Reps who failed to define a next step saw a 71% decline in close rates, the dated next step is the single highest-signal field in your pipeline. • Nearly 60% of forecasted B2B deals slip to the next quarter , so counting slips (not staring at percentages) tells you which deals are dying while there's still time to act. • Only 45% of sales organizations say their leaders have high confidence in their forecast . The fix is deleting opinion fields, not adding data-entry.
Section 2
Why more tracking makes the forecast worse, not better
Start with the thing nobody says out loud: a CRM (the software you store deals in, Customer Relationship Management) does not know anything. It holds whatever a human typed into it. So every field you add is a new opportunity for a busy person to type something hopeful, vague, or simply wrong. Picture a 12-person digital agency. One co-founder sells. The pipeline has eight stages, a "deal temperature" field, a probability percentage, and a "next action" free-text box that usually says "follow up." On Friday the founder looks at $240,000 of "weighted pipeline" and feels good. Three weeks later, two of the four big deals have gone silent and the quarter misses by 40%. Nothing in that CRM was a lie, exactly. But almost none of it was a fact. "Warm," "75%," and "follow up" are all the seller's mood, not the buyer's behavior. That is the trap: when most of the average pipeline is stale and less than half the data is accurate, "track more" really means "record more moods." More moods do not make a number you can bank on. This is why hygiene pays. InsightSquared found a clean pipeline produces forecasts that are 23% more accurate than pipelines with hygiene problems . The lift comes from deleting entries that were never information in the first place. Qualifying what gets into the pipeline is its own subject, the discipline of qualifying out before a deal ever enters, but even a well-qualified deal rots if you measure it with opinion fields.
Section 3
The hygiene test: did the buyer do it, or did the seller feel it?
Here is the one rule that replaces your entire "what to track" debate. For every field in your pipeline, ask: is this something the buyer actually did or agreed to, or is it something the seller thinks? Buyer-verified facts are hygiene. Seller opinions are theater. Sort every field into those two buckets and the answer to "what do I track" falls out by itself. Buyer-verified facts: • The prospect agreed to a specific meeting on a specific date. • The prospect sent the contract to legal / asked for a security review / requested references. • The prospect last replied, opened the proposal, or took any action on a given date. • The close date the buyer themselves named, and how many times it has moved. Seller opinions (theater): • "Probability: 70%." A number the rep invented. • "Stage: Negotiation." A label with no exit criteria the buyer ever crossed. • "Deal temperature: Hot." A feeling. • "Activities this week: 41 dials, 18 emails." Effort, not progress. The activity counts deserve a special mention because they look like data. But a dial is something your team did; it tells you nothing about whether the deal is alive. Counting outbound effort to gauge deal health is like weighing yourself by counting the times you opened the fridge. If you want to fix the demand side, fix the quality of the conversations, which is a discovery and qualification problem we work through in the handful of things every discovery call has to surface, not the count of attempts.
Section 4
What to track: the dated next step
The highest-signal field you will ever have is also the simplest: does this deal have a specific, mutually-agreed next step, on the calendar, with a date? Not "follow up next week." Not "circle back." A scheduled event both sides have agreed to: "Thursday 2pm, demo with their ops lead and CFO." A date you can point at. Gong's research, cited by Storylane, found that salespeople who skipped defining the next step saw a 71% decline in close rates . A deal without a defined next step is, statistically, already most of the way to dead, no matter how "hot" it feels. The sales author Jeb Blount, founder of Sales Gravy, puts the rule about as bluntly as it can be put: "Never, ever, ever, write this down, ever, ever, ever leave a meeting with a prospect if you haven't defined and set a next step." Translate that into a hygiene rule a founder can actually enforce: a deal with no future-dated, buyer-agreed next step does not belong in the forecast. It can sit in a "no next step" holding column where it is visible and uncomfortable, but it does not get counted in the number you commit to. This single rule does more for forecast honesty than any eight-stage funnel: it forces every "warm" deal to either produce a real calendar commitment or admit it has none. For the agency above, applying this rule cut their "real" pipeline from $240,000 to about $90,000 overnight. Then they realized the $90,000 was bankable and the missing $150,000 had always been fiction, they had just been forecasting from it.
Section 5
What to track: the close-date slip count
The second signal is the one almost nobody counts: how many times has a deal's close date moved? CSO Insights data, cited by Forecastio, shows nearly 60% of forecasted B2B deals slip to the next quarter . Slip is the normal failure mode of a forecast, not a dramatic "Closed Lost," just a quiet push, push, push until the deal evaporates. You cannot prevent the first slip. But you can count slips, and the count is an early-warning system. The rule of thumb: zero or one slip is normal. Two or more is a flag. A deal whose close date has been pushed twice is not "still in play." It is telling you, in the only language a CRM has, that something is structurally wrong, no real budget, no real authority, no real urgency, that no amount of optimism will fix. It usually means the close date was never anchored to the buyer's real process, the durable fix is setting close dates that actually mean something. This is why the slip count beats the probability percentage. The percentage is a guess that gets re-guessed every week. The slip count is a tamper-proof record of the buyer's actual behavior over time. When a rep says "it's at 80%, just slipped a bit," the slip count says "this is the third push." One of those is evidence. Most deals that die from slippage do it because nobody acted on the early pushes, the structural objection underneath the slip went unaddressed, which is squarely a closing problem we break down in why deals stall and what to do about it.
Section 6
What to track: days since the buyer last did anything
The third field is the simplest of all and it can be automated: days since the last buyer action, their last reply, their last proposal open, their last meeting. Not your last email. Their last move. The rule of thumb from the pipeline-hygiene literature is to flag any open deal with no buyer activity in roughly 14 days , alongside flags for no stage change in 30 days and a close date pushed two or more times. Fourteen days of buyer silence on an active deal usually means the deal is already over and you don't know it yet. That is the whole "what to track" list. A dated next step. A slip count. Days-since-buyer-action. Three buyer-verified facts. Everything else is optional and most of it is theater.
Section 7
What to ignore, on purpose
Ignoring is the harder discipline, because the theater fields feel productive. Permission to delete them: • The probability % slider. It is the rep's optimism dressed as math. If you want a probability, derive it from stage exit criteria the buyer actually crossed, not from a feeling. Otherwise, cut it. • Manual stage names with no exit criteria. A stage only means something if there is a buyer-verified event that moves a deal into it ("proposal sent and opened," "intro to economic buyer completed"). "Negotiation" with no defined entry test is just a word. • Vanity activity counts. Dials and emails sent measure your team's effort, which matters for coaching but is noise in a forecast. Do not let effort masquerade as progress. • "Just checking in" touches. A follow-up that moves nothing is not activity; it is the absence of a next step wearing a costume. If a touch did not produce a dated commitment, it did not happen, as far as the forecast is concerned. The credibility stakes are real: only 45% of sales organizations say their leaders have high confidence in their forecasting accuracy . The way you get into that 45% is not a better tool. It is refusing to count the fields that were never information.
Section 8
The BGA framework: The Two-Signal Pipeline (Next-Step-or-Dead)
Everything above collapses into one rule a solo-founder seller or a five-person team can run in any CRM, or a spreadsheet, with zero new software. A deal earns a place in your forecast only if it passes two tests: 1. It has a dated, mutually-agreed next step on the calendar. A specific event, a specific date, both sides agreed. No next step → the deal moves to a "No Next Step" holding column and is removed from the committed number. This is the rule with the sharpest teeth, because skipping the next step is associated with a 71% drop in close rates . 2. Its close-date slip count is zero or one. Track a tiny counter on every deal: each time the close date moves, +1. Two or more → the deal is flagged for a structural review (budget, authority, urgency) and pulled from the committed number until it earns its way back. This is your early read on the ~60% of deals that slip . Then run it on a weekly 20-minute cadence: 3. Sort, don't update. Friday's job is not "update the CRM." It is sort every open deal into Forecast (passes both tests), Holding (no next step), or Flagged (2+ slips). Three columns. That's the meeting. 4. Set the silence alarm. Add the one automated field, days since last buyer action, and flag anything past 14 days . This is the only field worth a piece of automation; wiring that alert is a small, classic AutomateOS job, the kind of low-effort signal automation that earns its keep because it removes a human judgment call. 5. Delete one opinion field this week. Probability %, deal temperature, or a stageless stage, pick one and remove it. You will not miss it, and the forecast gets more honest by subtraction. The metrics to watch over a quarter: your committed pipeline should shrink on day one (that is fiction leaving), then your hit rate against the committed number should climb. The target is the 23% accuracy gap a clean pipeline closes, earned by deleting fields, not adding them. If you want a one-page version to print and run your next pipeline review against, the checklist is built into our Template Pack.
Section 9
You're running the Two-Signal Pipeline right when…
You're running it right when your "committed" number is smaller than your gross pipeline and you trust the smaller number more. When your Friday review is a 20-minute sort into three columns, not an hour of typing. When no deal sits in the forecast without a date you can point at on a calendar, and a deal that slips twice triggers a hard conversation instead of a hopeful re-guess. When you have deleted at least one opinion field and the forecast got more believable. And when a salesperson saying "it's at 80%" makes you ask "what's the next step and how many times has it slipped?", because you both already know the percentage was never the point.