How a stage win rate and a slip are measured

Which deals are judged at all

Only closed deals, and only those that were open long enough to have been forecast. A deal created and closed inside 14 days was never forecast; it was logged after the fact. Those deals are counted and named on the tile Logged after the fact, and they are in no rate on the tab.

A closed deal carrying no created date is judged, because there is no span to measure it against. Where your store holds many of those, read the Closed deals judged tile as the outer edge rather than an exact population.

The other three tiles state the population directly: Closed deals judged, Won and Lost. Read them before anything else on the page, because every percentage below rests on them.

Where your store holds more closed deals than were read, the method note says so and gives both numbers — an even sample of the closed deals in that tool.

The stage win rate

For each stage, MarqIQ takes the judged deals whose history shows them in that stage at any point before they closed, and counts how many of those were won.

• A deal counts at every stage it passed through, so one deal appears in several rows. The rows do not add up to your deal count, and should not.
• A deal lost from a stage is counted at that stage. Excluding it would mean each stage's win rate measured only the deals that got past it, which is a different and much flattering question.
• Deals that are still open are in no row. A win rate is over closed deals.
• Where the stage belongs to one pipeline, only deals in that pipeline, or deals with no pipeline recorded, are counted against it.

The table prints four columns: the stage, what your pipeline says the probability is, what it has been, and the number of closed deals behind that. The fourth column is the denominator — quote it with the third.

When a stage is called optimistic

A verdict appears only with at least 30 closed deals behind the stage and a probability configured in the tool. Below 30 the cell is blank rather than guessed at.

The five-point margin is there so a stage set to 60% against a real 58% is not paraded as a fault. The finding raised from this is a fault rather than a risk when two or more stages are optimistic, or when the worst one is out by 20 points or more.

What a Commit turned into

Where your workspace has a forecast-category field, MarqIQ reads what it said 14 days before each deal was decided, and groups the judged deals by that value. The card What each forecast call turned into prints, for every category, how many were won out of how many.

Fourteen days is the look-back because it is far enough out to be a forecast and close enough in that the call should hold. A deal open for fewer than 14 days has no reading at that moment, which is why those deals are excluded from the whole tab.

A finding is raised on the commit-like category — the one whose value reads as a commitment rather than a best case or a pipeline bucket — only when it has at least 30 deals behind it and the top of its 95% range is still under 80%. Below 65% it is a fault rather than a risk.

Why only one or two owners are named

An owner is named only with five or more deals in that category. A single bad quarter on two deals is not evidence about a person, and a rumour is what it becomes once it is written down.

• An owner is listed as reliable when at least four in five of their deals in that category were won, on five or more deals.
• An owner is listed among the misses when their rate sits 15 points or more below the category's own rate, on five or more deals.
• Each name carries its two counts — “3 of 11” — so nobody is described by a percentage alone.
• The reliable owners are named before the misses, because the fix is usually to write down what they require before they use the category.

How far a close date slips

Slip is measured on won deals only, against the close date as it stood 30 days before the deal closed — the value read from history, not the one on the record now.

• A won deal open for fewer than 30 days has no reading at that moment and is left out.
• The figure needs at least 30 won deals with a readable date that far back. Below that, no slip is claimed at all.
• Three numbers are printed: the median days late, the share that closed no more than a week after the forecast date, and the share that closed more than a month after it.
• A deal that closed early is inside that first share. It is a count of deals not more than a week late, not of deals within a week either side.
• A negative median means the typical deal closed on or before the date it was forecast for, and the sentence says so rather than printing a minus sign.

A finding is raised when three in ten won deals closed more than a month late, and is a fault rather than a risk at five in ten. The damage is stated plainly: a forecast right about who will buy and wrong about when still misses the quarter, and hiring and spend are planned against the when.

What the tab cannot tell you

• Nothing about open deals. Every figure is on deals that have closed.
• Nothing about a field your tool does not keep history for. Where history could not be read, the headline says so instead of showing an empty table.
• Nothing about a stage that has fewer than 30 closed deals behind it. The row is still shown, with its counts, and no verdict.
• Nothing about deals logged after the fact, beyond how many there were.