The definition
Average deal size is total won value divided by the number of won deals in a period. Win eleven deals worth €330,000 together and your average deal size is €30,000. That is the whole formula. Everything difficult about this metric happens before the division. You decide which deals count, which value goes on them, and over which period you look.
Two averages are worth having. The mean is the calculation above. The median is the deal in the middle when you sort every won deal by value. On a healthy book the two sit close together. When they separate, the mean has been pulled by outliers. Sales teams quote the mean because it is bigger. Sales planning needs the median as well, because that is the deal a rep will actually see next week.
Why it matters
Take a team that won eleven deals last quarter. Ten of them landed between €12,000 and €18,000. One landed at €180,000. The mean is €30,000. The median is €15,000. If you set next quarter's plan on the mean, you tell six reps to find €30,000 deals in a market that produces €15,000 deals. The plan fails in week three, and nobody can point at the reason, because the reported average was correct.
The number also drives the maths elsewhere. It is one of the four inputs to sales velocity. It also sets the deal count you need behind a quota, which pipeline coverage then converts into a target. Get the average wrong and both of those numbers inherit the error silently. A team chasing coverage built on an inflated average is short of pipeline without knowing it.
How to measure it in Pipedrive
Build an Insights deal report filtered on won deals, over a full quarter. Display the sum of deal value next to the deal count. Pipedrive gives you the average directly, but keep the count visible so you always know how many deals the number rests on. Then add a second report grouped by deal value ranges, so the shape of the distribution is on screen next to the average.
Segmentation is where the metric earns its keep. Split the same report by pipeline, by industry, by lead source and by owner. Each split answers a real question: which market pays more, which channel brings bigger work, which rep sells up. Those splits need the segment as a proper field on the deal. Build it with custom fields, never by typing it into the deal title. If your deal values come from line items, keep them consistent using products and pricing. A deal priced by hand and a deal priced from the product catalogue do not compare.
Where teams get it wrong
The first mistake is one average for the whole company. A book with €4,000 renewals and €90,000 implementations produces an average that describes no deal anyone has ever sold. Segment first, then average. The second mistake is counting open deals. Early values are placeholders, often round numbers a rep typed to move on. Averaging them measures optimism, and it makes your forecast read high.
The third mistake is mixing currencies or mixing recurring and one off values in the same report. Pipedrive can convert currencies for you, but it cannot know that half your deal values are annual and half are first year totals. Write the convention down and enforce it. The fourth mistake is treating a rising average as good news on its own. It also rises when small deals stop closing. Read it beside win rate and deal count, never alone.
Questions
Should I use the mean or the median?
Report both. The mean tells you what the book is worth per deal, so it feeds capacity and quota maths. The median tells you what a normal deal looks like, which is what reps recognise. When the two numbers sit far apart, your deal sizes are skewed and one large customer is steering every average you publish.
Do won deals only, or all deals?
Won deals only, for the headline number. Open deals carry values that were guessed early and never corrected, so they drag the average toward optimism. Track the average value of open deals separately as an early warning that reps are quoting differently. Never mix the two in one figure.
How do discounts and recurring revenue fit in?
Put the deal value at the amount actually invoiced in year one, after discount. For subscriptions, decide once whether the deal holds annual value or first year total including setup, then write that rule down. Mixing the two conventions inside one pipeline makes every average and every forecast unreadable.
How often should the number be reviewed?
Quarterly for the headline, and immediately after any price change. A pricing move takes one full sales cycle to show up in won deals, so a monthly reading mostly measures noise. Below roughly 20 won deals in the period, read the individual deals instead of the average.