Average deal size is closed-won revenue divided by the number of closed-won deals you count in the same period. It looks like one number, but it tracks two different things, which deals you are winning and what price you are winning them at, and a report that blends both hides which one actually changed.
A manager who blames the wrong cause fixes the wrong problem. A drop driven by more small accounts closing gets treated as a pricing issue, the pricing conversation goes nowhere because pricing was never the problem, and the number drifts the same way again next quarter for the same reason nobody actually found.
What average deal size actually measures
Average deal size is closed-won revenue divided by the number of closed-won deals in the period, nothing more. Pull only deals that actually closed won. Leave open or lost deals in the denominator and the average drops for reasons that have nothing to do with how your real deals are sized.
The revenue basis you pick matters as much as the formula itself. A five-year contract worth $50,000 in total is a $50,000 deal measured as total contract value, and a $10,000 deal measured as annual contract value, according to HubSpot's breakdown of ACV versus TCV. Both numbers describe the same contract honestly. They are simply not the same number, and a report that mixes deals measured one way with deals measured the other way produces an average that means nothing specific.
Pick one basis, usually annual contract value for anything that renews or runs multiple years, and hold it across every period you compare. The moment the underlying basis shifts mid-quarter, whether a new rep logs contracts differently or a CRM field gets repurposed, the average moves on its own, and nothing about the chart tells you the deals themselves did not actually change.
Why the mean can lie: mean vs. median deal size
A single outsized contract can move the average more than ten ordinary deals combined, which is why the mean and the median are worth pulling side by side every period instead of trusting the mean on its own.
Pipeline Decoded's worked example makes the gap concrete. Twelve deals closed in a quarter total $340,000. The mean comes out to $28,333. The median, the deal sitting exactly in the middle once you sort them by size, comes out to $14,500, less than half the mean.
Ten of the twelve deals sit below the mean. Only one of them is anywhere near it.
One $166,000 contract is doing most of the work in that mean, and it is easy to miss unless you check both numbers. The gap is not a rounding difference either, it changes what you plan around. Forecasting off the $28,333 mean projects roughly $425,000 in quarterly revenue from that same pipeline. The $14,500 median projects closer to $217,500 from the exact same deals. The same split decides whether capacity planning assumes you need 71 deals or 138 deals to hit a $500,000 quota.
There is no universal cutoff for when the gap matters, but a rough rule holds up in practice. Once the mean sits one and a half times higher than the median or more, treat the mean as describing one deal, not your pipeline.
The other two reasons the number drifts: deal mix and dirty data
Outliers are not the only way average deal size moves without your actual deals changing shape. A shift in which segment is closing moves the blended number even when every individual segment's own average held perfectly steady.
Pipeline Decoded's data shows exactly this pattern. A blended average deal size rose 9 percent in a quarter while new-business deal size on its own actually fell 13 percent, because renewals, which run larger at that company, made up a bigger share of what closed that quarter. Read the blended number alone and the story is deals got bigger. Split it by segment and the real story is new business actually shrank.
Below the mix shift, a handful of ordinary data problems move the number with nobody noticing:
None of these are your deal sizes changing. They are the input data changing underneath a metric nobody re-checked before presenting it.
How to track it without fooling yourself
Segment before trusting a single blended number. Recompute average deal size within each product line, tier, and rep tenure band on its own, the same way you would isolate a mix shift. If every segment holds its own average steady and only the blend moved, the drift is a mix shift, a go to market fact worth planning around, not a pricing problem worth fixing. If a segment's own average moved too, that specific segment has a real pricing or packaging issue to chase.
Compare the trend against your own history before reaching for an external benchmark at all. HubSpot's own sales benchmark survey found reported average deal sizes spread from the $5,001 to $10,000 range up through deals over $100,000, with close to one in ten respondents unsure what their own average deal size even was. A spread that wide across real companies means an industry number tells you almost nothing about whether your own number this quarter is healthy. Your last four quarters are the only benchmark worth comparing against.
Average deal size also feeds two reports you likely already run, so track it there instead of as a standalone chart nobody else looks at. It is one of the four inputs behind pipeline velocity, and it sits next to stage duration on the short list of numbers that show whether a pipeline is actually moving at the pace your forecast assumes.
What a shrinking deal size can actually be telling you
A falling average deal size with a flat win rate is sometimes neither a pricing problem nor a mix shift. It can mean reps are closing against the budget one champion can approve alone, instead of the larger scope the full buying group would have signed off on.
A champion with a $15,000 discretionary budget closes a $15,000 deal every time, even when the account's real need, and the economic buyer's real budget, would support a $60,000 rollout across three departments. The rep never finds that out, because the rep never spoke to anyone who could confirm it, and the deal closes at the one number the champion was personally allowed to sign off on.
Buying group mapping exists for exactly that gap. Instead of guessing whether a bigger scope exists, a rep sees the roles typically involved in a purchase this size, who controls budget, who uses the product day to day, and who has to approve new vendors, before the proposal goes out at whatever number the first contact happened to mention.
Check this before blaming pricing or writing it off as a mix shift. Pull your five smallest closed-won deals from last quarter and see how many had more than one confirmed stakeholder attached. If most show just one, the deal size problem started before the first proposal ever went out, not at the negotiation table.
Where to start this week
Pull last quarter's closed-won deals and run both the mean and the median. If they sit far apart, one contract has been quietly writing the story you have been telling about the quarter, and it is worth knowing that before the next forecast review.
Then split the number by segment, product line, or rep tenure before anyone touches pricing. A falling blended average with every segment holding its own steady is a mix shift to plan around. A segment whose own average moved too is a pricing or packaging conversation worth having, but only in that segment.
Finally, check whether the smallest closed-won deals share a pattern of single-stakeholder closes. See how LeadLx supports the account side of that check, mapping who actually needs to sign off before a deal gets negotiated down to what one contact alone can approve, on the pricing page, starting with a free plan.