Marketing & GrowthFree
Lost Deal Pattern Analysis Report
A reason's share of your losses says nothing until you know how often you won with that same reason on the call.
Thurlow closed 640 opportunities in four quarters, won 184 and lost 456, a 28.75% win rate. Ravensworth was the alternative in 132 of the losses, 29.0%, the top line on every loss report the team had ever produced. Ravensworth was also the alternative in 66 of the wins. Thurlow's win rate against Ravensworth is 33.3%, above its own baseline. The company had spent two quarters building material against a competitor it already beats more often than it beats the field.
The move that makes this computable is reading the transcripts of the won deals. Thurlow's CRM named an alternative on 48.5% of losses and 22.3% of wins, because a rep fills that field when explaining a loss. Reading the calls raised won-deal coverage to 76.6%, 3.4 times as many. Every attribute then gets a win rate per encounter, and the ranking is by losses above what the baseline loss rate predicts rather than by share.
Written for whoever has to say something true about why the quarter went the way it did, to a board or to a product team. The review mining analysis gates competitor complaints through the same kind of test, and the competitor teardown is where a newly promoted competitor gets read properly. Send the capability findings into the product gap review, and take the surviving competitive claims to the battlecard pack before a rep repeats any of them.
Why the denominator is missing
A CRM records the reason, not the alternative. HubSpot's list of default deal properties carries a closed lost reason and, alongside it, the reason the deal was won. No competitor field appears anywhere in that list. So the schema captures why each side of the ledger happened and never captures who else was in the room, which is the one field a per-encounter win rate needs. Custom properties fix it going forward and do nothing for the deals already closed.
Segmenting a small dataset manufactures patterns. Thurlow's worst cell was public sector, a 5.9% win rate on 17 deals, 22.9 points under baseline. Its interval runs from 1.3% to 22.6%, so the honest statement is a shrug. NIST's handbook notes that with very few failures or a very small sample, symmetric limits from the normal distribution may not be accurate enough for some applications. So every cell carries a Wilson interval, and no cell under 25 encounters is reported at all.
Ranked by share of losses the order is Ravensworth, Padgett and then a bucket of deals with no alternative recorded. Ranked by losses above baseline it inverts: Instow at 9.7, an in-house build at 8.4 and the status quo at 7.1. Instow was last of six by share of losses and is faced in 58 deals, of which Thurlow wins 7. The excess concentrates in about 25 of the 456 losses, which is 5.5% of them.
How it works
Recover the alternative
Read the won calls as well as the lost ones, which is where coverage is missing.
Rate every attribute
A win rate per encounter for each competitor, segment, stage and concern raised.
Subtract the baseline
Excess losses per cell, so a big share of losses at an average rate ranks nowhere.
Suppress thin cells
Anything under the floor or inside its own interval is held back with its count.
What you get
- Every closed deal with the alternative it was lost or won against, sourced.
- A win rate per encounter for each competitor, segment, stage and capability concern.
- Losses above what the baseline loss rate predicts, which is the actual ranking.
- A Wilson interval on every cell, and a minimum cell size before reporting.
- The cells held back, with their counts, so nobody rediscovers them as findings.
- A Sheet of every cell and a Doc of the two or three patterns.
Common questions
Why is share of losses the wrong statistic?
Because it has no denominator. Ravensworth was the alternative in 29.0% of Thurlow's losses, more than anything else, and Thurlow's win rate against Ravensworth is 33.3% against a 28.75% baseline. A competitor you meet constantly turns up in a lot of your losses whether or not it is beating you.
We have no call recordings. Is this still worth running?
Partly. You can compute win rates on whatever attributes both sides of the ledger already carry: segment, size band, source, stage reached, sales cycle length. What you cannot do is rank competitors, because the alternative is recorded on your losses and not on your wins. That asymmetry was 48.5% against 22.3% here.
Does price ever come out as the pattern?
Not on this set. Price or budget came up on 187 of the 640 closed calls, and the win rate when it did was 28.88% against a 28.75% baseline. It is the most frequently raised concern in the corpus and it moves the outcome by a tenth of a point. Reason codes put it first.
What is an excess loss?
The losses in a cell minus what the baseline loss rate would have produced there. Thurlow loses 71.25% of everything, so 58 Instow encounters should produce 41.3 losses and produced 51. The excess is 9.7 deals. It converts a rate back into deals, which is what makes two cells comparable.
Won't a rep's memory of the competitor be wrong too?
It might be, which is why the alternative comes from what was said on the call rather than from a field filled in afterwards. Where a transcript names no alternative, the deal sits in an unknown bucket that is reported rather than dropped. That bucket was 112 deals here, and it wins at 38.4%.
How is this different from win-loss interviews?
Interviews ask a sample of buyers what happened. This reads every closed deal you already have and asks a different question, which is whether an attribute changes the outcome. The two fit together: an interview programme explains a cell, and this tells you which of the 30-odd cells is worth the interviews.
What actually gets delivered?
A Sheet with one row per cell, carrying encounters, wins, win rate, its interval, expected losses, actual losses and excess. A Doc with the two or three patterns that cleared both tests, the cells held back and why. Competitor-specific detail belongs in the competitor teardown, and where the whole set sits relative to each other belongs on the positioning map.
Lost Deal Pattern Analysis Report
Fill in the form and your workspace opens with the work already underway.