Every year we ask revenue leaders the same uncomfortable question: how close did your forecast land to what you actually closed? This year 412 teams answered, from 20-person startups to public companies, and 187 of them let us analyse the underlying opportunity data — 1.1 million deals in total. The gap between the best and the rest has never been wider, and it has surprisingly little to do with headcount or tooling budget.
This is our third annual benchmark. As in previous years, we asked about methodology, tooling, team structure and data practices, then matched survey answers to outcomes wherever teams shared their CRM data. Figures below are medians unless stated otherwise, and every company is anonymised.
The top line
The median team missed its quarterly forecast by 14%. The top quartile missed by less than 5%. Three findings stood out:
Fewer fields, better data. Top-quartile teams require a median of 11 opportunity fields. Bottom-quartile teams require 27.
Fresh data wins. Teams whose CRM updated within an hour of buyer activity were 2.4× more likely to land within 5% of forecast.
Bad news travels early. Accurate teams downgraded at-risk deals a median of 19 days before quarter end, versus six days for everyone else.
Team size barely registered. Top-quartile teams ranged from 6 to 400 sellers, and the typical top-quartile team actually ran a leaner RevOps function — one RevOps person per 14 sellers, versus one per 11 in the bottom quartile.
Speed beats sophistication
The most surprising finding was how little the forecasting method mattered. Teams using weighted pipeline, category-based commits and AI-driven models were spread almost evenly across every quartile. What predicted accuracy was data latency: how long it took for something that happened in the real world to show up in the CRM.
In bottom-quartile teams, that lag averaged 6.3 days. In top-quartile teams, it was under a day. You cannot forecast accurately on a pipeline that describes last week.
Latency compounds, too. A deal whose last activity is a week out of date isn’t just stale — it is reviewed, weighted and rolled up as if it were current. By the time the forecast call happens, every layer of the process has added confidence to information that was already wrong.
“We cut our required fields from 31 to 12, and our forecast got better the very next quarter. It felt like cheating.”
— Ruth Okonkwo, Director of RevOps, Kindred Health
What to do with this
If you change only one thing this year, measure your data latency. Pick 20 recent deals, compare when key events happened with when they reached the CRM, and share the number with your sales leadership. It’s rarely flattering, and it’s almost always the fastest path to a better forecast.
Then shorten it. Automate the updates that depend on activity data, cut every required field nobody uses in a forecast conversation, and move your at-risk review earlier in the quarter. None of this needs a new methodology — just a pipeline that tells the truth sooner.
The full report — with segment breakdowns, the survey instrument and our methodology — is free to download, no demo call required.
