Weekly deals won rose 24% after a sales forecasting model at a B2B SaaS

At a B2B SaaS, average weekly deals won rose 24% in the four weeks after a predictive sales model was adopted, almost one extra deal for every four. Pipeline conversion rose 14% over the same period. The model crossed the conversion probability of each funnel stage with the time and contact cadence in that stage. That way, the team learned where speed is critical and how many contacts each stage needs.

Profile

B2B SaaS · Series A

Problem

The sales team worked the pipeline without knowing which contact behaviours actually increased the chance of closing. Cadence depended on each rep.

Solution

Use the funnel's own data to find out, stage by stage, what closes deals, and turn it into a team routine.

What we did

We mapped the funnel stages and, in each one, the time spent and the volume of communications and follow-ups.
We built a conversion forecasting model, with initial accuracy of about 70%.
We crossed the conversion probability of each stage with the time and contact cadence in that stage, and showed the team where speed is critical and what the right cadence is.

Result

+24% average weekly deals won in the following four weeks and +14% pipeline conversion, measured as deals won over pipeline size.

Key learnings

Speed doesn't weigh the same at every stage. In some, acting fast increased the chance of closing by more than 10%. In others, it made almost no difference.
Each stage has its right number of contacts and follow-ups. Getting that cadence right increased the chance of closing by more than 30%.
Just measuring already improves the team's behaviour. When reps know what's tracked, they adjust.

Do it yourself

Export won and lost deals from the last 6 to 12 months from your CRM with, for each stage, the entry date, exit date and number of contacts and follow-ups.
In each stage, compare the close rate of deals that moved fast with those that took longer, and of those with more contacts with those with fewer. Where the difference is large, the stage gets a time and cadence target for the team, tracked every week.
Ask an LLM, after removing names and emails from the export:
I'll paste a CRM export with won and lost deals and, for each funnel stage, the entry date, exit date and number of contacts. For each stage, calculate the close rate by time-in-stage band and by number-of-contacts band, tell me in which stages speed and cadence make the most difference and propose a time and contacts target per stage.