The forecast is an opinion poll
Most SaaS forecasts are a roll-up of each AE's judgment, adjusted by a manager's gut. The CRM stages look like data, but a stage is just a field somebody changed. Two AEs can put identical deals in different stages, and the same AE can move a deal forward because a buyer said something encouraging.
That is why the forecast is optimistic early and wrong late. The optimism is honest. AEs genuinely believe the friendly call means progress. The problem is that nobody is checking the belief against what was actually said.
What actually predicts a close
Work back from the deals that closed and the ones that slipped, and the difference is rarely the buyer's tone. It is whether specific things happened on the calls: the economic buyer was present, the cost of the problem was named in the buyer's own numbers, a second stakeholder was involved, and every meeting ended with the next one booked.
Each of those is observable. Each is on a recording. And almost none of them show up in a forecast built from stage fields and AE confidence.
- Economic buyer never on a call
- Pain described but never quantified
- Single-threaded through one champion
- Next step agreed in principle but not on the calendar
Why pipeline reviews do not fix it
The weekly pipeline review is meant to catch this. In practice it becomes AEs narrating their deals and managers asking how confident they feel. The AE who tells the best story gets the benefit of the doubt, and the review measures storytelling rather than deal quality.
Inspecting a handful of recordings helps, but nobody has time to listen to every call. So the review falls back to opinion, and the forecast inherits the same blind spot it started with.
It starts as a skill problem
Underneath the forecast is a skill gap. AEs who are not confident asking for the economic buyer, quantifying pain or booking the next step on the call will skip those moves, and their deals will look healthy right up until they slip. Better forecasting methodology cannot fix a deal where the work was never done.
Training rarely closes that gap for long. 84% of what reps learn in training is gone within 90 days (Sales Performance International), and only 29% of companies can see whether training changed what their reps do. Customer AI & Role Play gives AEs a place to rehearse the uncomfortable asks against an AI buyer, scored in about 60 seconds, so the moves are habits before they matter on a live deal.
Forecast from the call, not the field
Intel Suite records, transcribes and scores every tagged call and syncs the result to HubSpot so it sits on the deal. Your own process carries 60% of the weight. If your standard is that a commit deal has had the economic buyer on a call and a booked next step, a deal that has neither scores low no matter what stage the AE put it in.
That turns the pipeline review into a different conversation. Instead of asking how confident the AE feels, you look at what the calls show and ask why the missing step has not happened yet.
One number per rep, and a forecast you can defend
The Compliance Score is one number per rep, 0 to 100, recalculated nightly: training completion 30%, daily tasks 25%, role play quality 25%, manager enforcement 20%. The dashboard ranks the team and sorts a coaching queue by urgency, so you know whose deals to discount before the quarter tells you.
A forecast you can defend is not a more sophisticated spreadsheet. It is one where every commit deal has evidence behind it, and the evidence came from the conversation, not the AE's mood.
The short version
- Most SaaS forecasts are a roll-up of AE opinion dressed up as CRM stages.
- What predicts a close is observable on the call: economic buyer, quantified pain, multiple stakeholders, a booked next step.
- Pipeline reviews drift into storytelling when nobody can check the calls.
- Underneath the forecast is a skill gap that training alone does not hold.
- Score calls against your own process so commit deals carry evidence, not confidence.
Questions owners ask
Should we switch to a new forecasting methodology?⌄
A methodology helps you ask better questions, but it still depends on the answers being true. If the inputs are AE opinion, a new framework produces a more organized wrong number. The bigger gain comes from checking each commit deal against what happened on the calls.
How many calls should a founder review before the forecast call?⌄
Ideally none by hand. The point of scoring every call against your process is that the missing steps surface on their own, so your time goes to the deals where the evidence is thin rather than to listening at random.
Is a bad forecast a sales leader problem or an AE problem?⌄
Both, and blaming either misses the fix. AEs need the skill to do the steps that make a deal real, and leaders need visibility into whether those steps happened. When both exist, the forecast tends to fix itself.
See how the system runs in SaaS.
Revenue OS for SaaSYour best month ever, and the plan to beat it.
Break Your Revenue Record