How a bloated, unreliable pipeline became a forecasting baseline leadership could actually trust - and what it took to get there.
Every month at our pipeline review, the same issue surfaced: we had enterprise deals that had been sitting open for six, eight, ten months with no real movement. We had built a pipeline decay report - a stacked vintage chart that tracked unweighted ARR by the quarter it entered the pipeline. It clearly showed that month after month our older cohorts weren't clearing. They were just sitting there, slowly bleeding but never dying.
The conversation would come up, someone would nod, and then it wouldn't come up again for another month. There was a section in our Rules of Engagement documentation about acceptable time-in-stage limits. There were just limited mechanisms to enforce them, and a fuzzy understanding of whose mandate it was to do so.
We were carrying roughly $10-13M in unweighted pipeline at any given time. A significant portion of that was deals older than six months with no recent activity - not officially dead, but not alive, just sitting.
The pattern had a name once we looked closely enough: stalled opportunities - defined as any deal three or more months in a stage with no touchpoint in the last 30 days - had grown from a manageable minority to roughly 40% of total pipeline. That number made our coverage metrics and forecast look healthy and robust. They weren't.
The stalled pipe was a symptom, but the disease was a sales process that hadn't evolved when the product did.
After raising the issue repeatedly in pipeline reviews without resolution, I pushed for a formal audit of our stage architecture. What we found was that our pipeline had one combined stage for discovery and demo - a rep would hold a meeting, move the deal to “Qualified,” and the SDR got credit for the booking. In practice this meant AEs were incentivized to move everything forward regardless of fit. Why close something out when moving it costs nothing?
We also had a non-linear “Stalled” stage built into our pipeline. I’m sure it had been created with good intentions, but it had become a graveyard - a holding area for reps to hoard accounts in the form of “active opportunities”. These sorts of stalled/holding stages consistently gum up pipeline regardless of rules and oversight.
The structural problem went deeper than that. We had launched multiple new products without updating our sales process to match. A consultative, multi-product sale needs a discovery call before a demo - you have to understand the customer’s environment before you can recommend anything. We were skipping that step entirely and then wondering why so many demos went nowhere.
SDRs deserved credit for qualified meetings they booked. The fix wasn’t necessarily to change how they got paid, but to insert a buffer stage so AEs couldn’t dump unqualified deals into the pipeline we used for forecasting.
We introduced a Discovery Call Held stage between initial contact and what we renamed Solutions Presentation Held (formerly “Qualified”). SDRs still got paid for qualified disco calls. AEs now had a gate before committing a deal to active pipeline. It was a win for both sides, and it meant our forecasting coverage metrics were finally counting the right things.
The stage rename was just one piece. I rewrote the full stage architecture from scratch. Not just what stages existed, but why each one existed, what gate logic it required, and what it meant from the prospect’s point of view at that moment in their buying journey.
| Before | After | Change |
|---|---|---|
| Demo Scheduled | Discovery Call Scheduled | Replaced to reflect the new emphasis on discovery before demo |
| — | Discovery Call Held | New stage - active buffer before pipeline entry |
| Qualified | Solutions Presentation Held | Renamed to reflect prospect action, not internal judgment |
| Progressing | Buyer Consideration | Gate logic tightened - requires documented pain and timeline |
| Negotiation | Negotiation | Stage definition clarified - verbal yes required to enter |
| Contract Sent | Pending Signature | More formal entry criteria - redlines/pricing already finalized |
| Closed Won / Lost | Closed Won / Lost | Loss reason taxonomy completely overhauled for closed-lost reporting |
| Stalled | — | Eliminated - a sub 2% historic conversion rate meant it wasn’t worth maintaining |
Every gate had a written rationale: what properties it required, why those properties mattered to the business, and how they got populated. The goal was that any AE joining the team six months later could read the process doc and understand not just what the rules were, but why they existed and what was needed from them throughout the sales cycle.
Before anything went live in HubSpot, we did one large cleanup sweep - manually auditing open opportunities and closing out anything that met the stall criteria. That initial purge was intentionally aggressive. Anything older than six months with no recent activity was closed out almost immediately. Reps were given clear instruction and deadlines to clear out the rest of their pipeline.
The vintage chart tells the story better than any summary metric. Both charts below are identical through June 2025. At July - when the overhaul went live - the right chart diverges.
The pipeline number went down. Leadership now had a forecasting baseline they could actually build a plan around - not a $12M stack where half the deals hadn’t had a touchpoint in months.
The harder change was cultural. We rolled the stage overhaul out alongside a more formal weekly pipeline review process - best case and commit calls, structured rep forecasting, the works. For a sales team that was largely homegrown or in their first enterprise SaaS role, that was a muscle that took time to build. Adoption wasn’t instant, but we had a few model citizens who actively coached the holdouts. The pushback wasn’t about the stage logic either. It was about suddenly having to defend your pipeline in a structured weekly call when you hadn’t had to before. It took time, but week over week we had reps coming into pipeline reviews more prepared and with cleaner pipelines.
The long-term vision included more automated enforcement: calculating mean time-in-stage from historical closed-won data per segment, flagging deals at one standard deviation, closing them out at two after a follow-up warning. We got the manual process working first and the automation piece was actively being built.
Pipeline hygiene problems are almost never about the pipeline. They’re about the incentives and processes upstream that let bad data accumulate in the first place. Cleaning the data without fixing the process just means you’re doing the cleanup again in six months.
Another thing - the stalled pipe wasn’t invisible. It came up in every pipeline review. The difference between it staying a recurring agenda item and actually getting fixed was my team finally getting fed up and escalating with enough specificity that leadership had to act. The analysis and the solution were ready, just waiting for the right moment to matter.
Retrospectively, I could have pushed harder on this pain earlier. But when the time did come, we attacked the problem with a commitment and veracity that delivered quick, org-changing results.