Short, practical pieces on the problems that come up regardless of which system you use. No email address required to read any of them, because gating a checklist behind a form has never made the checklist better.
A stage is a shared claim about what is true. If two people apply it differently, every number downstream is unreliable. The fix is short, testable criteria — "budget confirmed by someone who controls it" rather than "qualified" — written where people can see them.
Keep the list to four or five stages. Longer pipelines are usually describing internal activity rather than the buyer's decision, and activity does not predict close dates.
Deals with no scheduled next step, and deals that have sat in one stage past its normal duration. Between them these catch most of what silently inflates a forecast.
Review them weekly in a meeting that already exists. Teams that add a dedicated hygiene meeting tend to stop holding it within a quarter.
A migration is the only realistic chance you get to leave data behind. Contacts with no activity for years, deals that were never real, and custom fields nobody has filled in since the last migration — none of it improves by moving.
Map fields deliberately, import a sample first, and check the duplicate review before committing. An afternoon here saves months of mistrust in the numbers.
Making fields mandatory at record creation produces fast, imaginative garbage. Making them mandatory at the stage where they genuinely matter produces data.
Tie requirements to stage transitions instead: budget before Proposal, decision-maker before Negotiation. The friction lands where the deal justifies it.
Rules encode assumptions about a process. Encoding a process you are still changing means rewriting the rules every time it moves, and eventually leaving broken ones in place because nobody remembers what they were for.
Run manually for a few weeks. Automate the things that turned out to be genuinely repetitive, and give every rule a named owner.
Weighted pipeline, committed deals and raw open value answer different questions. Shown alone, each is misleading in a predictable direction — the weighted number flatters a fat early pipeline, the committed number hides everything not yet promised.
Show them together, and track the movement between reviews rather than the snapshot.
Rarely a feature problem. Almost always a cost-versus-return problem for the individual rep: maintaining the record takes real effort and the benefit accrues to someone else.
The systems that survive are the ones where the rep gets something back the same day — context before a call, a list of what needs doing, a handover that works when they are ill.
A free-text loss reason produces a hundred unique answers and no analysis. A fixed list of six or seven, chosen at close, produces something you can act on.
Include "no decision" as an explicit option. In most pipelines it is the largest category, and hiding it inside "lost to competitor" leads teams to solve the wrong problem.
The questions worth asking any vendor, including us: what does export cost, what happens to the data if we leave, how are stage definitions enforced, and what does the product deliberately not do?
We would genuinely like to hear it. These notes come from watching teams work, and they get better when someone argues with them.
Get in touch