Sales Cloud · audit of an existing org
Eight patterns,
and every one of them ends in the forecast.
None of these raise an error. The system runs correctly and reports something untrue, which is why they survive for years and why the conversation only starts when a quarter misses. Each one below has the check that finds it in your own org.
Reviewed 28 August 2026.
Before the eight
An audit and a health check answer different questions.
A health check inventories configuration: what exists, what is unused, what departs from best practice. It produces a list, and the list is often long and rarely decisive.
An audit asks whether the system produces numbers somebody can act on. It produces an argument about the forecast rather than an inventory. Both are legitimate work, and buying one while expecting the other is the most common disappointment in this category. Everything below belongs to the second kind.
Pattern 01
Stages that mean different things to different reps
The stage exists as a label. What it means is decided by whoever moved the record. For one rep negotiation means terms under review, for another it means a quote was sent, for a third it means the call went well. The pipeline report adds all three together.
How to check it
Measure how long deals sit in each stage, per rep, and compare the spread. Wide variance in the same stage across reps means the stage has no shared definition, whatever the field help text says. Then ask three reps to define one stage in a sentence, separately.
Pattern 02
Pipeline full of deals that are not deals
Opportunities with close dates in the past, no activity for months, and no realistic path to close. Nobody deletes them because nobody is asked to, and coverage ratios are calculated over the total.
How to check it
Filter for open opportunities with a close date already passed, then for those with no activity in ninety days. Do not argue about the right threshold: whatever number comes back, ask whether leadership knows it is in the coverage figure.
Pattern 03
Leads arriving in queues nobody owns
Assignment rules, round robin and status workflows were configured once. Over a year or two you accumulate queues whose owner is deactivated and round robin entries for people who have left. Leads land there and stop.
How to check it
List every queue and check its owner is active. Then find leads with no activity since creation and group them by owner. It is almost always one or two queues rather than a spread, which makes it cheap to fix once found.
Pattern 04
Probabilities from the setup wizard rather than from your business
The default probability on each stage came with the org. Nobody changed it, so the weighted forecast is arithmetic performed on numbers that describe a hypothetical company.
How to check it
Take the last four quarters of closed opportunities and calculate the actual conversion from each stage to closed won. Put that number next to the configured probability. The gap is what your weighted pipeline is wrong by, and it is usually not small.
Pattern 05
Half the customer contact never reaches Salesforce
Calls happen on a phone system, messages in a chat tool, meetings in a calendar that was never connected. The activity that would tell you a deal is alive is recorded somewhere the forecast cannot see.
How to check it
Pick ten deals that closed won last quarter and count the activities logged against each. Then ask the rep who closed them how many touches it actually took. The ratio between those two numbers is how blind the forecast is.
Pattern 06
The same customer as several accounts
One from lead conversion, one typed by a rep, one from an import, one from an integration. Each holds part of the history, and every report that groups by account is quietly averaging across fragments.
How to check it
Sort accounts by name and look for near duplicates, then check whether open opportunities sit on more than one of them. Finding this takes minutes and it changes what every account level report means.
Pattern 07
Field sprawl that reps route around
Custom fields accumulate the way debt does: one per quarter for a campaign, a report filter, a compliance requirement. The layout grows until filling it honestly costs more time than the rep has, so it gets filled dishonestly or not at all.
How to check it
Count the custom fields on Opportunity, then count how many are populated on deals closed in the last quarter. The difference between existing and used is the part that is costing adoption without returning anything.
Pattern 08
Dashboards whose filters describe an older company
The dashboard was correct when it was built. Then a record type was added, a team was renamed, a product line was split, and the filter kept excluding what it always excluded. It shows a number, it shows it confidently, and it is not the number anybody thinks it is.
How to check it
Take the three dashboards leadership actually opens and rebuild one number in each from the raw data by hand. If the two disagree, you have found why nobody trusts the reporting, and it was never about the data.
What they have in common
Every one produces a system that runs correctly and reports something untrue.
That is the property worth naming, because it explains why monitoring never catches them. There is no alert for a stage that means three things to three people. No error for a dashboard filter describing the company as it was two reorganisations ago. No warning that half the customer contact happened somewhere Salesforce cannot see.
It also explains the sequence in which they surface. Nobody investigates any of this while the numbers look plausible. The investigation starts after a quarter misses, which is the most expensive possible moment to begin.
Where to start
The probabilities, because it is an hour and it settles the argument.
Four quarters of closed opportunities, the actual conversion from each stage to closed won, placed next to the percentage configured on that stage. One hour of work, and the result is a single number: how wrong the weighted pipeline is.
Start there rather than with the loudest complaint, because every other item on the list is easier to discuss once that number is on the table. Arguments about pipeline hygiene and stage definitions tend to be about opinions until somebody produces arithmetic.
Asked often enough to answer here
Questions
What is the difference between a Sales Cloud audit and a health check?
A health check inventories configuration: what exists, what is unused, what is misconfigured against best practice. An audit asks a different question, which is whether the system produces numbers you can act on. The first produces a list. The second produces an argument about the forecast. Both are useful, and confusing them is why people buy one and expect the other.
None of these raise an error. How would we know we have them?
That is the defining property of all eight, and it is why they persist. Every one of them produces a system that runs correctly and reports something untrue. There is no alert for a stage that means three things, or a dashboard filter that describes last year. The checks above exist because the platform will never tell you.
Which one should we look at first?
The probabilities, because it is the cheapest check with the largest immediate consequence. Four quarters of closed opportunities against the configured percentages takes an hour and tells you how wrong the weighted pipeline is. Everything else on the list is easier to argue about once that number exists.
Can our own admin do this?
Most of it, yes, and that is why the checks are written out rather than summarised. What an outside pass adds is that it is not diplomatic: an internal admin has to keep working with the people whose stage definitions and dashboards are the problem.
What does it cost to have it done?
3,750 euro for 75 hours, which is the engagement described as the CFO not trusting the numbers. That covers the eight checks above, the attribution and data model work behind them, and definitions written down and agreed rather than assumed. A narrower diagnostic is 500 euro and runs against your own exports.
€3,750 for 75 hours. A narrower diagnostic against your own exports is €500, delivered within three working days and credited against the work. That number sits below what a partner would quote deliberately rather than as a discount: no delivery manager, no partner margin, and every price is published here.
Serhii Skrypnyk · Senior Salesforce Administrator and developer · 7 Salesforce certifications · on the platform since 2018. Reviewed 28 August 2026.