Analysing a sales funnel requires confirmed customer transitions between stages. First, the head of sales checks statuses against events in the deals, then counts transitions for a single group of customers as of a chosen date. Deals whose state is unknown are counted separately. Otherwise, the owner will mistake a gap in the records for a lost customer and start fixing sales where the data needs to be restored first.
During an audit of our client, a wholesale company, we found something odd: detailed analytical reports were available in the CRM system, but it was impossible to get a picture of the funnel from the database. A significant share of deals remained at the first stage regardless of subsequent work with the customer. Calls and tasks could only be found inside individual deal records.
At a meeting with our head of sales and business development director, it was described like this: “A deal comes in — and that’s it. To see what’s happening with it, you have to open it up.”
The tool had been bought and set up. Yet the summary report placed customers in different situations at the same stage. Looking at this picture, it is easy to conclude that salespeople cannot move deals forward. But first, you need to find out what is standing still: the work with the customer or the record of it.
A quick reference before discussing conversion
These four points can be forwarded to the head of sales before reviewing the report. The owner needs answers backed by evidence, rather than having to go through every record themselves.
| Question about the report | What the answer should include | If this is missing |
|---|---|---|
| What confirms the stage? | An event in the deal and its date | Establish the customer’s actual situation first |
| How many deals have an unknown state? | A separate count within the selected group | Do not treat the overall percentage as exact |
| Are we counting the same customers? | A single group of deals and the date as of which we assess the result | Recalculate transitions without mixing different streams of customers |
| When did the transition happen? | The event date, separate from the date the record was corrected | Do not present tidying up CRM as faster sales |
The order matters here. There is no point arguing about whether conversion is good when the head of sales has not yet established what went into the numerator. And you cannot compare salespeople if one records a transition straight away while another keeps the same information only in notes.
The same status hides different tasks
A deal at the first stage can be in at least three different states. The customer really has not moved forward yet. The customer has moved forward, but the salesperson has not changed the stage. Or the saved information does not tell you what happened at all.
In the first case, the head of sales examines the work with the customer: what is preventing the next action, who will take it and what deadline has been agreed. In the second, they bring CRM back into line with the actual situation and find out why the record fell behind. In the third, they recover the information through the salesperson responsible. Sending everyone the same instruction to “push harder to close” makes no sense here: the tasks are different.
The basis for a transition must be the same for the whole team. A step confirmed by the customer, a call recording or correspondence helps explain why a deal is at that stage. A completed call alone does not explain this: the conversation happened, but the buyer’s decision may not have changed. The head of sales chooses criteria that reflect the actual sales process and makes sure they are applied consistently.
The owner receives the number of discrepancies and their causes from the head of sales. If every answer requires calling in a salesperson and hearing the story again, the report has not yet replaced a round of verbal updates. It is too early to buy extra charts for it: they will use the same statuses.
Unknowns must stay in the calculation
The percentage is calculated like this: divide the number of deals that have reached the next stage by the number of deals in the selected group and multiply by one hundred. Both parts must refer to the same customers. You cannot take today’s enquiries and divide the transitions of customers who came in earlier by that number.
But before calculating, show how much of the group is unknown. For example, out of 100 deals, a transition is confirmed for 40, it has been established that no transition has happened yet for 40, and the state of 20 has not been determined. The confirmed proportion of deals that have moved forward is 40 divided by 100, or 40%. The unknown proportion is 20 divided by 100, or 20%.
If all 20 unclear deals have already moved forward, the actual proportion as of that date will be 60%. If none have, it will remain 40%. The owner sees a range from 40 to 60% and understands how much information is missing. You cannot remove the unknowns from the calculation: 40 divided by the remaining 80 gives a neat 50%, but for a different, incomplete group.
This is the proportion that has moved forward as of the report date. Customers who remain at the previous stage do not automatically count as having declined. Refusals, pending decisions and unknown states require different actions. While the head of sales gathers the missing information, confirmed problems in individual deals can already be examined; there is no reason to wait for a perfect database before taking any action.
A day of tidying up is not a day of growth
After deal records are corrected in bulk, the funnel will shift. The question is when the customers moved forward. If a buyer passed a stage earlier but the salesperson recorded it today, an old event will appear in today’s database changes.
The head of sales therefore separates the event date from the date the entry was made. If the original date cannot be recovered, it remains unknown. Making up a date in the past for the sake of a smooth chart means spoiling the next review: how quickly a deal progressed will seem precisely measured, even though it was not measured at all.
For the owner, this is a criterion for accepting the changes. After tidying up, the head of sales must show which transitions were recovered from the past and which happened under the new rules. Speed and conversion before and after can be compared using periods with comparable record-keeping. A single jump in a chart does not replace that comparison.
RENTROP’s AI ROP product analyses calls and correspondence alongside the deal record — this approach to sales department analytics helps link the status to what is happening with the customer. The service only reads data; the head of sales must correct records and ensure the rules are followed.
At that meeting, the team suggested starting with rules for keeping deal records. The finding was that the analytics they had paid for required a manual account of events: what was happening with the customer was inside the deal record, but was not reflected in the stage. The owner needs a report that the head of sales can explain using the shared data. If this means gathering the salespeople’s stories all over again every time, buying analytics has not yet led to putting it into use in the department.