I have previously demonstrated how a simple tweak in a sales process, such as replacing a few meetings early in the cycle with asynchronous content sharing, can result in double-digit growth in bookings.

That exercise remains largely theoretical until we know what the actual sales process looks like. As Peter Drucker once noted: “If you can’t measure, you can’t improve.”

In the earlier write-up, we modeled B2B sales processes as sequences of meetings:

\[M_1 \rightarrow M_2 \rightarrow \dots \rightarrow M_n\]

The following information was of particular interest:

  • The average number of meetings per opportunity
  • The meeting conversion rate
  • The average duration of the sales process

Armed with this information, one can design a strategy for increasing the process win rate and shortening the sales cycle. Due to exponential dependencies, even a small tweak can have a dramatic positive impact.

The problem is that most organizations do not have this information readily available. When they do have it, it tends to be unreliable. Fortunately, with a little bit of data science, getting reliable estimates can be relatively straightforward.

To illustrate the approach, I will use a single data source – the log of the sales team’s Zoom meetings. Zoom provides an API and a dashboard for retrieving the log. It can be massaged to look as follows:

Date Internal Participants External Participants
D1 IP1, IP2 EP1, EP2
D2 IP1, IP3 EP3
D3 IP2, IP4 EP2, EP4
D4 IP3, IP5 EP3, EP5

The first step is to group meetings by similarity. One might notice, for instance, that the first and the third meetings are similar because they have overlapping lists of participants.

There are powerful data science methods that can do this kind of analysis automatically and produce a table that looks like this:

Opportunity Sequence # Date Internal Participants External Participants
O1 1 D1 IP1, IP2 EP1, EP2
O1 2 D3 IP2, IP4 EP2, EP4
O2 1 D2 IP1, IP3 EP3
O2 2 D4 IP3, IP5 EP3, EP5

We now have a new column that identifies meetings as belonging to this or that opportunity. We have also sorted the entries by opportunity and by date and numbered the meetings.

The rest is pure arithmetic.

We can measure the time between the first and the second meeting in the process as the average of (D4-D2) and (D3-D1).

The conversion rate of the first step in the process can be calculated as

\[CR_1 = \dfrac{\text{Number of Second Meetings}}{\text{Number of First Meetings}}\]

The length of the sales cycle can be calculated as the average difference between the dates of the first and the last meetings.

There is a lot more information that can be extracted from the meeting log. For instance, one can notice that participants IP2 and IP3 play pivotal roles in opportunities O1 and O2 and that they are collaborating with participants IP4 and IP5 respectively. If IP2 closes the deal, IP4 must get points for the assist.

We can refine our analysis by adding data from other sources.

For example, opportunities in the meeting log can be mapped to opportunities recorded in Salesforce. By examining the outcomes of opportunities that are marked as closed, we can start making predictions about opportunities that are still in progress. We can also begin making recommendations for actions that lead to successful outcomes.

In summary, if you can map and measure the actual sales processes, you can implement a few minor improvements that deliver tremendous impact.

The effort that is required to collect the data and perform the analysis is often relatively straightforward. Every data scientist who is worth their salt must be able to assist.

If your data science team is too busy or unavailable, we must be able to help. Check out Morebell Revenue Intelligence and drop us a line.