Collaboration and Productivity

While collaboration can lead to new products, processes, and innovative ideas, too much of it may become counterproductive. A group of analysts studied quantitative measures of collaboration, such as the number of emails and meetings in order to better understand their impact on the productivity of a sales team.

An automated hardware manufacturer’s sales team of over 200 people had shown suboptimal performance. To increase sales and productivity, the company reached out to its people analytics team for assistance.

Purpose: Identify patterns and recommend actionable solutions based on data.

Photo by Mille Sanders on Unsplash

Step 1 — Business Question:
Why are attainment results uncharacteristically uneven across the sales department? Any solutions to increase sales quota attainment?

Description of Data Set:

Figure 1: Data dictionary Part 1

Figure 2: Data dictionary Part 2

Our analysis will focus primarily on the Sales Department, where we obtained the data on quota attainment. To begin, we examine the architecture of the sales team…

Frequencies of Employment Level in Sales Department:

ggplot(df_sales, aes(factor(level),
        fill = factor(level))) +
    geom_bar()

Figure 3: Levels in Sales Department, and relative # of employees

Most of the sales department is made up of “Junior IC” level employees, followed by “Support” and “Senior IC” employees.

Next, we look at the median quota results (% of attainment), by level of employee. The box plots display the minimum, maximum, median, and first and third quartiles.

Figure 4: Level of employees and attained quotas

The Senior Executives, Directors and Executives have been the top performers so far. The Junior IC level employees and Managers could benefit from learning relevant sales tactics that work from the Directors or Support teams to increase quota attainment. By focusing some of their efforts on improving their Junior IC employees and Managers, executives can help boost quota attainment while also driving employee growth.

Step 2 — Correlations Between Variables in Data Set:

Figure 5: Measuring the strength of various correlations between variables

Highest Correlating Variables in data set:
Centrality and emailcount— 0.70
Centrality and emailcount_int — 0.83
Emailcount and emailcount_int — 0.89
Emailcount and emailcount_ext — 0.87
Meetingcount and centrality — 0.59
Meetingcount and meetingcount_ext — 0.93
Meetingcount_int and meetingcount_ext — 0.89

Insights: These three factors — meetings, emails and centrality are all highly correlated with one another.

Mild Correlation:
Attainquota and emailcount — 0.28
Attainquota and meetingcount — 0.36
Attainquota and centrality — 0.35
Attainquota and meetingcount_int — 0.37

Insights: In order to understand the effect of collaboration on attainment % of quota, we look into email count, meeting count and centrality.

Part 3 — Visual Analysis: Histograms and scatterplots to explore the hypothesis and data categories/points

Get Lisa’s stories in your inbox

Join Medium for free to get updates from this writer.

Remember me for faster sign in

Histogram of Employees and their Sales Quota % attainment:

Figure 6: Histogram of sales quota attainment

Email Count and Attain Quota % (color coded by level of employee):

Figure 7: Attain quota % and email count, color coded by level

Based on our data, it appears that the relationship between email frequency and the quota attained is non-linear. When more emails were sent, there was a point at which the quota attained began to decrease. This suggests that collaborating through email can be an effective strategy up until a point, however excessive amounts of email could ultimately hinder productivity and progress.

Part 4 — Data Subsets — Non-Linear Relationships

We create data subsets to more accurately depict non-linear relationships.

Attain Quota % and Email Count, with Data Subsets:

ggplot(data=model_1, aes(x = emailcount, y = attainquota)) +
  geom_point(col = "blue") +
  geom_smooth(method = "lm", formula = y ~ x, data=subset(df_sales, emailcount<7500),
    se = FALSE, col = "coral") +
  geom_smooth(method = "lm", formula = y ~ x, data=subset(df_sales, emailcount>=7500),
    se = FALSE, col = "darkgreen") +
  theme_classic()

Figure 8: Attain quota % and email count, subsets of collaboration effects

Centrality and Attained Quota:

Centrality (the measure of how central the employee is to the flow of information within the organization) produces a similar result. This data too, was separated into subsets, to more accurately depict the non-linear relationship between quota attained and network centrality.

ggplot(data=model_1, aes(x = centrality, y = attainquota)) +
  geom_point(col = "blue") +
  geom_smooth(method = "lm", formula = y ~ x, data=subset(df_sales, centrality<0.30),
    se = FALSE, col = "darkgreen") +
  geom_smooth(method = "lm", formula = y ~ x, data=subset(df_sales, centrality>=0.30),
    se = FALSE, col = "purple") +
  theme_classic()

Figure 9: Attain quota and centrality, subsets of collaboration effects

This clearly depicts the negative effects of quota attainment with increasing centrality within the organization. (Higher centrality means employee is more central to data flows of information with networks.)

Then we included centrality squared in the data set, as a quadratic relationship, to achieve a better model fit.

Attain Quota OLS Linear Regression Model:

Figure 10: Regression analysis of effects on attained quota %

Interpretations on the coefficients:
Centrality — significant, the likelihood of attaining quota goes up.
I(centrality²) — significant, the likelihood of attaining quota goes down.
meetingcount — significant, the likelihood of attaining quota goes up.

Scatterplot of Interior Meeting Count and Attained Quota %:

Figure 11: Meeting count — interior and attained quota %

After 1,000 meetings, there seems to be a gradual decrease in productivity when achieving quotas. This is supported by the centrality and email count variables. (Note: A significant number of sales employees achieved their quotas without liaising with internal teams.)

Conclusion:
Collaboration can increase productivity, but up to a certain point. Too much collaboration can lead to lost time and result in decreased productivity.

Based on the findings of this analysis, there are a number of solutions that could be recommended to improve quota attainment. Improving individual sales processes through more structured sales protocols is one such solution.
Further analysis into high-performing employees could provide insight into why their sales techniques produce positive results. The team can utilise data analytics to determine which customers are more likely to purchase, based on historical data, and allocate more time towards those clients. This information can then be shared with all other employees within the sales department, as one potential way to improve quota attainment and increase sales performance overall.
To conduct further analysis, additional data would be required regarding quota attainment across departments beyond just the sales team. Conducting an additional analysis would allow us to examine how centrality, email and meeting count affects non-sales teams as well as their ability to achieve objectives.

Sources: Polzer, Jeff T., et al., (2018). Sarah Powers at Automated Precision Products, Harvard Business Review.