Size (sample size, that is) matters in closing deals
At SmarterMe, we are obsessed in helping sales professionals to be extremely productive. We do this by bringing intelligent insights to the most frequently used sales tool : email. By gleaning information from email contents, relationship among participants, frequency and timing of the customer exchanges and their correlation between the won deals ( and the lost ones too ), SmarterMe delivers “IQ”s such as critical sales follow-up reminders. Upon receiving the “IQ”s, the salesperson can file them under various CRM systems such as Salesforce.com, SAP and Oracle, all within the same SmarterMe’s app. SmarterMe helps sales to work smarter and to close more deals while they are on the go.
Epiphany
By working with current customer’s data, our data scientist team noticed that the number of follow up email exchanges at a specific stage of the sales cycle has significant correlation towards both closed and failed deals. By first separating the exchanges between the WON and the LOST deals and then plotting the average number of exchanges (number of exchanges for the WON deals / total number of WON deals and number of exchanges for the LOST deals / total number of LOST deals), we discovered there is a surge of exchanges right at around the 5th week of a 10-week sales cycle. And for the failed deals, this pattern takes place at the 8th week of the similar sales cycle.
Hypothesis
Based on such observation, we ask what if we can remind the user to follow up with customers right before the inflection point at which the deal would take off or fail. And if we can help recommend whom to follow up with and suggest the type of follow-up materials that are appropriate for the particular stage of the sales, wouldn’t that be great?
This could potentially save deals that would have tanked!
Questions remained
- Will the same inflection pattern emerge for the larger user population?
- If so, what exactly is the inflection number? Will it be the same for users coming from various companies and industries?
A large sample size comes to the rescue
To answer these questions, we first need to consider the inflection number as a random variable RV that exhibits a normal distribution with a mean μ and a variance σ^2, N(μ,σ^2). According to the central limit theorem, if an RV is a combination of a large number of independent RVs, then its distribution is also an approximately normal distribution with a certain mean and variance.
Hence, given a large enough sample size ( > 500 observations ) of deals with its corresponding inflection point ( time of the sales cycle when the surge of emails takes place / total time of the sales cycle ), the average inflection point will also have a normal distribution with a certain mean and variance.
As a result, a pattern for the inflection point emerge. Nevertheless, this pattern is being described by a normal distribution.
Are we confident?
“How does this help find the potential inflection point for any ongoing deals?” one might ask. We can estimate the true inflection point by examining the mean and the variance of the sample data we have collected. We do this by calculating the 95% confidence interval. Assuming the sample mean is μ and the sample standard deviation is σ ( square root of variance ), then the estimate will fall under the following normal distribution
where n is the sample size, the new mean is μ/n and the new variance is σ^2/n
The 95% confidence interval means that there is a 95% chance that the true inflection point will be found within the range
Notice that if we increase the sample size n, a much narrower range of confidence interval is yielded, hence a much precise inflection point can be selected. This number can be used to recommend the user to perform the follow up
Data matters
Many factors can lead to the surge of exchanges — desperation, new found partners, new discoveries. However, instead of identifying the various causes, we are relying upon data and patterns. By expanding the initial observations to a much larger data set, SmarterMe can reliably determine a best time for the follow-up reminders. Sales people are busy professionals, and they all can use the extra help in closing deals.
