Testing Marketing Hypotheses at WSES

CASE FACTS:

● We Sell Everything in Software(WSES )Inc., sold innovative off-the-shelf products and

had customers across the world. WSES specialized in providing software solutions for
different industries.
● Jack Williams cofounded WSES in 2005 and took charge as CEO in 2015 and was
instrumental in driving data-driven decision making.
● WSES was primarily a B2B (Business to Business) company, the sales cycle took
anywhere between 3 months to 1 year once the lead was generated by the salesforce.
● Ben Osborne, Vice President-Marketing at WSES.
● Jack Williams and his friend Prudy Perkins started with the creation of a services
product Procsys and in a decade, added an array of products to meet needs across
multiple industries
● Liz Smith who leads the data science team.

ISSUES:

● The “gut feel” base approach used by the sales team to pursue a lead was not the most
methodical.
● salesforce most often depended on its gut feel to allocate marketing expenses to an
opportunity.
● With increased competition, there was a pricing war, which naturally led to the loss of
customers to competitors.
● we are too focused on the UK and other European countries where the competition is
much higher.

PROBLEM STATEMENT:
How do I use data-driven decision-making to prevent money drain on non-potential clients and
markets, and regain the market share through strategic data-driven decision-making.

ASSUMPTIONS:

● The chance of winning a lead is not the same for different products.
● The chance of winning a deal in Africa is better than in the United Kingdom.
● Average size of a deal is at least 8 million USD
● Average size of a deal in different geographical locations is not the same.
● Winning chances are higher when the relative strength is higher.
● It is difficult to win leads from customers who are making low profit or making loses.

STATISTICAL METHODS TO BE USED FOR HYPOTHESIS TESTING:

● Two sample T-tes
● Chi-square test of independence.
HYPOTHESIS TESTING:

1. The chance of winning a lead is not the same for different products.
 ● Test: Chi-square test of independence with categorical variables ("product" and
"win/loss").
● Why: This test allows you to assess whether the observed association between product
and win/loss rate is statistically significant or could be due to chance.
2. The chance of winning a deal in Africa is better than in the United Kingdom.
● Test: Two-proportion z-test.
● Why: This test is suitable for comparing proportions (win rate) between two independent
groups (geographical locations).
3. Average size of a deal is at least 8 million USD.
● Test: One-sample t-test (assuming normally distributed deal sizes).
● Why: This test checks if the average deal size is statistically above the hypothesized
threshold (8 million USD) based on your sample data.
4. Average size of a deal in different geographical locations is not the same.
● Test: One-way ANOVA.
● Why: This test compares the means of deal size across multiple geographical locations,
allowing you to determine if there are significant differences.
5. Winning chances are higher when the relative strength is higher.
● Test: Logistic regression.
● Why: This test helps model the relationship between "relative strength" and the binary
outcome of "win/loss," while controlling for other potential confounding factors.
6. It is difficult to win leads from customers who are making low profit or making losses.
● Test: Chi-square test of independence with categorical variables ("customer profit" and
"win/loss").
● Why: This test helps assess if there is a significant association between customer
profitability and winning leads, although causation cannot be directly established with
this test alone.

CONCLUSION:

https://docs.google.com/spreadsheets/d/1O1YeCjLB-
_jPvUK1AAbmOLS4DCMp29abVHtZSazNmXQ/edit?usp=sharing