Customer Segmentation & LTV Modeling

Text

Situation:
"At GoOffer, marketing was splitting budget pretty evenly across LinkedIn, Instagram, and SEO — but nobody had actually checked whether those users were equally valuable once they signed up."

Task:
"So I built a customer segmentation and LTV analysis to figure out which channels were really bringing in profitable, long-term users."

Action:
"I started in SQL, joining users and payments data to calculate LTV, purchase frequency, and recency. Based on that, I created initial business segments — High-Value, Mid-Value, and At-Risk — using LTV thresholds.

Then in Python, I ran RFM analysis and k-means clustering on the same data to check if those segments actually held up statistically. The clustering shifted some of the boundaries — part of what I'd originally called Mid-Value behaved more like At-Risk based on actual purchase patterns — so I adjusted the segments to match what the data showed, rather than sticking with my initial guess.

I also built a Tableau dashboard so marketing could track LTV and channel value on their own, without coming to me every time."

Result:
"The analysis showed LinkedIn and SEO users had about twice the LTV of Instagram users, who churned faster. High-Value users — only 20% of customers — were generating almost 60% of revenue.

For the At-Risk segment, we launched a targeted email re-engagement campaign to bring lapsed users back.

Marketing reallocated budget toward LinkedIn and SEO, and ran that email campaign for At-Risk users. End result: LTV went up 12% from the budget shift, marketing spend dropped 15% from cutting waste on low-value channels, and churn fell 10% — mainly from the email reactivation."

Problem

• Budget pretty evenly: SEO, LN, IG

• users equally valuable after signed up

Responsable

• Customer segmentation and LTV analysis

• Bringing in profitable, long term users

Actions

• SQL: Users+payment = LTV, purchase frequency, recency

• Initial business segments - using LTV threshold

• Python: RFM, k-means clustering - segments held up statistically

• Shifted some of the boundaries: Mid-Val behaved At-Risk - purchase patterns

• Adjust segments

• DB: track LTV, channel value on their own

Insights

• LN, SEO 2x higher LTV than IG (churned faster)

• High-Val 20% - 60% of revenue

• AT-Risk: targeted email re-engagement campaign

• Marketing reallocated budget, run campaign

Results

• LTV +12%: budget shift

• Marketing spend -15%: waste on low-value channel

• Churn -10%: email reactivation