Retention Analytics · D2C Jewelry · Case Study
By identifying which entry products drive 4x higher lifetime value and restructuring acquisition strategy around that single insight.
The brand had grown to $15M in annual revenue on aggressive paid acquisition across Meta and Google. Revenue was climbing but so was CAC, and leadership had a nagging suspicion: too many customers were buying once and never coming back.
CAC climbed quarter over quarter as the brand scaled spend, with no corresponding improvement in retention or LTV to offset the rising costs.
Only 14 in 100 customers made a second purchase within 90 days, well below category benchmarks for jewelry brands at their revenue tier.
Dashboards showed healthy top-line growth but couldn't distinguish high-LTV customers from one-time buyers, making every acquisition decision a guess.
Customer data lived in silos: Shopify for transactions, Meta Ads for acquisition, a separate email platform for CRM, none connected at the customer level.
The Core Question Leadership Could Not Answer
"Are we acquiring customers who will come back, or paying to fill a leaky bucket?" Without connecting acquisition data to long-term purchase behaviour, there was no way to know. The business was optimising for volume at the expense of value.
Before any analysis was possible, four fragmented data sources needed to be unified into a single customer-level record, connecting every transaction to its acquisition source and mapping the full purchase journey over time.
| Platform | Data Contributed | Key Fields |
|---|---|---|
| Shopify | Transaction history | CustomerID, product category, order date, revenue, discount used |
| Meta Ads | Paid acquisition | Ad creative, audience type, click date, campaign, spend attributed |
| Google Ads | Paid acquisition | Keyword, campaign type, click date, spend, attributed revenue |
| Email Platform | CRM & engagement | Open rate, click rate, campaign dates, unsubscribe signals |
Early Finding
Retained customers generated 65% of total revenue despite being just over half the customer base. The LTV chart reveals stark differences by first purchase category, completely invisible in blended revenue reporting.
03 — The Critical Finding
Despite near-identical acquisition costs across both segments. The brand had been optimising for volume, not value. The product that built loyalty was already in the catalogue.
High-LTV Entry Product ✶
Silver Stacking Ring
$680
Average 12-Month LTV
Low-LTV Entry Product
Statement Piece
$170
Average 12-Month LTV
Four analytical layers were applied sequentially. Each answered a specific question and built the foundation for the next. Click each step to expand the detail.
Shopify transaction records, Meta Ads attribution data, Google Ads click data, and email engagement logs were merged at the customer level using email address as the primary key. Each customer record now carried their full history: first acquisition source, first product purchased, subsequent purchase dates, discount usage, and total spend.
For the first time, leadership could see a single customer's full journey from first ad click to most recent purchase in one place. This made cohort analysis and LTV modelling possible.
All customers were segmented by the category of their first purchase. 12-month LTV was calculated per segment, defined as total revenue per customer from first purchase date through 12 months. First product category proved a far stronger predictor of long-term value than acquisition channel, demographics, or discount usage.
Silver stacking ring buyers had a 12-month LTV of $680 on average. Statement piece buyers averaged $170. Charm bracelet customers were the second-highest LTV group at $410.
A churn prediction model was built using three behavioural signals: days since last purchase (recency), total number of site visits (frequency), and whether the customer had used a discount code. Customers in the top quartile of churn risk were flagged for proactive re-engagement before their window closed.
18% of at-risk customers who received targeted re-engagement sequences made a subsequent purchase within 30 days, customers who would otherwise have lapsed silently.
High-LTV customer profiles were traced back to acquisition sources to identify which creative angles and audiences were generating stacking ring buyers. Certain Meta creatives, specifically lifestyle imagery showing ring stacking, disproportionately attracted the highest-LTV customers but were under-funded relative to statement piece creatives.
Budget was reallocated toward creative angles correlated with high-LTV first purchases. CAC initially spiked slightly before dropping 23% within 90 days as the algorithm re-optimised toward higher-quality signals.
Results measured across three dimensions: acquisition efficiency, retention performance, and revenue impact. All metrics captured over a 90-day post-implementation window.
By reallocating spend toward audience signals associated with high-LTV buyers, CAC fell without any reduction in total ad budget.
Stacking ring buyers showed 4x higher 12-month LTV than statement piece buyers, the core insight driving every subsequent decision.
The repeat purchase rate nearly doubled, from acquiring higher-quality customers and re-engaging at-risk customers proactively.
Of customers flagged as high churn-risk, 18% made a subsequent purchase within 30 days of targeted re-engagement, before lapsing silently.
Total Revenue Impact
Combining improved repeat rate, reduced CAC, and at-risk customer recovery, calculated against existing traffic with no increase in total paid spend, the strategy delivered measurable, sustained revenue growth.
Additional Annual Revenue
from Existing Traffic
The same fragmentation problem that made this analysis necessary becomes significantly worse as brands expand across DTC, marketplaces, and retail. Cross-channel growth without unified data creates blind spots that compound at scale.
When the same customer buys on your DTC site, Amazon, and a retail partner, those transactions need to be connected. Without it, you optimise three siloed revenue streams instead of one customer relationship.
The products that perform well on marketplaces are not always those that build DTC loyalty. Understanding LTV by first product per channel changes which SKUs you promote where.
Not all paid channels deliver the same quality customer. An LTV-adjusted view often reveals a higher-CAC channel is cheaper per retained customer, a finding that changes budget allocation entirely.
At scale, churn is nearly invisible until too late. Models built on recency, frequency, and discount behaviour give a 30 to 60 day window to intervene before a customer is lost permanently.
A retention audit typically identifies the highest-leverage opportunity within one week, no commitment required. The insights above came from data the brand already had.
Jeremiah Ekpo-otu · ekpootujery@gmail.com