How a DTC Skincare Brand Increased AOV 31% With Market Basket Analysis on Shopify

Table of Contents

• The Situation: A $12M DTC Skincare Brand Hitting a Revenue Ceiling
• The Data Problem They Didn't Know They Had
• Running Market Basket Analysis on 14 Months of Shopify Data
• Three Findings That Changed Their Bundling Strategy
• Implementation: From Insights to Revenue in 90 Days
• The Results: 31% AOV Increase and Beyond
• Key Takeaways for DTC Brands at $5M–50M
• FAQ

The Situation: A $12M DTC Skincare Brand Hitting a Revenue Ceiling

In early 2025, a DTC skincare brand — let’s call them GlowBase — was doing $12M in annual GMV on Shopify. Their catalog had grown to around 85 SKUs across cleansers, serums, moisturizers, SPFs, and treatment products. The team of six handled everything from growth marketing to retention, and their Head of E-commerce was under pressure from investors to improve unit economics without dramatically increasing ad spend.

The numbers told a familiar story for mid-market DTC brands. Their CAC had climbed 28% year-over-year, largely driven by rising Meta CPMs. Meanwhile, their average order value had plateaued at $67 for three consecutive quarters. LTV:CAC was hovering around 2.8:1 — acceptable, but not the 3.5:1+ their investors wanted to see before their Series B.

GlowBase had tried the obvious moves. They’d launched a few bundles based on gut feeling and merchandising logic: a “starter kit” with cleanser + moisturizer + SPF, a “glow routine” set, and a seasonal gift box. These bundles sold okay — about 12% of orders included a bundle — but they weren’t moving the AOV needle in a meaningful way.

The core problem was that GlowBase was building bundles based on what their merchandising team thought customers wanted to buy together, not what customers were actually buying together. That gap between assumption and reality was costing them real revenue.

The Data Problem They Didn’t Know They Had

GlowBase’s analytics setup was typical for a brand at their stage. They had Shopify’s built-in analytics, Google Analytics 4 (partially configured), a Klaviyo integration for email, and Triple Whale for attribution. Their Head of E-commerce could tell you top-line metrics — revenue, conversion rate, returning customer rate — but couldn’t answer the question that actually matters for bundling: which products do customers buy together, and in what sequence?

Shopify’s product analytics shows you which products sell the most. It doesn’t show you which products are purchased in the same cart, which products are purchased sequentially across orders, or which product combinations drive the highest average order values. That’s a completely different analytical question — and it requires market basket analysis (MBA) to answer it.

The team had actually looked into building their own analysis. Their growth marketer had some SQL knowledge and attempted to export Shopify order data and run co-occurrence calculations in a Google Sheet. After three days of work, they had a messy spreadsheet that showed the top 20 product pairs by frequency — but no confidence metrics, no lift calculations, no sequential purchase patterns, and no way to distinguish between products that genuinely drive each other’s purchases versus products that simply sell a lot individually.

This is the analytics dead zone that mid-market DTC brands consistently fall into. Your Shopify reports tell you what sold. Your attribution tools tell you where customers came from. But neither tells you how products relate to each other in the customer’s mind — the insight that actually drives bundle strategy, cross-sell placement, and post-purchase flows.

Running Market Basket Analysis on 14 Months of Shopify Data

GlowBase connected their Shopify store to Affinsy and ran a market basket analysis on 14 months of transaction data — roughly 168,000 orders across their full catalog of 85 SKUs. The analysis completed in under five minutes and returned three types of insights their existing tools couldn’t provide.

Association rules with confidence and lift metrics. Rather than just showing which products appeared together most often (raw frequency), the MBA calculated support (how often a combination occurs), confidence (the probability of buying Product B given that a customer bought Product A), and lift (whether the co-purchase rate is higher than random chance). These statistical measures separate genuine purchase relationships from coincidental co-occurrence.

Sequential purchase patterns. Beyond same-cart analysis, the MBA identified which products customers purchased on their second and third orders. This revealed the natural “product journey” that GlowBase’s most valuable customers followed — critical data for designing post-purchase cross-sell sequences in Klaviyo.

Basket-level AOV analysis. The tool segmented orders by which product combinations they contained and calculated average order value for each basket type. This showed which combinations actually drove higher spend — not just which ones occurred frequently.

Three Findings That Changed Their Bundling Strategy

Finding 1: Their best-selling bundle was their worst-performing combination

GlowBase’s flagship “Starter Kit” — cleanser + moisturizer + SPF — was their most purchased bundle by volume. But the MBA revealed a problem: this combination had a lift of just 1.1x, meaning customers who bought the cleanser were only 10% more likely to add the moisturizer and SPF than random chance would predict. In practical terms, most customers buying this bundle would have purchased those products anyway. The bundle discount was cannibalizing margin without driving incremental revenue.

Meanwhile, a combination the team had never considered bundling — their Vitamin C serum + their niacinamide toner + their hyaluronic acid moisturizer — had a lift of 3.4x and a confidence of 62%. Customers who bought the Vitamin C serum were more than three times as likely to add the toner and moisturizer as chance would predict. This was a genuine purchase driver hiding in the data.

Finding 2: Post-purchase cross-sell timing was off by two weeks

The sequential analysis revealed that GlowBase’s most valuable customers followed a predictable product journey. First purchase: typically a single hero product (Vitamin C serum or retinol treatment). Second purchase: 18–24 days later, adding a complementary product. Third purchase: 35–42 days later, expanding into a full routine.

GlowBase’s existing Klaviyo post-purchase flow sent a cross-sell email 7 days after first purchase. The data showed this was too early — customers weren’t ready to expand their routine yet. The sweet spot for cross-sell conversion was days 16–22, when customers had used the first product long enough to see results and were primed to add a complementary step.

Finding 3: A hidden “gateway product” was driving their highest-LTV customers

The MBA identified something the team had completely overlooked: their $28 exfoliating toner — a mid-priced product that wasn’t a top seller by volume — was the most common first purchase among customers who went on to make 4+ orders. Customers who entered through the toner had an average LTV of $340, compared to $180 for customers who entered through their best-selling Vitamin C serum.

This didn’t mean the toner was a better product. It meant the toner attracted a specific customer profile — experienced skincare users who already had a routine and were looking to add to it. These customers were pre-qualified for higher spend and longer retention. This insight completely changed GlowBase’s acquisition strategy for high-LTV customers.

Implementation: From Insights to Revenue in 90 Days

GlowBase took the MBA findings and implemented changes across three areas over 90 days. Here’s what they did and how each change impacted revenue.

Phase 1 (Days 1–30): Bundle redesign

The team retired their three existing bundles and launched four new ones, each based on high-lift product associations from the MBA. The key change was pricing strategy: instead of offering a flat percentage discount on bundles, they priced new bundles at a 12–15% discount (compared to the previous 20–25%) because the data showed these products had genuine co-purchase affinity. Customers didn’t need a steep discount to buy them together — they already wanted to.

They also restructured their Shopify product pages to show “Frequently Bought Together” recommendations based on actual MBA data rather than Shopify’s default algorithm. The top three cross-sell recommendations on each product page now reflected the highest-lift associations for that specific product.

Phase 2 (Days 30–60): Post-purchase flow optimization

GlowBase rebuilt their Klaviyo post-purchase email sequence using the sequential purchase data. The new flow:

• Day 3: Usage tips for the purchased product (no cross-sell)
• Day 14: “What’s working” check-in with a soft mention of complementary products
• Day 18: Primary cross-sell email featuring the highest-confidence complementary product from the MBA data
• Day 22: Follow-up cross-sell with a time-limited 10% incentive
• Day 35: “Complete your routine” email featuring the full bundle

Each email was dynamically personalized based on the customer’s first purchase, pulling the specific cross-sell recommendations from the MBA association rules.

Phase 3 (Days 60–90): Acquisition targeting

Using the gateway product insight, GlowBase created a separate Meta ad campaign specifically promoting the exfoliating toner to experienced skincare audiences. They used interest targeting focused on skincare-savvy demographics (ages 28–45, interests in retinol, AHAs/BHAs, multi-step routines) rather than broad DTC beauty audiences. The CPA for this campaign was 15% higher than their general acquisition campaigns, but the 90-day LTV of acquired customers was 89% higher — a dramatically better return on ad spend.

The Results: 31% AOV Increase and Beyond

After 90 days of implementation, GlowBase measured the impact across their key metrics:

The AOV increase alone added approximately $2.9M in annualized revenue at their existing traffic levels — without spending a single additional dollar on acquisition. The LTV:CAC improvement pushed them past the 3.5:1 threshold their investors required, directly supporting their Series B timeline.

But the most surprising result was the margin improvement. By reducing bundle discounts from 20–25% to 12–15% while simultaneously increasing bundle attach rates, GlowBase improved their blended gross margin by 4 percentage points across the entire business. The lesson: when you bundle products that customers genuinely want together, you don’t need to discount as aggressively to drive adoption.

Key Takeaways for DTC Brands at $5M–$50M

GlowBase’s story isn’t unique. The patterns they discovered — gut-feel bundles underperforming data-driven ones, cross-sell timing being off, hidden gateway products — show up consistently across mid-market DTC brands that run their first market basket analysis. Here’s what you can take away regardless of your vertical.

Your best-selling bundle probably isn’t your best bundle. High-frequency product combinations don’t necessarily have high lift. If two products both sell well individually, they’ll appear together in carts often — but that doesn’t mean they drive each other’s purchases. Run an MBA to find the high-lift combinations that represent genuine purchase affinity, then build bundles around those.

Cross-sell timing matters more than cross-sell content. GlowBase’s cross-sell emails had the same products before and after optimization. The difference was when they sent them. Sequential purchase analysis tells you the natural cadence of your customers’ buying journey — match your email timing to that cadence instead of guessing.

Gateway products aren’t always your hero products. The product that drives the most first-time purchases isn’t necessarily the product that attracts your highest-LTV customers. MBA can identify which entry points lead to the longest, most valuable customer relationships — information that should directly inform your acquisition strategy and budget allocation.

Smaller discounts work when product affinity is real. GlowBase cut their bundle discount nearly in half and saw bundle attach rates double. When you bundle products with genuine co-purchase affinity (high lift), customers don’t need a steep discount to buy them together. The discount becomes a nudge, not a bribe.

This analysis pays for itself in a single day’s revenue. GlowBase spent less than an hour setting up and reviewing their MBA results. The resulting bundle and cross-sell changes generated an incremental $8,000+ in daily revenue. For mid-market brands, the ROI on this type of analysis is measured in hours, not months.

FAQ

How much order data do I need to run a useful market basket analysis?

For statistically meaningful results, you generally need at least 5,000–10,000 orders. GlowBase had 168,000 orders across 14 months, which gave very high confidence in the results. If you’re a smaller brand with 1,000–5,000 orders, an MBA can still surface directional insights — you’ll just want to validate the top findings with a few weeks of A/B testing before making major changes. Most DTC brands on Shopify at $5M+ GMV have more than enough data.

Can I run market basket analysis on Shopify without a data scientist?

Yes. That’s precisely the gap tools like Affinsy are built to fill. You connect your Shopify store, the analysis runs automatically on your order data, and the results come back in a visual dashboard with plain-language explanations of each association rule. GlowBase’s Head of E-commerce ran the analysis herself in under five minutes with no SQL or statistics background.

How is market basket analysis different from Shopify’s product recommendations?

Shopify’s built-in product recommendations use a simplified co-occurrence algorithm — essentially “customers also bought.” MBA goes further by calculating statistical measures like confidence and lift that separate genuine purchase drivers from coincidental co-occurrence. It also analyzes sequential purchases across orders, not just within a single cart. The difference is the quality and depth of insight: MBA tells you why products sell together and how strongly the relationship is, not just that they appeared in the same carts.

What ROI can a mid-market DTC brand realistically expect from MBA?

Based on anonymized data across DTC brands in the $5M–$50M range, MBA-driven bundle and cross-sell optimizations typically produce a 15–35% increase in AOV within the first 90 days. The variance depends on your current bundle strategy (brands with no existing bundles see the biggest lifts) and catalog complexity (more SKUs generally means more hidden opportunities). The investment to get started is minimal — most brands can go from data to actionable insights in under an hour.

Does this work for categories beyond skincare?

Absolutely. MBA is category-agnostic. The statistical principles apply whether you sell skincare, supplements, pet food, home goods, or apparel. Any DTC brand with 50+ SKUs and repeat purchase behavior will find actionable product associations in their data. Categories with natural “routine” or “system” purchasing behavior (skincare, supplements, fitness, pet care) tend to show the strongest associations, but we’ve seen significant results across all DTC verticals.

Unlock Hidden Revenue in Your Shopify Data

GlowBase’s story illustrates a reality most mid-market DTC brands are living but haven’t quantified: your order data contains product relationships that can drive meaningful revenue growth — if you have the right tools to surface them. Market basket analysis turns transaction history into a strategic bundling, cross-sell, and acquisition playbook.

If you’re running a DTC brand on Shopify at $5M–$50M in GMV and your bundling strategy is based on merchandising intuition rather than statistical analysis, you’re leaving revenue on the table. The fix takes less than an hour.

Try Affinsy with your data and see which product associations are hiding in your Shopify orders. Your first analysis is free.