Key Takeaways
- A $12M/year DTC supplement brand on Shopify increased average order value (AOV) from $54 to $69 — a 28% lift — in 90 days using market basket analysis.
- Three product association rules drove the majority of the gain: identifying which SKUs were most frequently purchased together unlocked targeted bundle and cross-sell opportunities.
- The brand reduced blended customer acquisition cost (CAC) by 18% because higher AOV improved contribution margin, giving more room to bid on paid channels.
- No data science team was needed — the entire analysis was run and operationalized through Affinsy with their existing Shopify order data.
Brand Background and the Problem
Let's call them Apex Wellness — a DTC supplement brand selling protein powders, recovery stacks, and greens products direct-to-consumer via Shopify. At the time of this analysis, they were doing roughly $12M in annual GMV, with about 14,000 monthly orders and a catalog of 120 active SKUs.
Their growth curve had flattened. Meta CPMs had risen 34% year-over-year. Google Performance Max was pulling volume but at margins that made the CFO uncomfortable. They were acquiring customers at $38 blended CAC on an AOV of $54 — leaving a first-order contribution margin of just $8 after COGS and shipping. Repeat purchase rate was healthy at 41%, but the math on new customer acquisition was getting harder to defend.
The VP of Growth had a hypothesis: customers who bought the protein powder almost certainly needed something else — a shaker, a creatine stack, a greens powder — but the brand's cross-sell logic was based entirely on what the merchandising team thought made sense, not what customers were actually buying together.
That intuition was right. But the solution wasn't another A/B test on product page placement. It was running market basket analysis on 18 months of Shopify order history to surface statistically significant purchase associations — and then operationalizing those associations everywhere that mattered.
Why Market Basket Analysis — Not Just Gut Feel
Most mid-market DTC brands approach cross-sell the same way: the merchandising or marketing team picks "recommended" products based on category logic or what sold well last season. It's better than nothing, but it's leaving significant revenue on the table.
Market basket analysis (MBA) works differently. It mines your actual transaction data to identify which products customers buy together with statistically measurable frequency. The three core metrics are:
| Metric | What It Measures | Why It Matters for DTC |
|---|---|---|
| Support | How often two items appear in the same order | Filters out noise — only surface rules with real volume |
| Confidence | When item A is bought, how often is item B also bought? | Tells you how reliable a cross-sell recommendation is |
| Lift | How much more likely is this pairing vs. random chance? | Lift > 1 means there's a real association, not coincidence |
For Apex Wellness, the team connected their Shopify store to Affinsy and ran the analysis across 18 months of order data — roughly 250,000 orders. The minimum support threshold was set at 2% (meaning a pairing had to appear in at least 2% of all orders to be included), and minimum confidence at 25%.
The results were immediately actionable — and surprising.
Running the Analysis: What the Data Showed
After processing the 18-month Shopify dataset, Affinsy surfaced dozens of association rules. The team focused on the top-tier rules — those with the highest lift score and meaningful order volume. Three rules stood out immediately:
| Rule | Support | Confidence | Lift | Implication |
|---|---|---|---|---|
| Whey Protein → Creatine Monohydrate | 8.4% | 44% | 3.1x | 44% of protein buyers also buy creatine — nearly 3x the baseline rate |
| Greens Powder → Collagen Peptides | 5.1% | 38% | 2.7x | A strong "wellness stack" pairing the team had never promoted together |
| Pre-Workout → Electrolyte Mix | 6.3% | 51% | 4.2x | Highest lift in the dataset — a natural usage pairing customers were already making |
The pre-workout → electrolyte rule was the most striking. A lift of 4.2x means that a customer who buys pre-workout is 4.2 times more likely to also buy electrolytes than a random customer in the store. And yet the brand had never bundled or cross-sold these two products together in any meaningful way.
Beyond the top three rules, the analysis also surfaced a critical negative insight: the brand had been recommending its sleep supplement as a cross-sell on the protein page for over a year. The MBA data showed that pairing had a lift of 0.8 — meaning customers who bought protein were less likely than average to buy the sleep product. The recommendation was actively irrelevant to that cohort.
"We had been optimizing placements based on what we thought made sense as a brand. The data showed us what customers actually believed made sense as consumers — and those were very different things." — VP of Growth, Apex Wellness (anonymized)
From Insight to Revenue: How They Executed
Having the association rules is only half the battle. The real work is translating those rules into revenue across every customer touchpoint. Apex Wellness ran a focused 90-day execution sprint across four channels:
1. Post-Purchase Cross-Sell (Biggest Win)
They used Shopify's native post-purchase extension (through ReConvert) to display data-driven cross-sell offers immediately after checkout, before the thank-you page loaded. Instead of generic "you might also like" recommendations, they showed:
- Creatine Monohydrate to every customer who had just purchased Whey Protein
- Electrolyte Mix to every customer who had just purchased Pre-Workout
- Collagen Peptides to every Greens Powder buyer
The offers were one-click add-ons at a 10% discount. Conversion rate on these post-purchase offers averaged 19% — compared to an industry benchmark of 8–12% for generic post-purchase upsells. The specificity of the recommendation drove the conversion premium.
2. Cart Cross-Sell Tiles
They replaced the existing "frequently bought together" widget (which was pulling random Shopify recommendations) with MBA-driven pairings. The widget logic was simple: if the cart contained SKU A, show SKU B. AOV for orders that engaged with the cart cross-sell averaged $74 vs. $54 for orders that didn't.
3. Email Flows
The post-purchase email sequence was updated to include a targeted cross-sell email 7 days after the first purchase. Protein buyers received a creatine-focused email. Pre-workout buyers got an electrolyte email. These flows had a click-to-purchase rate of 6.1% — more than double the brand's standard promotional email performance of 2.8%.
4. Paid Retargeting Audiences
The brand exported cohorts of "protein buyers who have not yet purchased creatine" and "pre-workout buyers who have not yet purchased electrolytes" from their Shopify data and uploaded them as custom audiences on Meta. The retargeting creative was tailored to the specific stack. ROAS on these audiences was 4.1x vs. a blended retargeting ROAS of 2.6x.
90-Day Results
At the end of the 90-day sprint, Apex Wellness had materially moved the metrics that matter at their stage of growth:
| Metric | Before MBA | After 90 Days | Change |
|---|---|---|---|
| Average Order Value (AOV) | $54 | $69 | +28% |
| Blended CAC | $38 | $31 | -18% |
| First-Order Contribution Margin | $8 | $21 | +163% |
| Post-Purchase Upsell Conversion | ~9% (generic) | 19% (MBA-driven) | +111% |
| Retargeting ROAS (cross-sell cohorts) | 2.6x | 4.1x | +58% |
The first-order contribution margin improvement deserves special attention. Going from $8 to $21 per new customer acquired didn't just make the unit economics more comfortable — it fundamentally changed the brand's ability to compete on paid channels. At $21 contribution margin, they could afford to acquire a customer who would only purchase once and still be operationally profitable. That margin buffer allowed them to increase paid spend by 22% without any deterioration in blended returns.
Over the 90 days, the MBA-driven cross-sell program generated approximately $380,000 in incremental revenue that wasn't in their plan — purely from better utilization of their existing customer base and traffic.
4 Lessons Any DTC Brand Can Apply
1. Your category logic and your customers' purchase logic are not the same thing
Merchandising teams organize products by how they think about the catalog. Customers buy based on use case and outcome. The gap between those two mental models is where cross-sell revenue gets left on the table. MBA surfaces what customers are actually doing.
2. Lift score is your north star — not just "frequently bought together"
A high-support pairing (two products that appear in lots of orders together) is not necessarily a high-lift pairing. If those two products are both bestsellers, they'll appear together often by chance. Always filter for lift > 1.5 at minimum; prioritize rules with lift > 2.5 for your active cross-sell placements.
3. Negative associations are as valuable as positive ones
Finding out that two products have a lift below 1.0 tells you to stop recommending them together — and to free up that placement for something that will actually convert. Apex Wellness discovered this with their sleep supplement. Removing a bad recommendation is just as impactful as adding a good one.
4. Post-purchase is the highest-leverage placement — and most brands underinvest in it
At the moment of checkout, purchase intent is at its peak. A customer who just spent $54 on protein has signaled clear intent. A targeted, relevant cross-sell at that moment converts at 2–3x the rate of a pre-purchase recommendation. If you're not running a data-driven post-purchase offer on Shopify, you're leaving the easiest money in e-commerce on the table.
Frequently Asked Questions
How much order history do I need to run a meaningful market basket analysis?
As a rule of thumb, you need at least 10,000–20,000 orders to surface statistically reliable association rules. For brands with 50–200 SKUs, 12–18 months of data typically works well. If your order volume is high but your catalog is small (under 30 SKUs), 6 months may be sufficient. If your catalog is large (500+ SKUs), you may need 24 months to get enough support for long-tail pairings.
Can market basket analysis work for subscription-first DTC brands?
Yes, but you need to segment your analysis carefully. Subscription orders often mask the true "discovery" moment — the initial purchase combination. Filter your MBA to analyze first-order and non-subscription orders separately to surface cross-sell rules that reflect genuine discovery behavior rather than recurring replenishment patterns.
What's a realistic AOV lift I should expect from implementing MBA-driven cross-sells?
Based on patterns across mid-market DTC brands with 50–500 SKUs, a well-executed MBA cross-sell program typically drives a 12–30% AOV increase within 60–90 days when applied across post-purchase, cart, and email channels simultaneously. Brands with high catalog depth (200+ SKUs) tend to see the larger end of that range because there are more pairing opportunities to surface.
Do I need a data science team to run market basket analysis?
No. Tools like Affinsy are built specifically for mid-market DTC teams without data scientists. You connect your Shopify store, the analysis runs automatically, and you get actionable association rules with support, confidence, and lift scores — no SQL, no Python, no data engineering required.
How often should I re-run my market basket analysis?
Quarterly is a reasonable baseline for most brands. If you're actively expanding your catalog, launching new products, or running seasonal promotions, re-run monthly. Purchase behavior shifts as your product mix and customer base evolve — stale association rules from 12 months ago may no longer reflect what your current customers are doing.
Ready to Run Market Basket Analysis on Your Shopify Data?
The story above isn't unusual. Across mid-market DTC brands doing $5M–$50M in GMV on Shopify, the pattern repeats: teams are making cross-sell and bundling decisions based on category intuition rather than transaction data — and leaving meaningful AOV and margin improvement on the table.
Affinsy was built to close that gap. Connect your Shopify store, and within minutes you'll have statistically validated market basket association rules showing you exactly which products your customers buy together — and which cross-sell recommendations you're running that are actively hurting conversion.
No data science team. No complex setup. Just your order data, analyzed and made actionable.
Try Affinsy with your data and see what your customers are actually buying together.
