Rebuilding an e-commerce brand that had stalled at one million

The problem wasn’t marketing

The brand had been sitting at about $1M a year for several quarters and couldn’t seem to get past it. When we first sat down with the leadership team they already had a theory: marketing spend was inefficient, and the fix was better acquisition. They’d tried buying more traffic. It hadn’t moved the number.

So before agreeing with anyone, we spent the first few weeks pulling the data together. Shopify, the CRM, point of sale and Google Analytics had never been looked at side by side, and once they were, the story was pretty clear. Acquisition was fine. People were arriving. It was everything after that.

• Cart-to-checkout conversion was under 20%. Comparable brands run 30% to 40%.
• Repeat purchase rate was flat at around 12% and had been for a long time.
• The referral program had under 5% participation, which made it more admin than it was worth.
• Checkout was five steps long, and a couple of those steps collected information nobody in the business ever used.
• Support was two people working out of a shared inbox with no macros, no triage and no escalation path. They were tired.
• Product pages opened with specifications and never got around to why you’d want the thing. There was no social proof anywhere near the buy button.

None of that is unusual. What was unusual was how confident the marketing theory had been, given that nobody had checked.

What we did

We ordered the work by how cheap it was against how much it would move. Checkout was quick and would show results in weeks. Retention was slower and more structural. Doing it in that order meant the second piece was funded by evidence from the first, which made the harder conversations easier.

Getting the facts straight. We reconciled twelve months of data across the four systems into one view of the customer journey, ran customer interviews, and read through the support tickets ourselves, which is slow and worth it. We also benchmarked ten comparable DTC brands on checkout length, post-purchase sequencing and referral mechanics, so that “below average” could be a number instead of a feeling.

Checkout and the product pages. We took checkout from five steps to three and dropped the fields that fed nothing downstream. Product pages were rebuilt to lead with what the product does for the buyer, the navigation was simplified, and we put reviews and a bit of reassurance at the points where people were hesitating.

Retention and referrals. The tiered referral scheme became a single $10 credit, which people could actually explain to a friend. Post-purchase emails were split by order value and product category instead of sending everyone the same sequence, and we added loyalty prompts based on what a customer had spent over time. On the product side we redesigned the bundles, added upsells at the cart, and introduced seasonal drops and limited runs so there was a reason to buy this month rather than eventually. Once volume picked up, we had a much better hand when we went back to suppliers to renegotiate.

Support. We wrote a macro library, set up a triage queue with a proper escalation path, and built dashboards in Looker Studio so the team could see conversion, retention and support load without asking anyone for a report. The two-person team is still running all of it.

Rebuilt artifact

Before

After

1. Checkout in five stepsThree-step checkout
2. Unused checkout fields$10 referral credit
3. Tiered referral structureSegmented email flows
4. One email flow for allMacros and triage queue
5. Support with no triageReal-time dashboards

Where it landed

• Revenue grew from $1M to $5M over the same period, alongside acquisition work the team ran themselves.
• Customer retention improved 22%.
• Lifetime value rose 31%.
• Gross margin expanded 20%.
• Referral participation went from under 5% to 18%.