how we found hidden revenue for a shopify fashion brand

before we audited other people’s stores, we ran one of our own - a minimalist watch brand. this is the analysis we wish we’d run on ourselves, applied to six years of real shopify data, start to finish.

we get asked whether the fte analysis works. the most honest answer we have is a story we don’t usually lead with: freedom to exist started as a watch brand. we sold minimalist watches - a fashion accessory - direct-to-consumer on shopify from 2015 until we closed the store in february 2021.

so before we ran a single client audit, we ran our own store through the exact process we now use for others. every order, every customer, every product - six years, read back through the data. we already know how this story ended, which is what makes it the most honest case study we can show you. here is what was hiding in plain sight.

the dataset

this is a real, closed brand - not a synthetic model. across its life it took 1,846 paid orders from 1,613 customers, generating £124,725 in revenue. healthy enough on the surface. but the single most important number isn’t in the totals - it’s in what happened to price.

the average watch sold for £119 in 2017 and £33 by 2020 - a 72% fall. and here is the trap: 2020 set the order-count record (658 orders) while producing 40% less revenue than 2019. more orders looked like growth. it wasn’t. it was the same demand, bought at ever-lower prices.

the discount data shows how it happened. a third of all orders (33.2%) used a code - but the average hides the slide: 36% in 2017, 54% by 2019. a welcome offer is defensible; what we actually ran was a rolling calendar of monthly codes - july, august, september, october, november - that put the store on permanent promotion. £10.4k of margin was given away in codes alone.

the six weeks that were the business

november and december accounted for 34.2% of all lifetime revenue. december alone took £24.9k, and december 2017 was never beaten. the gift wrapping service confirmed it from inside the basket - added to 20.9% of all orders. this was a gifting brand whether it called itself one or not, yet marketing effort was spread evenly across the whole year.

the two colourways that carried the range mapped straight onto that gifting story. rose gold & tan sold 312 units for £24.1k - nearly twice any other colourway, a fifth of all watch revenue on its own. one range read “for her”, one read “for him”. a catalogue this concentrated argued for fewer products bought deeper, not a wider range - the long tail of colourways tied up cash without earning its keep.

the second purchase that never happened

this is where the real hidden revenue was. 88.6% of customers bought exactly once. for a durable product that’s partly structural - a watch lasts years - but it compounds the price problem: when almost every sale is to a new customer, every sale carries full acquisition cost, and cutting prices cuts straight into the only margin there is. repeat revenue is what makes discounting survivable, and this business had almost none.

the fix was already in the catalogue. interchangeable straps were the built-in repeat purchase - a £9–29 way to make one watch feel new. the idea was right; the execution never landed.

the timing signal was there too. among customers who did come back, the median gap was 134 days, and a third returned inside a month. a strap-led email to owners 60–90 days after purchase - care tips, the strap range, a gentle “make it feel new” - would have landed exactly where the data says returners were already moving. near-pure-margin revenue from customers we’d already won, sitting untouched.

who actually funded the brand

the most valuable 162 customers - the top 10% - spent an average of £194 each and together made up 25.4% of all revenue. the remaining 90% averaged £64: essentially one watch. 162 people is a knowable group, small enough to thank by name, and with straps as the natural second product it was the obvious audience for early colour releases and an owners’ offer. a flat lifetime-value curve isn’t a failure - it’s a brief. it tells you retention revenue has to come from a deliberate second product, not from habit.

the money that quietly left

the report’s most expensive quiet number: refunds took back £1 of every £12. £11.2k was returned across the brand’s life - 8.3% of gross revenue - when healthy non-apparel ecommerce runs low-single-digit rates. watches invite returns (wrist fit, strap feel, gifts that missed), but at this level refunds erased most of a typical year’s growth. a refund log is a free research programme nobody reads: reason codes, sizing guidance and honest strap-width photography were the cheap fixes we never made.

search built the brand, social got the credit

this was a minimalist watch brand with the visual identity of an instagram-native business - and instagram-native businesses are usually assumed to be won on instagram. the order-level referrer data said otherwise. of 1,964 lifetime orders with a traceable source, search engines drove 900 orders and £55,765 - 46% of revenue. social media, across facebook, instagram, pinterest and youtube combined, drove 80 orders: 4.3%.

this doesn’t mean the social presence was wasted - a consistent visual identity probably supported the searches once they happened, and email undercounts anyone who opened on mobile then bought direct. but any brand assuming social parity with search, before checking its own referrer data, is making a £50k mistake in one direction or the other. we didn’t know which until we looked.

a genuinely international small brand

for a brand this size, the geography was a surprise. the uk was home - 68% of revenue - but nearly a third of the business came from abroad, and not as strays: real, repeat order counts across seven countries. the us alone was £15,430 across 194 orders, a 12% revenue share won with zero us-specific marketing.

the list that outlived the store

the last asset was the one we under-used most. 89.3% of a 4,039-strong contact file accepted marketing - including 2,426 subscribers who never bought. a consented list 1.5× larger than the customer base. the constraint was never the audience; it was what got sent to them. with straps unmerchandised and no owner journey, the list mostly carried the discount calendar.

what we do with this now

none of these findings needed new traffic or a bigger budget. they were all sitting inside data the store already owned - orders, customers, products, discounts, refunds. we just weren’t reading it. that’s the entire premise of what freedom to exist does today: we run this same analysis for live shopify brands, while there’s still time to act on it.