E-COMMERCE
I run a Shopify app P&L. This is what the job taught me.
Most operators who talk about e-commerce have run a store or sold to one. I sit on the other side of the install button, carrying the P&L for three Shopify apps used by tens of thousands of merchants, which means I see the same failure modes across a very large sample rather than in one brand.
THE OPERATING REALITY
Five things about e-commerce SaaS that are obvious once you have run it and surprising until then.
Install and uninstall both take a minute
There is no procurement, no contract and no switching cost. A merchant can be gone before your onboarding email lands. That makes day-one experience a revenue lever rather than a UX preference, and it makes any retention plan that starts at day thirty already late.
Revenue arrives on more than one rail
App-store billing and direct billing behave differently, report differently and cancel differently. Most reporting is built against whichever rail came first. That is the single most common reason an e-commerce SaaS number is wrong, and it is invisible from the chart.
Free installs are not customers
Active installs is the most quoted and least useful number in the category. It mixes free, trialling, paying and dormant, and it is usually cumulative rather than current. Retention measured on that base tells you almost nothing about the business.
The category commoditises faster than the roadmap
Whatever the feature was, someone ships it at a third of the price within a year, and the platform itself may ship it for free. Competing on capability is usually the losing game. Pricing, positioning and the surface you already own are the ones worth playing.
Support is the product for a long stretch
In a category where the tools are near-identical, response time and competence are what the reviews actually reward. That is a customer success question, and treating CS as a cost centre in this category is a strategic error rather than a budget one.
WIRED, NOT DESCRIBED
The reporting layer for my division reads these directly. Knowing what each one can and cannot tell you is most of the job.
| Billing | Stripe · Shopify Partner · app-store payouts |
| Merchant behaviour | PostHog · GA4 · product event streams |
| Lifecycle | Customer.io · Klaviyo-adjacent email · in-app messaging |
| Support | Intercom · HelpScout · review surfaces |
| Commercial | HubSpot · partner and agency pipeline |
| Search | Ahrefs · app-store listing performance |
Stripe · Shopify Partner · app-store payouts
PostHog · GA4 · product event streams
Customer.io · Klaviyo-adjacent email · in-app messaging
Intercom · HelpScout · review surfaces
HubSpot · partner and agency pipeline
Ahrefs · app-store listing performance
BRAND SIDE AND APP SIDE
The mechanics rhyme. A DTC brand and a Shopify app both live on thin margins, a fickle install-or-buy decision, and a stack of tools that disagree with each other about the same number.
The billing-rail problem is universal
Any brand selling through more than one channel has the same defect waiting: a metric wired to one rail and reported as if it covered all of them.
Cohort truth beats headline truth
Headline retention flatters. The cohort view is where the answer is, on both sides of the install button.
The operating cadence is the same shape
Named owner per number, a weekly rhythm that surfaces movement, and a worklist rather than a dashboard people have to remember to open.
IF YOU RUN E-COMMERCE OPERATIONS
I am not taking consulting work. I am interested in operating roles, and I am always happy to compare notes on how a reporting stack is wired. The check below is eighteen questions and takes about ten minutes.
START A CONVERSATION