The four buckets.
Most Shopify app fees fit one of four buckets: Software expense (general SaaS not tied to specific revenue), Marketing expense (apps that drive customer acquisition), COGS (apps that flex with order volume and are part of fulfillment), or Selling expense (apps that flex with revenue but are not COGS).
A theme app like Shogun is Software expense. An email tool like Klaviyo is Marketing. A shipping calculator like ShipStation is split between Software (the subscription) and COGS (the per-shipment cost if any). A loyalty app like Smile.io that pays out points as discounts is Selling expense. Putting them all under "Apps" obscures unit economics.
COGS or operating expense?
The test for COGS treatment: does the cost flex with order volume in a near-1:1 way, and does it relate directly to fulfilling the order? If yes to both, COGS. If either is no, operating expense.
Stripe and Shopify Payments fees flex with transaction volume; they go in Processing fees as part of Selling expense (some accountants put them in COGS, both are defensible). A fixed-fee shipping calculator like ShipStation that costs $99/month regardless of volume is Software expense, not COGS. A per-shipment label cost from EasyPost that flexes with shipments could go in COGS.
- Variable per-order: COGS or Selling expense
- Fixed monthly subscription: Software expense or Marketing
- Customer-acquisition driven: Marketing
- Operations infrastructure: Software expense
A worked classification.
A 12-app stack: Klaviyo ($150/mo), Yotpo Reviews ($70/mo), Smile.io Loyalty ($50/mo), Shogun Page Builder ($99/mo), ShipStation ($99/mo), Recharge Subscriptions ($99/mo + 1.25% per order), Privy ($30/mo), Tidio Chat ($39/mo), Bold Bundles ($30/mo), Avada Air Reviews ($19/mo), Loox ($35/mo), Plus Shopify Plus base.
Classification: Klaviyo + Privy + Loox + Avada + Yotpo = Marketing ($304). Smile.io + Recharge fees = Selling expense ($50 + variable). Shogun + ShipStation + Tidio + Recharge base + Bold Bundles = Software expense ($297). Plus subscription = Software expense or COGS depending on Plus pricing model. The variable Recharge fee at 1.25% × monthly subscription revenue flows separately as a Selling expense.
Free vs paid app trials.
Many apps offer 14-30 day free trials before billing. The free trial period is not an expense. Many merchants accidentally accrue trial-period costs that never materialize. Watch for this in monthly close: any "trial active" status should not have an accrual.
When the trial ends and billing starts, the first month’s invoice is the first expense. Track app sign-up dates so trials and billing-start dates align with the GL. Apps that bill annually (with a discount) capitalize as a prepaid expense and amortize monthly: $1,200 annual billed in January is $100/month expense, with prepaid asset $1,100 declining to zero over the year.
App refunds and switching costs.
Switching apps is common. You install Klaviyo, run for 6 months, switch to Omnisend. The Klaviyo final invoice is an expense; the new Omnisend trial is not yet. The data migration cost (often a one-time setup fee from the new app or a freelancer) is a one-time expense, often Software setup or Marketing setup.
Annual subscriptions you cancel mid-term may or may not refund pro-rata. If they refund, debit Bank, credit Prepaid expense for the unused portion. If they do not, the prepaid asset writes off: debit Software expense, credit Prepaid expense. Most Shopify apps do not refund annual; budget accordingly when committing to annual deals.
App expense as a margin diagnostic.
Total app expense as a percentage of revenue is a useful health check. For most Shopify SMBs, app expense should be 1-3% of revenue. Above 5% and the stack is bloated; you are paying for apps that do not move metrics enough to justify their cost.
Audit the stack quarterly. For each app, identify the metric it is supposed to move and whether it has moved that metric. If you cannot articulate why an app is paid for, cancel it. Brands operating in lower-cost markets often benchmark relatively low on app-expense ratio (1.5-2.5%) because local-currency revenue stretches farther against USD-priced apps than equivalent volumes in higher-cost markets.
Where Nonari fits.
Nonari pulls Shopify app billing data and classifies each app into the four-bucket framework with sensible defaults you can override. The classification feeds the P&L immediately so Software expense, Marketing, Selling, and COGS are clean monthly. Multi-currency handling for USD-priced apps consumed by a non-USD-functional org converts at month-end rate with FX isolated.
For merchants tracking ROAS and contribution margin, having Marketing expense cleanly separated from Software expense matters. The CAC calculation needs the right denominator and numerator. With apps misclassified, your CAC is wrong, your contribution margin is wrong, and your operational decisions follow. The accounting setup is not a back-office concern; it is an operational input.