Summary

  • Expensify reported US$5.924 million of interchange revenue in Q2 2026, up 11.8% from US$5.297 million. It recognises that interchange gross within revenue because it concludes that it is principal under its issuing-bank and processor arrangements.
  • Customer cashback rose 18.8% to US$2.950 million and is recorded as contra revenue. Interchange less cashback therefore increased only 5.7%, to US$2.974 million.
  • A separate Visa volume incentive reduces cost of revenue. It fell from US$0.8 million to US$0.2 million. Adding that disclosed cost offset to interchange less cashback produces a provisional three-line receipt of US$3.174 million, down 12.2% from US$3.613 million.
  • The bridge is not card gross profit. Expensify does not separately disclose card-user count, spend, processor cost, support, fraud, credit exposure, funding cost or the allocation of shared platform expense.
  • Subscription revenue fell 6.4%, paid members fell 1.8%, total revenue fell 5.3%, and consolidated gross profit fell 12.1%. The next useful disclosure is a card contribution schedule that keeps gross interchange, customer rewards, network incentives and direct operating costs visible on the same page.

One growth rate, three accounting locations

Expensify led its second-quarter highlights with interchange revenue of US$5.9 million, 12% higher than a year earlier. The number is useful. It shows that transactions on the Expensify Card generated more network revenue even as the average number of paid software members declined from 652,000 to 640,000. It does not show how much of that revenue remained after rewards and the programme’s direct economics.

The reason is not an obscure adjustment. It is the shape of the reported accounts. Expensify earns interchange when card transactions are authorised and settled through the programme it runs with The Bancorp Bank and Visa. The company says its contracts with the issuing bank and payment processor make it the principal, so it recognises all interchange generated as gross revenue.

Cashback appears in the same revenue section with the opposite sign. Expensify treats rewards as consideration payable to customers. The company either applies the amount against a customer receivable or pays it in the following month, and the income statement records it as contra revenue. The Q2 revenue table therefore shows US$5.924 million of interchange and negative US$2.950 million of cashback before arriving at total revenue.

The second rebate lives below revenue. Visa pays volume-based incentives through a vendor arrangement. Expensify records those amounts as a reduction of cost of revenue when earned. The incentive is economically relevant to the card programme, but it never enters the interchange growth headline and it does not appear beside cashback in the revenue table.

Three lines, then, answer three different accounting questions. Gross interchange describes network revenue controlled by Expensify. Cashback describes consideration returned to customers. The Visa incentive describes vendor consideration that offsets delivery cost. A reader who stops at the first line sees the card programme before either rebate has done its work.

The first rebate consumed half the interchange gain

The year-on-year bridge begins with a US$627,000 increase in interchange, from US$5.297 million to US$5.924 million. Cashback increased by US$466,000, from US$2.484 million to US$2.950 million. Subtract the second movement from the first and only US$161,000 of additional interchange after cashback remains.

That arithmetic changes the growth rate materially. Gross interchange expanded by 11.8%. Interchange less cashback expanded by 5.7%, from US$2.813 million to US$2.974 million. Cashback absorbed roughly three quarters of the incremental interchange dollars. This does not mean rewards are waste. Cashback may be necessary to win spending, change payment behaviour and keep the card competitive. It means adoption should be judged after the price paid for that behaviour, not before it.

The distinction is particularly important because Expensify’s core subscription revenue moved the other way. Subscription fees declined by US$2.094 million, or 6.4%, to US$30.770 million. The card’s extra US$627,000 of gross interchange was not large enough to offset that fall even before cashback. After cashback, the net incremental amount was only US$161,000.

Other revenue was US$122,000, leaving total net revenue of US$33.866 million, down 5.3% from US$35.764 million. The card is becoming a larger part of the revenue mix partly because it is growing and partly because subscriptions are shrinking. Those are different forms of mix shift. One signals a successful adjacent product; the other raises the burden that adjacent product must carry.

Expensify does not disclose the number of card users, active cards or purchase volume. It would therefore be wrong to divide the interchange or the post-cashback amount by 640,000 paid members and call the result card yield. Paid members are users billed on Collect or Control plans; they are not a disclosed cardholder cohort. A clean-looking unit metric built from mismatched populations would conceal more than it explains.

The second rebate reversed the remaining progress

Visa volume incentives provide the second adjustment. Expensify earned US$0.2 million in Q2 2026, compared with US$0.8 million a year earlier. The US$600,000 decline was larger than the US$161,000 increase in interchange after cashback.

For a provisional receipt, add the incentive to interchange less cashback because it reduces cost of revenue rather than revenue. The result is US$3.174 million in Q2 2026: US$5.924 million of interchange, less US$2.950 million of cashback, plus US$0.2 million of vendor consideration. The same construction for Q2 2025 gives US$3.613 million. The disclosed three-line receipt declined by US$439,000, or 12.2%.

This bridge is an analytical presentation, not a company KPI and not a proposed accounting reclassification. Each component remains where Expensify reports it. The purpose is to keep two economically connected rebates in view while respecting their different accounting treatment.

Nor is US$3.174 million card gross profit. Cost of revenue includes customer support, hosting, credit-card processing fees, third-party software, finance-lease amortisation, capitalised-software amortisation and outsourced engineering. Expensify does not allocate those expenses between software subscriptions and the card programme. Public filings also do not provide card-specific fraud losses, funding costs, bank fees or operational headcount. The bridge stops before those missing lines.

That restraint matters. Calling the receipt gross profit would turn an evidence improvement into false precision. The defensible conclusion is narrower: the headline increase in interchange did not survive the combined movement in the two disclosed rebates. Whether the fully allocated card contribution also fell cannot be determined from public data.

Why the Visa incentive may move independently

Volume incentives are not necessarily a fixed percentage of interchange. Network agreements can contain thresholds, tiers, timing rules, portfolio qualifications and true-ups. Expensify describes the amount as consideration for certain volume-based incentives and records it when earned, but it does not publish the formula.

The quarterly fall from US$0.8 million to US$0.2 million could therefore reflect a lower rate, a threshold effect, programme mix, timing or another contractual condition. It is not proof that card spend fell. Interchange itself rose, which points away from a simple volume-collapse story. The important observation is that the cost offset did not scale with the reported interchange line in this period.

This is an underappreciated dependency. Expensify can control product design, customer rewards and the commercial push behind its card. It does not unilaterally control network incentive schedules. If gross interchange grows while vendor consideration becomes less generous, the programme must recover the gap through better customer economics, lower direct cost or a changed rewards structure.

The six-month comparison shows the same tension at lower intensity. Interchange rose to US$11.465 million from US$10.331 million. Cashback increased to US$5.586 million from US$4.758 million. Interchange less cashback therefore rose to US$5.879 million from US$5.573 million. Visa incentives fell to US$0.5 million from US$1.0 million. The six-month three-line receipt was US$6.379 million versus US$6.573 million, down about 3%.

Quarterly and half-year views together suggest the incentive line deserves monitoring rather than a one-off explanation. They do not establish a permanent deterioration. The formula is undisclosed, and later periods may reverse the movement. But an investor should not model interchange growth as if every network-related offset rose automatically beside it.

The consolidated gross margin is consistent with the warning

Expensify’s Q2 gross profit declined from US$18.577 million to US$16.330 million, a fall of 12.1%. Gross margin narrowed from 52% to 48%. The percentage change is strikingly close to the 12.2% fall in the provisional card receipt, but that numerical proximity is not causation.

Total revenue fell because lower subscription activity and higher cashback outweighed greater interchange. Cost of revenue rose 2% to US$17.536 million. Expensify said higher amortisation of capitalised software was the main reason, partly offset by savings from using more artificial intelligence in place of human agents. The Visa incentive was only one component of that net cost line.

The proper reading is structural. Card adoption can improve gross interchange while consolidated gross profit deteriorates because cashback reduces reported revenue, network incentives change, subscriptions contract and shared delivery costs move. A single card headline cannot explain a mixed software-and-payments income statement.

The software amortisation detail adds another clock. Q2 amortisation of capitalised software was US$2.2 million, up from US$1.9 million. Capitalised development costs reached US$39.420 million before accumulated amortisation. AI support savings may lower current personnel costs, but capitalised product work reaches gross profit later through amortisation. The operating model moves costs across time as well as across functions.

That does not invalidate the company’s free-cash-flow progress. Expensify reported US$8.4 million of quarterly operating cash flow and US$6.4 million of free cash flow. Cash generation and gross-margin quality can improve or weaken on different schedules. The question for the card is whether recurring transaction economics can help replace declining subscription revenue without demanding proportionally more rewards, incentive support or balance-sheet capacity.

A card programme also occupies the balance sheet

Expensify Card posted collateral for funds held for customers increased to US$11.764 million at June from US$10.530 million at December. Earned interchange restricted cash rose to US$1.843 million from US$1.654 million. These balances are not expenses, losses or deductions from the provisional receipt. They show that the programme has a treasury and settlement surface in addition to an income-statement surface.

The distinction between collateral, restricted cash and cashback liability is essential. Posted collateral supports funds held for customers. Earned interchange restricted cash reflects cash whose use is restricted. Cashback liability, US$616,000 at June, represents rewards accrued but not yet applied or paid. None can be substituted for the current-period US$2.950 million cashback expense.

As transaction activity grows, the card may require more collateral, settlement assets and operational liquidity. The cash may remain an asset, but restricted assets cannot be deployed as freely as ordinary corporate cash. A card programme that produces attractive contribution can justify that use of balance-sheet capacity. One that depends on rising rewards and volatile vendor incentives needs a higher return to do so.

Expensify does not disclose enough information to calculate that return. Investors would need average and peak collateral, settlement timing, losses, interest or funding effects, and card-specific contribution. The reported balances nevertheless prevent a purely asset-light reading of the payments strategy.

The half-year cash-flow statement supplies a broader caution. Operating cash flow was US$8.551 million, down from US$16.040 million a year earlier. Many factors determine that change, and the card balances cannot be assigned sole responsibility. It does mean the payments thesis should be tested in cash and balance-sheet terms, not only through a growing revenue label.

The missing disclosure is a contribution waterfall

A useful card schedule would begin with purchase volume and gross interchange. It would then subtract customer rewards, processor and issuing-bank fees, fraud and credit costs, direct programme operations, and incremental support. Network incentives would appear as a separate positive line. The bottom would show contribution before and after allocated platform cost.

No confidential merchant names are needed. Expensify could disclose ranges, indexed values or year-on-year movement. It could separate active card companies from software-only companies and show whether card adoption improves retention or expands paid-member activity. Without that cohort view, the card may be succeeding as a transaction product while the wider customer relationship continues to shrink.

The schedule should also reconcile the balance sheet. Average card collateral, settlement assets, restricted interchange cash and reward liabilities would show how much operating capacity accompanies each dollar of contribution. A return-on-required-capital measure would prevent an apparently high-margin payments stream from being evaluated without its liquidity needs.

The second-quarter filing already supplies the architecture. Interchange, cashback, Visa incentives, paid members, subscription revenue and several restricted balances are disclosed. The missing step is to connect them. That would let investors distinguish four possibilities: more spend from existing card customers, more card adoption among existing software customers, genuine new-customer acquisition, or a mix shift caused by declining subscriptions.

Until then, the two-rebate test is a disciplined interim measure. It does not solve the allocation problem. It stops the most visible gross number from carrying an economic claim that the published accounts do not support.

What the headline should have to prove

The bullish case is plausible. A card can deepen product usage, create transaction revenue, reduce software churn and give Expensify a wider distribution proposition. Gross interchange growth is one piece of evidence. Greater automation could lower support cost. New bank, accounting and travel integrations could expand the addressable transaction base.

But the proof must survive the order of the accounts. Interchange should grow faster than rewards over a sustained period. Network incentives should be explained well enough that a temporary tier benefit is not capitalised into the model. Direct processing and operating costs should not absorb the remaining spread. Card customers should retain or expand software subscriptions. Required collateral should earn an adequate return.

Q2 2026 did not provide that full proof. Gross interchange increased, but cashback took most of the incremental dollars and the decline in the Visa incentive more than removed what remained. Subscription revenue and paid members still contracted. The three-line receipt therefore gives a more cautious answer than the 12% headline.

That answer is neither that the card is failing nor that accounting presentation is misleading. It is that a payments product must be read across revenue, contra revenue, cost offsets and the balance sheet. Expensify’s disclosure makes the first reconstruction possible. The next task is to publish the missing direct-cost and cohort lines so the market can decide whether card growth is becoming contribution growth.

Sources