Dext built a business on the receipts nobody wants to type

A receipt is a small piece of paper with a remarkable talent for creating work. Someone has to collect it, read it, decide what the purchase belongs under and get the numbers into the books. Multiply that little errand across a business, then across an accountant’s clients, and the pile becomes a market. Dext built its business there.

The company began as Receipt Bank in 2010, founded by Alexis Prenn and Michael Wood. Recalling its origins after the announcement of its sale to IRIS, Prenn placed the pair “at a desk in my wife’s office”. He also supplied a useful correction to the tidy founding story: the belief that they were in the expenses business lasted three weeks.

That recollection captures Dext’s trajectory. A receipt was the entrance to a larger problem: the repetitive work required before financial records become useful. Today, the company sells bookkeeping automation to businesses and the professionals who keep their accounts. Its expanding ambitions still depend on a modest test. Does the software remove more work than it introduces?

The money was in the typing

Dext’s core workflow is easy to picture. A business owner photographs a receipt, forwards an invoice or uploads a document. The software extracts the information, helps organise it and makes it available for review. Once published, the data goes into connected accounting software. The product turns a document someone would otherwise have to transcribe into an accounting record someone can check.

That distinction explains its place in the market. Xero, QuickBooks and Sage are destinations for the data. Dext concentrates on collecting and preparing it, with tools for the practice managing many businesses as well as the business submitting its own documents. An accountant can oversee clients; employees can submit expenses; an owner can stop treating the inbox as a filing cabinet.

Setup matters. Dext’s getting-started guidance describes importing the chart of accounts, suppliers and contacts from connected software. Those are the coordinates that make the captured numbers usable. A clean scan alone cannot decide how a particular business wants to organise its books. If an accountant manages the account, that professional will typically handle publishing.

The model gives Dext two routes to the customer. Businesses can subscribe directly, while practices can provide accounts through their own subscriptions. The person photographing the lunch receipt need not be the person buying the software. That arrangement puts accountants in an influential position: they experience both the paperwork arriving and the effort required to finish it.

Thirty seconds, repeated often enough

The most persuasive customer evidence comes in small units. In a Dext-published case study, Singapore’s Harvest Accounting estimated that the software saved 30 seconds of data entry per invoice and about 20 hours a week compared with its previous manual process. Founded by Matthew Phua and Bryan Zhao, the firm had grown from five clients to more than 100 by the time described in the study.

“We estimate 30 secs is saved per invoice in terms of data entry.”Harvest Accounting · Dext customer case study

Harvest also described using a cloud workflow with colleagues and clients in different countries, reducing its need for physical office space and document storage. These are the firm’s estimates and experience, rather than a promise for every subscriber. Their usefulness lies in showing what to measure: the repeated action, its frequency and the work that remains.

Another Dext case study offers a different lens. A practice leader named Johann reported 3,323 documents processed in 30 days, with 67% automatically published to QuickBooks Online. That leaves a substantial remainder outside the automatic publishing route. The percentage is more informative than a blanket claim that bookkeeping has been automated. It shows where a team can look next.

For a prospective buyer, the copyable experiment is straightforward: choose one client or business, time the existing process, then run a representative batch through a trial. Count corrections and document chasing as well as typing. Begin automation with predictable suppliers. Expand only when the result survives review. The objective is fewer minutes spent finishing the books, not simply more documents uploaded.

The receipt app outgrew its name

Receipt Bank became Dext in 2021. The same year, the company acquired Greenback, a move that widened its attention from purchase paperwork to digital sales. Dext Commerce launched in August, collecting and categorising data from e-commerce, point-of-sale and payment platforms including Amazon, Shopify, PayPal and Stripe.

Online selling creates an especially awkward collection problem. Records arrive from different services in different formats. The Commerce launch addressed the work of gathering and consolidating them before sending revenue information to Xero or QuickBooks Online. A photograph of a receipt would never solve that whole problem. Buying Greenback supplied another route into the financial record.

Expansion also created separate products for customers to navigate. In March 2025, Dext’s release notes described bringing key tools from Prepare, Precision and Commerce together. Data Health & Insights brought accounting-data checks into the main platform as a paid add-on; Commerce Lite brought in e-commerce transactions. The stated benefit was less switching between products.

The change reveals a practical tension in software growth. Each additional task offers another reason to subscribe, but another interface can become another chore. Consolidation was Dext’s response to that friction. Its expertise now spans reading documents, connecting accounting systems and helping practices inspect the quality of the resulting data.

Investors followed the paperwork

The financing arrived well before today’s rush to attach AI to office work. Receipt Bank received $10 million from Kennet Partners in 2016 and a $50 million Series B from Insight Venture Partners in 2017. In January 2020, it announced $73 million in combined equity and debt, led by Insight Partners with Augmentum, Kennet and CIBC participating.

The 2020 announcement said business customers had doubled in 2019 to more than 360,000. The money was intended to broaden the product suite and support expansion in Europe, Australia and North America. Those are dated company figures, but they establish an important point: Dext had a substantial document-processing business before generative AI became a default software sales pitch.

Hg became a major shareholder in 2021 through an ownership transaction. Then IRIS completed its acquisition of Dext on December 23, 2024. Sky News reported a price of about £525 million; the parties did not disclose the transaction terms. An acquisition price belongs in the ownership story, separately from capital raised to run and expand the company.

IRIS’s rationale was to connect Dext’s bookkeeping platform with IRIS Elements and develop a more complete accountancy workflow. The companies also emphasised continued choice of accounting software. That makes the acquisition interesting beyond its price: customers are being offered closer integration while being told their existing ledger connections will remain supported.

A subscription has to beat the spare minutes

Dext charges for that convenience. On September 7, 2026, its UK business pricing advertised an entry plan at £290 billed annually, equivalent to £24.17 a month excluding VAT, for five users and 250 documents a month. The advertised monthly-contract starting price was £30. Accounting practices build packages around clients and selected features.

The entry price is only part of the calculation. Extra line-item and statement extraction can cost more; Commerce Lite and AI Assist are additional purchases. The business AI Assist launch offer is £7.50 monthly through October 31, 2026. Dext says uncapped launch usage ends then, making the offer a dated price rather than a permanent assumption.

Payments extends the bill again. The UK business payment page advertises 25p per Faster Payment and a one-time £15 connection fee, excluding VAT, on top of the Dext subscription. Availability and supported integrations depend on the account and region. Paying from the workflow could remove another handoff, but buyers need to price the version they can actually use.

Competition keeps that calculation honest. Hubdoc is bundled with eligible Xero subscriptions, making basic capture a tough place to demand another fee. AutoEntry uses credits for processing documents. Dext’s broader practice workflow, data checks and automation must justify the difference for each buyer. A business with few documents and satisfactory existing capture has less reason to add another subscription.

The work left for a person

Dext’s stated vision is to make accounting effortless. Its careers materials frame the culture around speaking up, taking responsibility and working together. The product’s more useful promise is specific: remove routine handling so people can spend attention elsewhere. AI Assist moves further into categorisation and field updates, where helpful suggestions still depend on the business context.

The company’s own practice tools make the remaining work visible. Its Time spent insight groups activity by clients and colleagues to support workload planning and repricing discussions. Dext explicitly describes colleague activity as a capacity-planning tool rather than a performance evaluation. The distinction matters: more clicks are not necessarily more value, and fewer clicks do not prove a finished job.

Dext’s story began with an idea that quickly proved too narrow. It has since followed the tasks around financial data: collect it, prepare it, inspect it and increasingly act on it. The test remains pleasantly unglamorous. After the scan, the suggestion and the subscription invoice, does the accountant have less work to do? Every receipt gives Dext another chance to answer.