A small Bangkok accounting firm goes digital, Part 7: retrospective — every step must hold up to market testing

This is the last chapter of the series. Across the previous six, this small Bangkok accounting firm went from “intake by mail and LINE, tracking in Excel, bookkeeping by hand” all the way to clients submitting online, an intake list, data-driven tiers, a PWA portal and AI extraction with review. This chapter looks back at why the road was walked this way.

Back to the start: why not buy it all at once

On day one, this firm was not short of anything — it was short of everything: scattered submissions, intake kept in memory, pricing by private judgment, keying done in overtime. And it had been disappointed by off-the-shelf cloud accounting systems three times:

• they lock in their own accounting logic, and decades of client history across dozens of clients cannot be migrated;
• per-seat subscriptions do not add up over years, and still do not solve how clients submit vouchers;
• they sell software, not a workflow for collecting documents and talking to clients.

Three disappointments point to one decision: digital is the end goal, but it must be built in stages. Not because the firm does not need it now — it desperately does — but because buying everything at once is expensive, does not match the current workflow, and nobody has designed how it lands.

Why every step follows this order

Looking back, the order was not random. Every step paved the road for the next:

Step	What it did	What it left for the next step
Website and voucher uploads	Clients submit voucher images online every month	a steady, trackable stream of vouchers
Internal intake list	Client, month, item count, missing status at a glance	replaced Excel and memory, accumulated real data
Data-driven tiers	Graded by monthly volume, overage charges	transparent pricing, clients understand usage decides price
PWA client portal	Tiers, profile, monthly status, on-demand upgrades	clients see and upgrade on their own, lower communication cost
AI extraction with review	OCR extraction, structured drafts, bookkeeper review	keying hours saved, judgment stays human

Each stop changes one small slice of the flow, so bookkeepers, clients and the owner can keep up; and each stop leaves data behind that becomes the input for the next.

Every stop answers three questions

This is the most important principle of the whole series: in digital, no step passes on “feels like it should exist.”

Before and after each step, answer:

1. What problem does it solve? Does it point at a real, specific pain — lost submissions, missed chases, unexplainable quotes, keying overtime — or just “others have it, so we should too”?
2. How do we prove it? Is there a comparable change after launch — upload volume up, missed chases down, clients upgrading themselves, keying hours down?
3. Is it worth it? Do the numbers add up — build cost, maintenance, bookkeepers’ learning time, multiplied by time — still worth it?

If the answers do not hold, the step is not done, or it waits. Building in stages is not delay — it is putting every step on a verifiable market fact.

What the owner can take home

• Digital is the end goal, but build in stages — no step skips proof and runs on imagination;
• Submission is the foundation: how clients get vouchers in decides whether anything else is even worth discussing;
• Track before scaling up: the intake list clarifies what is missing before tiers and extraction can be discussed;
• Tiers come from data: real volume grades with overage charges make pricing transparent and quotes consistent;
• Let clients see it: the PWA portal lets clients check tiers, watch status and upgrade on their own, and only then does communication cost come down;
• AI saves hours, not judgment: OCR extraction, human review before posting, responsibility always with the licensed CPA;
• Ask first at every step: what problem does it solve, how do we prove it, is it worth it.

The series ends here, the firm’s road does not

Digitalization has no finish line: voucher types change, client volume changes, AI recognition improves. But this firm has now accumulated trackable intake data, a clear tier structure and a habit of “verify before invest” — every future step has a basis. Building in stages is not an end — it is building every step on something visible and verifiable.

This is the end of the series. The whole series follows a small Bangkok accounting firm on its journey into digital; the scenario is drawn from our real implementation experience in accounting services and does not depict any single client. The intake list behind the voucher uploads, service-tier management, PWA client portal and OCR/AI extraction described here are independent system projects with their own scope, not part of the standard website package.