Disclose Only Reconciled Numbers: Why HOPE Does Not Publish 2023 to 2025 Figures

In July 2026 I opened the company's revenue report and saw a number that looked healthy. It was wrong. Our accounting software, fed by hand, ran a few days behind and was missing 125 orders for the month. Worse, when I traced it back, 53 places in our own code were quietly asking that same software for revenue whenever they needed it. For nine days the company was blind to its own sales, and nobody raised an alarm.

That incident taught me more than any finance book. It also explains why, when we raise capital, HOPE discloses figures only from 2026 and nothing for 2023 through 2025.

Why I do not disclose 2023 to 2025

HOPE was registered on 23 August 2023. In the first three years we sold products, served customers and had some good months. But I do not have a reconciled ledger for that period. There are numbers in accounting software, numbers in chat messages, numbers in the memory of the people who did the work. They do not agree with each other, and I cannot prove which is right.

When an investor asks what revenue was in 2024, the easy answer is to pick the best-looking of the three numbers. I chose the harder answer: no reconciled ledger, no disclosure. A number that cannot be checked is not an asset of the company. It is a debt of credibility, and it gets called in exactly when you need credibility most.

The figures we share start in 2026, when the ledger became reconciled. Net revenue, after VAT, for the first nine months of 2026 is about VND 3.57 billion (about US$137K). The average of the last three months is about VND 364 million per month, a run-rate of about VND 4.4 billion a year (about US$170K). That is small next to many investor decks. But it has one virtue that larger numbers often lack: it holds up under cross-examination.

Three sources that check each other, and one door

HOPE's revenue today is built from three independent sources.

• The order ledger: orders are parsed automatically from the order chat group where staff close sales with customers.
• J&T waybills: every parcel shipped has a waybill with its creation time and cash-on-delivery value.
• Bank statements: at month end, the money that actually arrived is matched against the first two.

The accounting software is still there, but only for cross-checking, not as the source of the number. There is one trap we state in every report: when a customer pays by transfer in advance, the cash-on-delivery value on the waybill is zero, so waybills alone always understate real money. That is why we need all three sources rather than any one.

More importantly, we put this rule into the code itself. Only one single door is allowed to return the company's revenue. Every report, every AI advisor, every daily bulletin must pass through it.

A rule written in words cannot stop code.

We had written "do not take sales numbers from the accounting software" into our operations handbook. Five days later the code was still reading the accounting software as if the rule did not exist. A rule only takes effect once it becomes the structure of the system, meaning there is only one road to the number.

Three small lessons from three incidents

The 9,256-lead trap. Our prospect table once showed 9,256 rows, a number that made people excited. When we filtered by creation date, 9,098 of those rows turned out to be one bulk import on a single day. The real leads were the remainder, far fewer. Since then we have a rule: never use a table's total row count as a monthly figure; always filter by creation date and look at the distribution first.

If a scrape comes back blank, stop and do not write. One of our data sources once lost its login session, and the scraper overwrote the real figure of 1,481 with zero. Nobody acted in bad faith; the system did exactly what it was told. Now any scraper that returns zero rows must stop, keep the old number, raise a warning flag, and a watchdog monitors every source. The principle is failure must be loud: a system that goes quiet when it breaks is more dangerous than one that breaks noisily.

Every report carries a recommendation. A report that only lists numbers is unfinished. Our operations advisor bot runs around the clock, reads live data, and every report it sends must come with a concrete recommendation.

All three lessons share one pattern: wrong numbers are rarely the work of someone lying. They are wrong because the system allowed a shortcut, and the shortcut produced a comfortable number.

One set of numbers for every application

One thing is rarely discussed: many funds today use AI to read applications. They hand the form, the deck, the data room and the referral letter to a model and let it look for inconsistencies. A company that states revenue one way on one form and another way on the next will be caught in seconds, and usually nobody tells the company.

So we keep one single set of numbers for every application: the same revenue figure, the same customer count, the same method, the same starting point of 2026. Paying customers number 1,294 (unique phone numbers, retail, March to September 2026). Paid advertising in 2026 is about VND 34 million, roughly 1% of revenue. What we cannot measure, we label as an estimate or leave out. For example, we have no reconciled 2025 data, so we make no claim about long-term retention; we say only that 40.5% of customers who first bought between March and June 2026 have bought again.

This discipline does not make our application look better. It leaves nothing in it to catch us on.

What I hope investors take away about HOPE

I do not hope you read VND 3.57 billion and feel impressed. I hope you look at how we answer the question "where does this number come from?" Three sources check each other. One door in the code. Automatic safeguards for blank data, bulk-imported data and late data. And a willingness to say "we do not have that yet" instead of filling the gap with a comfortable figure.

I believe a young company can lack many things, but it must have one: trust in its own books. That cannot be bought after the funding arrives.

Check us. HOPE's numbers can be verified, and we want you to verify them.