A Yearly Target Only Means Something When It Becomes Today's Customer Count

I have a small habit that sounds dull when I describe it. Whenever there is a revenue target for the whole year, I sit down and divide it. Revenue divided by the average value of an order gives a number of orders. Orders convert into customers. Customers, through a closing rate, convert into the number of people who must message us to ask about buying, every single day. Once I typed that exact chain of questions to an AI assistant: what is this target in Vietnamese dong, and for a product at this price, how many orders and how many customers are needed? I did not think of it as a method then. I just wanted to know whether the target could walk on the ground.

I later understood why that division reassured me more than many thick plans. A yearly target is a story. A daily customer count is an event. Everyone can agree with a story. An event is either enough or short this morning, and I know what to do about it today.

Why a yearly target cannot be measured by itself

A twelve-month target gives everyone a reasonable excuse to wait. Short this month, there is next month. Short this quarter, there is next quarter. By the time the gap is large enough for everyone to see, the time to fix it has run out.

A distant target also cannot give a concrete instruction to anyone. A salesperson cannot work with the sentence "we reach figure X this year". That person needs to know how many conversations must be opened and answered today. The content team needs to know how many enquiries the channels must bring in today. A target becomes work only when it has an address and a deadline of today.

The third point is subtler. Two targets with the same revenue can demand two very different worlds of work: one sells fewer orders of higher value, the other sells many small orders. Without the division you cannot see that difference, and the team pours its effort into a place that is not the bottleneck.

The chain: from revenue down to daily conversations

The numbers in this section are hypothetical, for illustrating the method. They are not HOPE's targets and not HOPE's actual figures.

Suppose a yearly net revenue target of VND 6 billion (about US$230K). The chain has four steps.

• Revenue to orders: if the average order is worth VND 600,000 (about US$23), we need 10,000 orders a year, roughly 27 a day.
• Orders to customers: if a customer buys 1.25 orders a year on average, we need 8,000 customers, roughly 22 a day.
• Customers to conversations: if 15 of every 100 people who message us buy, we need about 146 new conversations a day.
• Conversations to each team's work: each team takes its part: how many a channel brings in, who replies, and how fast.

The best thing about the chain is that it exposes the levers. For the same VND 6 billion:

• If each order is fuller because we sell a complete set of suitable products, say VND 800,000, orders fall to 7,500 and the daily conversations needed fall to about 110.
• If the sales team lifts the closing rate from 15% to 20% and nothing else changes, the conversations needed also fall to about 110.
• If more customers come back for a second purchase, the number of new customers we must find shrinks.

Three paths, three different pressures on three different teams. The division does not tell me which path is right. It forces the company to say which link it is betting on, and then to measure whether the bet holds.

Set it beside real numbers, not beside wishes

The division is worth something only if the other side is real. That is why I treat the essay on publishing only reconciled numbers and this one as a pair. If the input is a miscount, the chain is just a precise way to compute something false.

HOPE's real starting point, from the ledger reconciled since the start of 2026: net revenue for the first nine months was about VND 3.57 billion (about US$137K); the average of the last three months is about VND 364 million a month, a run-rate of about VND 4.4 billion a year (about US$170K). There were 1,294 paying customers, counted by unique phone number, retail, from March to September 2026. Among customers who first bought between March and June, 40.5% have bought again, and customers who bought two or more times generate about 66% of revenue. I described that yardstick in returning customers are the real measure.

Any real target of ours is tested through the same chain, starting from these numbers rather than from wishes. When counting daily conversations, the old trap is still there: filter by creation date, never take the total row count of a table. A daily number counted wrongly is more dangerous than no number, because it gives a false sense of comfort.

Connecting it to the DO, MEASURE, LEARN loop and the morning report

At HOPE the working loop is always: produce a real result, measure it against the plan, learn from the deviation, repeat. Dividing the target into daily numbers is what gives "measure against the plan" a meaning. Without a daily plan line there is only "measure", no "against".

In practice, the morning operations report does this. An operations advisor bot runs continuously, reads live figures from the order ledger and the waybills, places them next to the day's required level, and reports. Our rule is that every report carries a recommendation, not just a reading of numbers. An illustrative report looks like this.

Illustration, hypothetical numbers: Yesterday there were 84 new conversations against a required 110. The closing rate was 18%, above the assumed 15%. The week is about 9% behind the plan line. Assessment: the shortfall is in the source of conversations, not in closing. Recommendation: look at which content pieces brought in the most enquiries last week and raise the cadence of that type; do not change the advertising budget until each channel has been reviewed.

Two details matter. First, the required level is recalculated every day: the remaining part of the target divided by the days remaining. That way today's shortfall is not hidden until the last quarter; it is added to tomorrow's requirement where everyone can see it. Second, the bot only recommends. It reads broadly and writes narrowly, as I described in AI reads broadly, writes narrowly. Ad autopilot is off by default, every budget increase is capped at VND 1 million a day, and the AI has no permission to move money. The decision belongs to a person.

Response speed sits in the same picture. In the 30 days to 10/10/2026, the median reply time to customer messages was 1 minute, and 80% of messages were answered within 15 minutes. A conversation that opens and is not answered in time is a piece of the "needed every day" number wasted.

What investors and partners should read from this

• Our plans can be challenged. For any target you can ask about the three assumptions beneath it: order value, closing rate, return rate. Each has a measurable number attached.
• The capital we raise attaches to links of the chain. The brand and content portion aims at the number of conversations. The sales team portion aims at the closing rate. The repeat-customer programme aims at how many new customers we must find. A spend that cannot say which number in the chain it moves is a spend not yet understood.
• The limits of the method. The division is a feasibility check, not a prophecy. If an assumption is wrong, the result is wrong. So every quarter we review the assumptions against measurements, and we do not keep an old assumption just because it is already in the plan.

Closing

I still remember the feeling of the first time I divided: a yearly target suddenly had the shape of a Tuesday morning. Since then I have believed something simple. A target that cannot be turned into today's work is not yet a target; it is a wish. At HOPE, every morning there is a number to reach, a number reached, and a recommendation for the gap between them. The company grows from such mornings added together.