Facebook ads for Philippine online sellers: the money leaks at the door, not at the click

Facebook reports a purchase when the order is placed. In a market that runs on cash on delivery, the money arrives days later, when a rider hands over the parcel and the customer pays. Those are two different events, and the gap between them is where a lot of Philippine sellers quietly lose their margin.

This matters because sellers optimise what they can see. If your cost per purchase looks acceptable, you scale. But if some of those purchases never convert into cash, you have scaled the wrong number, and the loss grows in proportion to how well the ad performed. A good campaign can lose more money than a mediocre one.

Two figures get cited, and they disagree because they measure different things. A Rakuten Insight survey found that 86% of Philippine respondents said they had paid via cash on delivery for online shopping. An FIS survey put cash on delivery at 27% of online purchases. The first is a share of shoppers who use it at all, the second a share of transactions. Both can be true at once, and together they say the same thing: this is not a fringe payment method you can design around.

You will also see a widely repeated claim that cash-on-delivery orders return or fail at some specific rate, usually a range in the teens or twenties. We went looking for the primary source behind that figure and could not find one. Individual seller anecdotes exist, but we are not going to restate someone else's unsourced number as though it were measured. Your own failure rate is knowable and specific to you, and it is the only one worth acting on.

Before changing anything about your ads, work out the one ratio that matters. It takes a spreadsheet and an afternoon.

• Take one month of orders that came from ads. Count them.
• Count how many of those were actually delivered and paid.
• Divide your total ad spend for that month by the number of paid orders, not by the number of orders. That is your real cost per customer.
• Compare it to what your ads manager told you. The difference between those two numbers is what the delivery step is costing you.

Most sellers we talk to have never run this calculation, because the two halves of it live in different places: the ad platform on one side, the courier dashboard and a notebook on the other. Nothing about it is difficult. It is just nobody's job.

Once you know the size of the leak, the fixes are unglamorous and mostly free. Two messages carry most of the value, and both fail for the same reason: they depend on a human remembering, at exactly the moment the business gets busy.

• Confirm before dispatch. A short message asking the buyer to confirm the order and address before it ships. This catches the impulse order, the wrong address, and the buyer who has changed their mind, while it still costs you nothing but a message. A parcel you never sent cannot fail delivery.
• Follow up the moment delivery fails. A failed attempt is not automatically a lost sale; often the person was at work. But the window is short, and by the time the parcel is back at the hub, the sale is usually gone.

Both are worth doing manually first. If confirming orders by hand for two weeks visibly reduces your failed deliveries, you have proven the mechanism works for your business before spending anything on automating it. If it does not move, automating it would only have made a useless step faster.

With the delivery leak measured and the two messages running, your cost per paid order becomes a real number, and advertising decisions get much easier. You can tell which products survive the door and which ones attract buyers who refuse on arrival. Those are usually not the same products, and it is a distinction no amount of audience testing will surface for you.

It also changes what you optimise for. Some sellers find that pushing prepaid options, even with a small discount, is worth more than any creative improvement, because a prepaid order is finished. Others find one product line is responsible for most of their refusals. Neither insight is available while your only visible number is cost per purchase.

We build this kind of automation, and we are based partly in Manila, so this is the market we watch most closely. But the honest order of operations is the one above: measure the leak, run the two messages by hand, and automate only the part that has already proven it works. Anyone who tells you to start by rebuilding the funnel is selling the rebuild.

• Cash-on-delivery adoption figures are as reported by Rakuten Insight (86% of Philippine respondents had paid by cash on delivery) and an FIS survey (27% of online purchases), both as cited publicly. They measure different populations and should not be averaged or combined.
• We deliberately excluded the commonly quoted cash-on-delivery failure and return percentages, because we could not trace them to a primary study. If we cannot see a number in its source, it does not appear here.
• The measurement method above is arithmetic, not research. It gives you your own figure, which is the one that should drive your decisions.
• Nothing on this page is a claim about results we have produced for a client.