Importing on instinct is common: you spot a product selling somewhere, order a few hundred units, and wait. The problem is not the decision itself. It is that the decision has no data behind it. This is about measuring real demand before your money sits in a warehouse, using search signals, platform sold counts, and small controlled pre-tests.
Why instinct is not enough
When you place a first order from 1688, the time from placing the order to stock landing in your warehouse is typically 20 to 35 days, depending on the shipping route and product type. During that window you cannot adjust anything. Goods on the way are goods on the way.
If you ordered because you "saw a lot of people buying on TikTok" or "a friend said it is selling well," you are forecasting demand on someone else's impression, not your own data. That impression might be right. It might also be reflecting a peak you just missed, while your shipment is still at sea.
Validating demand does not remove risk entirely. It replaces instinct with something more concrete before your capital is locked.
Measuring demand through search signals
Search volume is unmet demand. When someone types a product keyword, they are signaling intent to buy, not just interest.
Google Trends is the first free tool to use. Type the product name in Vietnamese (the way real buyers search, not the technical name), set the region to Vietnam, and read the 12-month chart. You are looking for two things:
- A trend that is rising or still early. If the line has been climbing over the past three to six months, more people are discovering the product. If the line is flat or declining, the fast-growth phase may already be behind you.
- Year-round versus seasonal demand. Steady search volume across the full year means demand does not depend on a specific occasion. A sharp seasonal spike means you have to import and sell within a narrow window. Miss it and you hold inventory for another year.
Beyond Google Trends, check the autocomplete suggestions in the Shopee and TikTok Shop search bars. Those suggestions reflect the real keywords users type most. If typing half the product name already surfaces it at the top of the list, search volume is meaningful.
Reading platform sold counts to confirm real buying
Search signals tell you people want to find something. Sold counts tell you people are actually paying money. Both signals need to point the same direction.
On Shopee, open the product page and check the units sold figure (usually displayed under the product name, like "1.2k sold"). On TikTok Shop, browse the product tab for each shop and note the order count. Rather than looking at just one shop, run through five to seven different shops selling the same item and write down the numbers.
Three things to watch:
- Whether sold counts are spread or concentrated. If one shop has 10,000 orders and every other shop combined is under 200, that demand is tied to the channel, not the product. If five shops each have a few thousand orders, demand lives in the product and there is a lane for you.
- Recent velocity, not lifetime total. A product with 5,000 cumulative orders over three months but a flat week-over-week count may already be at its ceiling. What you need to know is how many units it is selling right now, not historically.
- The price band. Record the selling prices across those shops. If prices are clustered at the low end, the market may already be squeezed. If the band is still wide, there is room to enter at a reasonable price without racing to the bottom.
Build a simple table: product name, platform, shop, selling price, estimated recent order count. Five to seven rows of that data give a usable picture.
Cross-check both platforms
Durable demand usually leaves marks on both Shopee and TikTok Shop, not just one. TikTok Shop is where demand appears early and sharp. Shopee is where it tends to stay longer and level out. When both show a signal, you are looking at demand with a floor.
If TikTok is hot but Shopee shows almost nothing, the most likely explanation is a video-driven trend that will cool once the clip leaves the algorithm. This risk is highest when all the sales volume on TikTok traces back to one or two large creator accounts.
The reverse is a better position: Shopee selling steadily for a while, with TikTok just starting to pick up. The demand has been proven on the older channel and is now getting amplified. That is worth a closer look.
Run a small pre-test before committing to a large batch
Even when signals look good, you still do not know whether the product will sell from your shop, at your price, with your images and copy. The only way to find out is to test with real inventory.
A small pre-test is not about ordering a handful because you are cautious. It is about ordering enough to get a statistically reliable sales signal. For a mid-priced product (in the VND 100,000 to 300,000 retail range), at least 30 to 50 completed orders is the minimum needed before the data means anything. Fewer than that and the noise drowns the signal.
During the test, track three specific numbers:
- Listing conversion rate. How many page visits versus how many orders placed. If you get views but no orders, the issue is in pricing, photos, description, or the product itself not convincing buyers.
- Return and cancellation rate. High orders paired with high returns is a bad signal. Causes include product quality lower than expected, images that misrepresent the item, or a product that does not match the actual need.
- Ad cost per order, if you run paid traffic. If you need to spend heavily to pull each order, the product's real margin cannot support scale.
After the test, you have three outcomes: selling well with decent margin means expand; selling but needing price or image adjustments means adjust and re-test; not selling or margin too thin means stop and move capital to another SKU. All three outcomes are better than importing a large batch with no evidence at all.
What 1688 data can tell you about demand
Before locking in a supplier, you can also read signals from 1688 itself. These are not retail demand signals; they are signals about import demand from other wholesale buyers.
On 1688, find the product and look at the transaction count (the sold figure shown on the listing) over the past 30 days. A high number with buyers from many different locations confirms real purchasing activity. This cuts two ways: it validates that demand exists, but it also signals that competition on the supply side is strong.
Also note how many suppliers are selling the same product. If dozens of suppliers list the identical item at nearly the same price, the barrier to entry is near zero. A late importer has no cost advantage and can only compete through marketing or product differentiation.
Signals have limits
No method guarantees you will not make a wrong call. Platform data is a record of the past, and you are betting on the future. A few things to keep in mind:
- Sold counts on platforms do not separate organic sales from ad-driven volume. A shop spending heavily on ads can produce impressive numbers that do not reflect natural market demand.
- TikTok signals move fast. A product can travel from peak to decline in two to four weeks when it is purely trend-driven. That window is shorter than the lead time for importing from 1688.
- Seasonality can make signals look better than they are. If you research a product right at the peak of its season, the numbers look strong. Three months later when your stock arrives, demand may have retreated.
These limits do not mean you should ignore the data. They mean the data should be read in context, not mechanically.
Bottom line
Validating demand before importing replaces the question "will this sell?" with concrete evidence rather than a guess. Search signals, sold counts spread across multiple shops, a two-platform cross-check, and a controlled small-batch test: those four layers together do not give you certainty, but they give you a foundation for committing capital that instinct alone cannot provide.