Ecommerce product research is usually taught as a hunt for a magic item, which is why so many Pakistani stores launch with a product nobody wanted and a store owner who cannot work out what went wrong.
It is closer to a filter than a hunt. You remove everything that cannot work in this market, then test what survives.
What Should You Filter Out First?
Start by deleting categories rather than searching for winners. It is faster and it is where the costly mistakes live.
| Filter | Avoid | Why it matters locally |
|---|---|---|
| Weight | Over about 5 kg | Courier charges scale with weight and eat thin margins |
| Fragility | Glass, ceramics | Breakage on local routes, and refusals on arrival |
| Sizing | Fitted clothing, shoes | Wrong size is the top return reason and returns are paid twice |
| Margin | Under about 2.5x cost | Cannot absorb a refused delivery and still profit |
| Explanation needed | Complex technical items | If it needs a demo, a cold ad will not sell it |
| Availability | Sold in every general store | No reason to wait 2 days for delivery |
That last row catches more beginners than any other. If somebody can walk 5 minutes and buy it, your delivery time is a disadvantage rather than a service.
Why Does Margin Matter More Here?
Because cash on delivery charges you for failure.
A refused parcel costs you the delivery leg out, the return leg back, and the packaging, with no revenue against it. So a product with a 40% margin and a 25% refusal rate can be unprofitable while every dashboard shows healthy sales.
The practical rule: price at a multiple of landed cost that leaves room for the parcels that fail. Work it per 100 orders rather than per order, because that is the only way the return rate becomes visible in the arithmetic. The full COD breakdown sets out the sums.
This single constraint eliminates most of the cheap, low-ticket products beginners are drawn to. A 400 rupee item cannot carry a 250 rupee delivery charge twice.
Where Should You Actually Look?
Not at “winning product” lists, which are by definition already saturated by the time they are published.
- Your own irritations. Things you have searched for locally and failed to find easily are a real signal.
- Marketplace search, filtered by recency. What people are actively searching for now rather than what ranks on reputation.
- Running Facebook and Instagram ads in your market. If someone has run the same creative for weeks, it is working. That is free research.
- Wholesale markets. Walk Hall Road or Shah Alam and ask what is moving, not what is in stock.
- Complaints under competitors’ posts. The gap between what is sold and what people wanted is an offer, not just a product.
The fourth one is underrated and almost nobody doing online courses actually does it. A supplier telling you what shifted this month is better data than any tool.
How Do You Validate Without Buying Stock?
Build a real product page and spend a small budget sending traffic at it.
Not a survey. Not asking in a group. Not counting likes. People are generous with opinions and stingy with money, and only one of those predicts a business. The test is whether a stranger who does not know you tries to complete a checkout.
Run it until you have enough results to read rather than for a fixed number of days. A handful of clicks tells you nothing either way, and the most common testing error is spreading a small budget so thin that the outcome is noise presented as a number.
Watch 3 things: whether people reach the product page and stay, whether they start checkout, and what it costs you per attempted order. The third is the one that decides whether this is a business.
Should You Sell Something Nobody Else Sells?
Usually not, and this is where ambition misleads people.
A product with no competition may have no competition because there is no demand. Competition is the cheapest evidence of a market that exists, and for a first product that evidence is worth more than novelty.
Your edge does not have to come from the product. It can come from faster delivery, a clearer page, better photos, confirming orders before dispatch, or simply answering WhatsApp quickly. Those are easier to build than a new category and they are what customers actually notice.
Save the unique product for when you have a working store and can afford to be wrong.
Does Dropshipping Work in Pakistan?
For testing yes, for scaling rarely.
The constraint is delivery time. Under cash on delivery, acceptance falls as the gap between ordering and arrival grows, because enthusiasm decays and the buyer has committed nothing. A product arriving in 2 days gets accepted far more often than the same product arriving in 3 weeks.
So a model built on long international shipping fights the payment method the market runs on. You can test demand that way. You will struggle to build a profitable operation on it.
Holding modest stock of a validated product, dispatched same day or next day, beats it on the only metric that pays you, which is delivered orders rather than placed ones.
How Do You Check Supply Before You Commit?
A product you cannot restock quickly is a liability dressed as an opportunity.
The failure runs like this. You test, it works, you scale the ad spend, and 10 days later you are out of stock with a campaign that was finally profitable. Turning that campaign off and on again resets the learning phase, so you lose the performance as well as the sales.
So before the test, not after, establish 4 things with the supplier. What is the real landed cost at the quantity you will actually buy, not the quantity on their price list. How long does a reorder take. What is the minimum they will sell you. And can they hold stock for you if you commit.
Ask what happens when they run out too. Many small Pakistani suppliers are reselling themselves, and a product with one fragile source is a product you will eventually have to apologise for.
What About Selling Internationally?
It changes the maths in your favour on margin and against you on everything else.
Selling abroad means prepaid payment rather than cash on delivery, which removes the refusal problem entirely and removes the courier float. That is a genuine advantage and it is why some Pakistani sellers go straight to international marketplaces.
What it adds is shipping cost, customs paperwork, longer delivery windows, returns you cannot practically accept, and payment infrastructure you have to set up properly. None of those is impossible and all of them are work.
The usual sensible order is to learn the mechanics locally, where feedback is fast and mistakes are cheap, then take a proven product international once you know it sells. Trying to learn both at once is where people stall.
How Many Products Should You Launch With?
One, properly, rather than twenty hopefully.
A general store with 50 items spreads your attention, your budget and your learning across 50 things you understand shallowly. A single product store forces you to get the page, the offer, the ad and the fulfilment right, and those skills transfer to everything you add later.
Add the second product once the first is profitable and boring. Boring is the signal. It means the system works without you watching it.
What Does a Sensible Research Week Look Like?
- Days 1 to 2. Build a shortlist of 10 products that survive the filters above. Do not fall in love with any of them.
- Day 3. Check supply. Can you get it locally, at what cost, how fast, and can you reorder quickly? A product you cannot restock in a week is a liability.
- Day 4. Work the margin per 100 orders at a realistic refusal rate. Most of the shortlist dies here, which is the point.
- Days 5 to 7. Build 1 real product page for the survivor and start a small paid test.
- Following week. Read the cost per attempted order. Scale, adjust the offer, or go back to the shortlist.
Step 4 is the one people skip because it is arithmetic rather than discovery. It is also the step that prevents most failures.
Common Mistakes to Avoid
- Buying stock before any validation, which converts a research question into a storeroom problem.
- Choosing on excitement rather than on margin and weight.
- Asking friends and family, who will be encouraging and wrong.
- Launching 30 products so that nothing gets proper attention.
- Ignoring restock speed, then going out of stock exactly when an ad starts working.
- Testing on a budget too small to read, and concluding from noise.
- Picking something available in every corner shop.
Final Thoughts
Ecommerce product research is mostly subtraction. Filter on weight, fragility, sizing, margin and whether the thing is already on every shelf nearby, and the field narrows fast.
Then validate with money rather than opinions, and do the per-100-orders maths before you buy anything. A product that survives both is not guaranteed, but it has already passed the tests that kill most Pakistani stores in their first 3 months.
Working through this on a real store, with real suppliers and a real budget, is how our Ecommerce and Shopify Mastery course is structured. Current batches are on the apply page.
Frequently asked questions
How do I find a winning product?
You do not find it, you filter for it. Start from margin, weight, fragility and whether the problem is obvious, then validate demand with a small paid test before buying stock. Most people reverse this and buy first.
What margin do I need in Pakistan?
Enough to absorb a refused delivery and still profit. Under cash on delivery you pay the courier on parcels that come back, so a thin margin that works in theory loses money in practice. Work it per 100 orders, not per order.
Should I dropship or hold stock?
Holding stock usually wins locally, because delivery speed drives acceptance under COD and a long shipping wait produces refusals. Dropshipping suits testing, not scaling, in this market.
How do I know if there is demand?
Spend a small budget sending traffic to a real product page and watch whether anyone tries to buy. Interest surveys, friends and family opinions and social media likes all predict almost nothing.
Is it better to sell something unique or something proven?
Proven, almost always, for a first product. A product nobody sells may be unsold because nobody wants it. Competition is evidence of demand, and your edge comes from offer, service and speed rather than from novelty.
How much should I spend testing?
Enough to generate a meaningful number of results rather than enough to feel safe. A tiny budget spread over a week produces noise you cannot read, which is worse than not testing.