Facebook and Instagram ads for a small online store in Pakistan

A small store can lose money on Facebook and Instagram ads even when the ads work. The ads send people; the store isn't ready to turn them into delivered, paid orders, or the owner judges the results by likes instead of margin.

This guide is for a store running its first paid campaigns on Meta's platforms. It won't quote costs per click or returns on ad spend: those depend on your product, audience, season and creative, and a number from someone else's campaign will mislead you. Ads Manager's menus and options change too, so this sticks to the principles that don't. It covers when you are ready, what has to exist before you spend, how to structure and budget a first campaign, how to read the results, and when to stop or scale.

Get organic sales first

Ads amplify what already converts. If the people who find you through your own Instagram, WhatsApp and word of mouth aren't buying, paid traffic won't change that. It will just lose money faster.

You are ready to advertise when:

• Strangers have bought, not only friends and family.
• The product page answers the questions people used to ask you in DMs.
• You have reviews from real buyers.
• Returns are under control, and you know the return rate of the product you will advertise.
• You can handle more orders — stock, packing and dispatch — without falling behind.

If you are not there yet, getting your first 100 orders and selling on Instagram come first.

What has to exist before you spend

A pixel on your store. Without one, Ads Manager can't connect a purchase on your store to the ad that led to it, and you are left judging ads on clicks. On Storecrafto you create a pixel in Meta's Events Manager (it may be labelled a dataset there), generate a Conversions API access token for it, and paste both into your store's Marketing page. The store then reports add-to-cart, checkout and purchase events to your own ad account, so purchases are attributed in Ads Manager and the data stays yours. Place a test order and check that the purchase shows up in Events Manager before you launch anything — the setup is in the ad tracking docs.

A checkout that works on a phone. Buy something yourself on mobile data with each payment method you offer. Ads bring impatient strangers, and they won't forgive friction you have learned to live with.

COD confirmation. Some orders from strangers will be impulse buys, typos and fakes. Every one you dispatch unconfirmed costs freight both ways, on top of the ad money already spent to get it. Confirm cash-on-delivery orders before dispatch — on the Growth and Pro plans, Storecrafto can ask for that confirmation on WhatsApp automatically. The full method is in how to reduce COD returns.

Your break-even number. Work out what a delivered order leaves you after the product, packaging, delivery, gateway fees and your share of returns. That is the most you can spend on advertising per delivered order before each one loses money, and without it you can't read any result. How to price products online covers the sum.

Someone to answer. Ads produce comments and DMs — "price?", "delivery to Quetta?", "COD available?" Every one left unanswered is a click you paid for and wasted.

A simple first campaign

Keep the first campaign boring. Every extra moving part makes the result harder to read.

• One objective: sales. Pick the objective aimed at purchases, with the purchase your pixel reports as the result you are paying for. Objectives built around traffic or engagement get you clicks and likes, and you can't bank either.
• One product, or a few that belong together. Send people straight to the product page, not your homepage.
• One or two audiences, not ten. Splitting a small budget across many audiences leaves each with too little spend to tell you anything.
• Target where you deliver well — the cities and areas where your courier is reliable and your buyers already come from, not the whole country by default.
• Creative that shows the product in use: on a person, in a home, in a hand. Video shot on a phone is fine, and it looks like the posts around it. Show the product in the first seconds.
• Put the price and delivery terms in the caption, so the "price?" comments answer themselves.
• Two or three genuinely different creatives — a demonstration, an unboxing, a customer's review — so you learn which angle works.
• Write in the language your buyers use, whether that is English, Urdu or Roman Urdu. Your DMs will tell you which.

The basics of shooting it yourself are in product photos without a photographer.

Budget: small, daily and time-boxed

Decide the test budget before you start, as an amount you can afford to lose completely. If losing it would hurt the business, it is too big for a first test.

Then set a modest daily budget and a fixed test window — at least a full week, so it covers weekdays and a weekend — and leave the campaign alone until the window ends. Judge it once, at the end, against your break-even number.

Size the test so it can actually tell you something: it needs enough budget to produce several purchases at your break-even cost. A test that can only afford one or two can't tell a good ad from a lucky one.

Reading the results

Likes, reach and comments feel like results. They aren't. The numbers that matter:

• Cost per purchase, from Ads Manager: what you spent divided by the purchases your pixel recorded.
• Cost per delivered order, which you work out yourself. A cash-on-delivery order counts as a purchase the moment it is placed, including the ones that will be refused at the door, so divide your spend by the orders that actually arrived and were paid for.
• Margin after returns, set against that cost. That comparison is the only verdict that counts.

Illustration only, with made-up round numbers: if an ad set shows ten purchases and three of those COD parcels come back, you paid to get ten orders, earned a margin on seven and paid return freight on three. The ad set that looks cheapest in Ads Manager isn't always the one that made money.

The rest of the path tells you what to fix:

• Plenty of people see the ad, few click: the creative isn't stopping them.
• They click, few add to cart: the product page, the price or the trust signals aren't convincing.
• They add to cart, few buy: the checkout, the delivery charge or the payment options — see why buyers abandon carts.

On Storecrafto, the store's analytics separate visits from paid ads from search and social visits, and the funnel shows where visitors drop out between arriving and buying.

Retargeting people who didn't finish

Shoppers who added to cart or reached checkout without buying are your warmest audience: they already wanted the product. Because your pixel reports those events, you can build an audience of them in Ads Manager and show them a follow-up ad.

• Keep it recent and small — people who left in the last few days, not everyone who ever visited.
• Exclude buyers. An ad for something a customer just bought wastes money and annoys them.
• Answer the objection with reviews, your delivery promise, "COD available" or your WhatsApp number — whatever might have stopped them.
• Don't lead with a discount every time. If leaving always earns a discount, buyers learn to leave.

This runs alongside your store's own abandoned-cart reminder emails, which reach the buyers who left an email without costing you any ad spend.

The traps that burn money

• Boosting posts because they got likes. A popular post isn't necessarily one that sells, and a quick boost may not be aimed at purchases at all. Check what you are paying for, or build the campaign in Ads Manager.
• Targeting the whole country. You pay to reach places your courier serves poorly and your delivery promise can't cover.
• Changing everything daily. Change the audience, budget and creative every morning and you will never know which change did what.
• Ignoring returns. A campaign that looks profitable on purchases can lose money on delivered orders.
• Spending before the pixel works. Without purchase data you are judging on clicks, and clicks don't pay.
• Sending ads to the homepage. The buyer came for one product; make them hunt and they leave.
• Leaving comments unanswered. Every "price?" left hanging is a paid click walking away.
• Advertising what you can't restock. An ad that works is the worst possible moment to run out.

When to stop, and when to scale

Stop, or pause, when the test window ends with your cost per delivered order above break-even and no sign of improving, or when orders arrive but refusals eat the margin. Before you stop, find which step failed — the ad, the product page or the checkout. The fix is different for each, and the product may deserve a second test with that one thing fixed.

Scale when your cost per delivered order stays comfortably below break-even across the whole window. Raise the budget in steps rather than all at once, and judge each step over its own window, because what worked at a small budget doesn't always hold at a larger one. Make sure the rest of the store can keep up — stock, packing, confirmations, courier pickups — and keep new creative coming, since the same ad shown to the same people tends to wear out.

The short version

• Ads amplify what already sells — get organic sales and real reviews first.
• Before you spend: a working pixel, a tested checkout, COD confirmation and your break-even number.
• Start with one sales objective, one or two audiences and creative that shows the product in use.
• Set a budget you can afford to lose and a fixed test window, and leave it alone until the end.
• Judge on cost per delivered order against margin after returns — not likes, not clicks.
• Retarget recent cart abandoners with the answer to their objection, not an automatic discount.
• Scale in steps, and only as fast as your stock and dispatch can follow.