The thing that drives sales on Instagram is a piece of creative that shows the product being used, pointed at a product page that loads in under two seconds, with a retargeting pool behind it. That's most of it. Follower count, posting cadence, hashtag strategy, Reels trends, the grid aesthetic: all of that sits somewhere between mildly useful and irrelevant. If your Instagram marketing for ecommerce is built around a content calendar rather than around a creative testing loop and a paid budget, you are running a media brand that happens to sell things, and the economics of that are terrible.
We build and run Shopify stores for D2C brands in India, the UAE, the US and Australia, and the pattern is consistent across all of them. The stores that make money from Instagram treat it as a performance channel with a content input. The stores that don't treat it as a branding exercise and then wonder why the revenue line is flat after eight months of daily posting.
What Instagram is actually good at
Three jobs, in order of how much revenue they produce.
Demonstration. Products that are hard to explain in a static listing sell well here. A skincare texture, a saree drape, a folding pram, a kitchen appliance doing the annoying task fast. Fifteen seconds of the thing working beats any amount of copy.
Discovery at scale, paid. Meta's delivery system will find buyers for you if the creative gives it enough signal. That's the whole machine. You are not buying an audience any more, you are buying a chance for the algorithm to figure out who responds to a specific video.
Retargeting and warm-audience closing. Cheap, high-ROAS, and the reason your prospecting numbers look better than they are. Be honest with yourself about how much of your reported return is just people who were going to buy anyway.
What Instagram is bad at: considered high-ticket purchases with long research cycles, anything that needs a comparison table, and building durable demand that survives you turning the spend off. It does not compound the way search content does. Stop paying and the traffic stops that afternoon.
Run the arithmetic before you run the ads
Most founders we talk to know their ROAS and nothing underneath it. Here's the calculation that matters, with numbers you should replace with your own.
Say you put ₹1,00,000 into prospecting in a month. At a ₹250 CPM you buy 4,00,000 impressions. A 1.2% click-through rate gives 4,800 sessions, so your effective cost per click is about ₹20.80. Your product page converts at 1.4%, which is realistic for cold Instagram traffic on mobile. That's 67 orders. At an average order value of ₹1,850 you've booked ₹1,23,950. ROAS 1.24. Looks survivable.
Now apply reality. Sixty percent of those orders are COD, so 40 orders. RTO on cold-traffic COD runs high; assume 25%, so 10 parcels come back. You're left with 57 delivered orders worth ₹1,05,450. The 10 returns still cost you forward and reverse shipping, roughly ₹120 each, so subtract ₹1,200. At a 55% gross margin your contribution is about ₹56,800, against ₹1,00,000 of ad spend. You lost ₹43,000 and your dashboard said 1.24.
Two levers fix this and neither is a new audience. Push landing page conversion from 1.4% to 2.2% and you get 105 orders instead of 67, which at the same margin and RTO rate turns the loss into roughly a break-even first order with a repeat purchase to come. Or cut COD share by making prepaid genuinely attractive: a 5% prepaid discount costs you less than one RTO in three. Razorpay's UPI intent flow on mobile converts well enough that a lot of the COD habit is just default behaviour, not preference.
Do this arithmetic monthly. If you can't state your delivered-order contribution margin, your ROAS target is a guess.
Creative is the targeting
Interest targeting stopped mattering years ago. Broad audience, one campaign, let the system learn. The variable you control is what you feed it.
Practical rules from running these accounts:
- Test concepts, not variations. Five different hooks beat five colour grades of the same video. A concept is a distinct reason to buy: price, speed, before-and-after, objection handling, someone else's opinion.
- The first two seconds decide everything. Open on the product doing something or on a face saying a specific sentence. No logo card, no slow pan.
- Shoot vertical, shoot on a phone, keep it under 20 seconds for prospecting. Polished brand films underperform against a founder holding the product in reasonable light.
- Burn-in captions. A large share of the feed is watched muted.
- Expect a 1-in-6 hit rate at best. Budget for the misses. If you are producing four videos a month you are not testing, you are hoping.
Festive season in India distorts all of this. From roughly the second week of September, CPMs climb as every brand in the country bids at once. Plan creative production in August, not October, and accept that your September ROAS target should be lower than your July one because the auction is more expensive, not because your ads got worse.
The landing page is part of the ad
This is where most Instagram marketing for ecommerce quietly leaks money. You spend ₹20 to buy a click and drop it onto a homepage carousel, or onto a product page carrying nine apps, a hero video and a review widget that blocks render.
A user coming off a Reel is on 4G, on a mid-range Android, with about three seconds of patience. If your Largest Contentful Paint is 3.5 seconds, a meaningful chunk of the traffic you just paid for never sees the product. We've taken product pages from 2.8s LCP down to under 1.6s by removing two abandoned apps, deferring the review script, and serving properly sized images, and the conversion difference on paid mobile traffic was larger than anything we achieved by rewriting the copy.
Some specifics that matter more than they should:
- Send traffic to the product page, not the collection or the homepage. The ad already did the selection.
- Match the creative. If the video showed the blue one, the page opens on the blue one.
- Price, delivery timeline and return policy above the fold. Indian buyers scroll for those three before anything else.
- One primary button. Buy Now over Add to Cart for single-product ads, because the cart page is a leak.
If you want a number before you commit to work, run SwiftStore against your store to see where the score sits and what's dragging it, or read how we approach Shopify speed optimization when the theme itself is the problem.
Creators: pay for the footage, not the follower count
The most reliable use of a creator budget is buying content you can run as ads. A creator with 12,000 followers who makes good vertical video is worth more than one with 300,000 who posts once and disappears, because you can put paid spend behind the asset and reach whoever you want.
Get usage rights in writing, for paid media, for at least six months. Ask for the raw files. Whitelisted posting from the creator's handle often outperforms the same video from your brand handle, which tells you something uncomfortable about how much your brand identity is worth to a stranger scrolling at 11pm.
Barter works below roughly ₹2,000 product value and stops working above it. Above that, pay cash and treat it as a production cost.
Organic Instagram: what it's honestly worth
Keep the profile alive because people check it before buying, especially for a brand they've never heard of. Nine recent posts, clear product shots, a working link, replies in the DMs. That's the bar for trust.
Chasing organic reach as a growth strategy is a poor use of a small team's hours. Unless you have a genuine content talent in-house, the same effort spent on creative testing and product page work returns more. We say this to clients who have hired a social media manager and are disappointed with the result, and it is rarely welcome.
The exception is categories where the founder is the product: coaching, styling, niche food, anything where personality carries. There, organic can carry real volume. You'll know if that's you.
Measuring it without lying to yourself
iOS attribution being what it is, Meta's reported ROAS overstates. So does Shopify's last-click. Use three sources and triangulate:
- A post-purchase survey on the thank-you page asking how they heard about you. One question, five options. Directionally more useful than any pixel.
- Blended MER: total revenue divided by total ad spend across all channels. Crude, honest, hard to game.
- A holdout test. Turn Instagram spend off for seven days in a quiet week and watch total orders, not attributed orders. Uncomfortable, informative.
Set UTMs properly on every link so Shopify's reports at least agree with themselves. And check GST treatment on your ad invoices if you're billing from India, because reverse charge on Meta invoices catches people out at filing time.
What the first 30 days should look like
Week one: fix the product pages you intend to send traffic to. Speed, images, delivery promise, prepaid incentive. Nothing else.
Week two: produce eight pieces of creative across four concepts. Phone camera is fine.
Weeks three and four: one broad prospecting campaign, one retargeting campaign, no interest stacking, budget you can afford to lose in full. Kill creative that doesn't clear a 1% CTR after 8,000 impressions. Keep what works, make three variants of it.
Then repeat, forever. That loop is the strategy. Everything else is decoration.
If your Instagram numbers look fine at the top and bad at the bank, the problem is usually on the store, not in the ad account. Send us the URL through a free audit and we'll tell you which of the two it is.