Nine times out of ten, when a founder tells us their Instagram ads aren't converting, the ads are fine. The click-through rate is healthy, the CPM is normal for the category, and the creative is doing exactly what it should. What breaks is everything after the tap: a landing page that takes four seconds to paint on a 4G phone, an offer that doesn't match the ad, a COD-heavy order mix that quietly eats the margin, or attribution that makes a profitable campaign look dead. Before you rewrite the hook, do the arithmetic. It usually tells you where the money is going.
Run the numbers before you touch the creative
Here is a month of spend from a mid-priced apparel store, using round figures you can swap for your own.
Spend ₹1,00,000. CPM ₹180, so roughly 5,55,000 impressions. CTR of 1.2% gives about 6,660 clicks, which puts CPC at ₹15. Site conversion rate of 1.0% gives 67 orders. At an AOV of ₹1,450 that is ₹97,150 in revenue. ROAS 0.97. The founder sees that number and starts testing new hooks.
Now the part most people skip. On that ₹1,450 order: COGS ₹500, shipping ₹80, packaging ₹25, payment gateway at 2% is ₹29. Contribution before ad spend is ₹816. Your actual CAC is ₹1,00,000 divided by 67, so ₹1,492. You are spending ₹1,492 to earn ₹816. Every order loses ₹676 and you cannot fix that with a better thumbnail.
Break-even CAC is ₹816. At ₹15 a click, that is 54 clicks per order, which means you need a 1.84% conversion rate. You are at 1.0%. So the question is not "why aren't my Instagram ads converting", it is "where do I find another 84% of conversion rate, or ₹600 of contribution margin". Those are two very different projects, and the second one is often easier.
The in-app browser is a different, slower internet
Instagram traffic does not land in Chrome. It lands in Meta's in-app webview, usually on a mid-range Android phone, usually on mobile data, usually with a cold cache. Whatever your desktop PageSpeed score says, that is not the experience you are buying.
We audit paid traffic on a throttled 4G profile and look at one number: how long until the hero image and the price are on screen. On a stock Dawn build with four or five apps injecting scripts, an LCP of 3.5 to 4.5 seconds is ordinary. Getting that under 1.5 seconds is mostly unglamorous work. Preload the hero image and stop lazy-loading it. Serve the product image at the width it actually renders at instead of a 2000px original. Defer the review widget, the currency switcher, the popup app and the chat bubble until after first interaction. Remove the apps you stopped using in March but never uninstalled, because their script tags are still there.
If you want a baseline before committing to a rebuild, SwiftStore will scan the store, fix what can be fixed automatically and track the score as you keep shipping changes. For theme-level surgery, deferred hydration and app-script triage, that is our speed optimization work, and it is the single highest-leverage thing most paid-traffic stores can do in a fortnight.
Ad says one thing, landing page says another
A reel opens on a specific product in a specific colour with a specific claim. The link goes to the homepage, or worse, to a 200-product collection. The visitor now has to do the work of finding what they just saw. Most of them don't.
Match the page to the promise. If the ad shows the olive linen shirt, the destination is the olive linen shirt PDP with olive preselected. If the ad says "flat 30% off", the price on the page shows the discounted figure, not the MRP with a code the buyer has to remember at checkout. If the ad is a founder talking about how the product is made, the PDP needs that story above the fold, not buried under a size chart.
This sounds obvious. We still find it broken on the majority of accounts we audit, usually because the media buyer and the person who owns the theme are different people who don't talk.
COD is why your Instagram ads aren't converting on paper
Take the same 67 orders. Say 60% choose cash on delivery, so 40 COD orders. If your RTO rate on COD is 22%, that is roughly 9 orders that never complete, plus forward and return freight of about ₹150 each burnt on the way. Revenue drops to about ₹84,000 and you have lost another ₹1,350 in shipping. Meta still reports 67 purchases, because the pixel fired at order confirmation. Your bank account disagrees.
Things that actually move this: a prepaid discount of ₹75 to ₹100 shown at checkout, COD available only above a cart threshold, a COD fee of ₹49 that most buyers will pay a small amount to avoid, and an OTP or WhatsApp confirmation on every COD order before it is packed. Also worth doing: rank your pincodes by RTO and turn COD off for the worst ones. Not popular internally. Works.
UPI is the other half of this. If UPI intent is not one of the first two options on your Razorpay checkout, you are pushing people toward cards they abandon or COD you don't want.
You are probably measuring two different things
Meta's default reporting window credits a purchase to an ad if the click happened within seven days or the view within one day. Shopify credits the last non-direct touch. GA4 does something else again. So Meta says 67 purchases, Shopify's marketing report says 41 from Facebook and Instagram, and everyone spends the standup arguing.
The fix is not to pick a winner. It is to pick one number as the source of truth for money decisions, and use the others directionally. We use Shopify orders as the denominator and total ad spend as the numerator, calculate blended CAC for the month, and hold that against contribution margin. Everything else is for optimising inside the platform.
While you are in there: check that Conversions API is live and deduplicating properly against the browser pixel, and look at event match quality. If it is low, you are sending Meta thin signals and its optimisation is guessing. Passing hashed email and phone from checkout usually lifts it within a week.
Budget spread too thin to learn
₹1,00,000 a month is roughly ₹3,300 a day. Split across six ad sets, that is ₹550 each. At a ₹1,450 AOV and a 1% conversion rate, an ad set at ₹550 a day produces well under one order a day. Meta's optimiser cannot learn from that, so it never exits the learning phase, so performance stays random, so you conclude Instagram doesn't work for your category.
Consolidate. One or two campaigns, broad targeting, five to eight creatives inside a single ad set, Advantage+ placements on. Let the algorithm do the segmentation it is actually good at. Kill the interest stacks you built in 2021.
Creative problems that are real
Some are. The ones we see most: the first frame is a logo instead of the product, so the hook is wasted; the ad is a polished brand film when the account rewards a phone camera and a person talking; there is no price anywhere, so unqualified clicks flood the site and tank the conversion rate; and every variant tests the same idea with a different font.
A cheap diagnostic: look at hook rate, meaning three-second views over impressions, and hold rate, meaning thruplays over three-second views. Weak hook rate is a creative problem. Strong hook rate with weak CTR means the ad entertains but doesn't sell. Strong CTR with weak site conversion sends you back to the landing page, which is where this article started.
Checkout, and when Shopify Plus is not the answer
Standard Shopify checkout converts well. Most abandonment at this stage comes from things you control anyway: shipping cost appearing for the first time on the last step, a coupon field that invites people to leave and hunt for codes, mandatory account creation, and a five-field address form when Shopify can autofill most of it from a pincode.
Founders at ₹8 to ₹15 lakh a month ask us whether upgrading to Plus will fix their conversion rate. Usually the honest answer is no, and the ₹1.8 lakh a month is better spent on landing pages, creative volume and fixing RTO. Plus earns its keep when you need checkout extensibility for custom logic, scripts on shipping and discounts, multiple storefronts or B2B, and at that point our Plus team can scope it properly. Before that, it is an expensive way to avoid the actual work.
Sometimes it is the offer
The uncomfortable case. If your product is priced 40% above three near-identical listings a buyer can find in ten seconds, and the only difference is your packaging, no amount of media buying fixes that. Same for a category with a genuinely long consideration cycle where a cold Instagram click was never going to buy today. In those accounts, the money goes into email and WhatsApp capture, a first-order incentive that clears margin, and repeat purchase, with paid social funding the top of the funnel rather than paying for itself on day one.
We would rather tell a founder that in week one than take a retainer to optimise something that cannot be optimised.
What to do this week
Open a sheet. Column A: last 30 days of spend, Shopify orders from Meta-tagged traffic, AOV, COGS, shipping, gateway fee, RTO rate. Work out contribution per order and actual CAC. If CAC is above contribution, write down the conversion rate you need to break even, then load your top-spending ad's landing page on your own phone through the Instagram app and time it.
That exercise takes an hour and settles most arguments. If you want a second pair of eyes on the numbers and the page, our free audit covers both, and we will tell you if the problem is the ads.