Meta ads for a new Shopify store: the first 90 days

Before you spend ₹1: get tracking right

Meta's delivery system learns from the events your store sends it. If those events are wrong or missing, you're paying it to learn the wrong thing.

Checklist:

• Meta Pixel installed through Shopify's Facebook & Instagram channel (or your theme), firing ViewContent, AddToCart, InitiateCheckout and Purchase.
• Conversions API turned on, so events are sent server-to-server as well as from the browser — this recovers events that browsers and ad blockers drop.
• Domain verified in Meta Business Manager.
• Purchase value passing correctly (test with a real order).
• GA4 running alongside, so you have a second view of what's happening.

COD needs a decision: a COD "purchase" fires when the order is placed, not when cash is collected. If your COD returns are high, Meta will happily optimise for buyers who later refuse delivery. Two fixes: confirm COD orders on WhatsApp, and judge results on delivered revenue in your own reports, not only on Meta's dashboard.

The budget maths that matters

Meta says an ad set usually exits the learning phase after about 50 optimisation events in a 7-day window. Until then, results are unstable. If it never reaches that, it shows Learning Limited — not a penalty, but a sign the system doesn't have enough data.

So size budget from the event you optimise for:

Daily budget per ad set ≈ expected cost per event × 50 ÷ 7

Example: if you expect a purchase to cost ₹700, one ad set needs about ₹700 × 50 ÷ 7 = ₹5,000 a day to exit learning on purchases. Many new stores can't do that yet. Two honest options:

1. Consolidate. Fewer ad sets, each with more budget. Two well-funded ad sets beat eight starved ones.
2. Optimise for a more frequent event first (for example, add to cart), then switch to purchases once you have volume.

Days 1–14: foundation and creative testing

Goal: find creative that stops the scroll and gets the right people to the store.

• One prospecting campaign, broad targeting (age, country and gender only if your product requires it). Let Meta find buyers — it's usually better at this than narrow interest stacks.
• Test creative, not audiences. 4–6 genuinely different ideas: a founder video, a problem/solution demo, a customer review, a product-in-use reel, a static with a single bold claim.
• Keep the store ready: fast product pages, clear price, delivery and return information, COD availability visible.
• Don't edit daily. Big changes (targeting, optimisation event, large budget jumps, adding ads) restart learning.

What to watch: thumb-stop rate (3-second views ÷ impressions), click-through rate, cost per landing page view, add-to-cart rate. Purchases will be noisy this early.

Days 15–45: find the offer that sells

Goal: turn attention into orders at a cost you can live with.

• Kill clear losers, keep winners running. Add 2–3 new creatives a week based on what worked (same angle, new execution).
• Test offers, one at a time: free shipping threshold, bundle, first-order discount, prepaid discount. The offer often moves results more than the ad.
• Start retargeting people who viewed products or added to cart in the last 7–30 days, with reviews, FAQs and urgency that's real (stock, delivery time).
• Try Click-to-WhatsApp for considered or higher-priced products. Many Indian buyers want to ask one question before they pay. Reply fast — an AI or automated first reply helps outside working hours.
• Fix the store, not just the ads. If add-to-cart is high but checkout completion is low, the problem is on the site.

Days 46–90: scale carefully

Goal: spend more without breaking what works.

• Raise budgets gradually — small steps every few days rather than doubling overnight, which can restart learning.
• Duplicate winners into a new ad set when you want a big jump, rather than shocking the original.
• Keep the creative pipeline going. Winning ads tire. Plan new ones every week.
• Retention starts paying back: repeat-order reminders, review requests and loyalty bring cheaper second orders, which is what makes paid acquisition profitable.

How to judge results honestly

• Contribution margin per order (after product cost, shipping, payment fees and returns) — not ROAS alone.
• Delivered revenue, not placed orders, if you sell COD.
• Blended cost per acquisition across all channels, because Meta's attribution and your store's numbers will never match exactly.
• New-customer revenue vs returning — scaling should grow new customers.

The 90-day checklist

• Pixel + Conversions API verified with a real test order
• COD confirmation running (if you sell COD)
• One prospecting campaign, broad, few ad sets
• 4–6 different creative ideas live, 2–3 new each week
• One offer test at a time
• Retargeting on from week 3
• Budgets raised in small steps
• Weekly report on delivered revenue and contribution margin

We run this playbook as part of our Meta ads service and the Growth System that sits behind every store we build. Want the store, the tracking and the ads from one team? Build your package.

Sources

• Meta for Business — Learning phase guidance (about 50 optimisation events in 7 days; Learning Limited), summarised by AdManage and Lebesgue (what restarts learning)

Checked in November 2026. Meta changes delivery and attribution regularly; treat budgets as starting points.