Plan your first Meta ads campaign with AI — a ₹300-a-day plan for a small Indian store.

Before you spend a rupee: what ₹300 a day actually buys

₹300 a day is ₹2,100 a week and ₹9,000 a month. That is a real amount of money for a small store, so be clear about what the first week is for. It is not a profit plan. It is a purchase of information: which product strangers stop for, which hook they click, and what a click costs you. Treat the ₹2,100 as tuition and you will make better decisions with the next ₹2,100.

Meta sells impressions, priced per thousand (the CPM you will see in Ads Manager). The price moves with the audience, the season and how much people like your creative — big brands push it up in October and November. Nobody can honestly tell you your CPM before you run; Ads Manager tells you after the first day. The arithmetic is simple once you have it: daily budget ÷ (CPM ÷ 1000) = impressions per day. If your CPM turns out to be ₹200, ₹300 buys about 1,500 people seeing the ad each day.

The arithmetic: your break-even ROAS comes from your margin

ROAS means return on ad spend — sales divided by ad cost. A ROAS of 3 means ₹3 of sales for every ₹1 of ads. Everyone quotes ROAS, but the only number that matters for your store is the one where you stop losing money, and that depends entirely on your margin. The formula is: break-even ROAS = selling price ÷ profit per order before ads. Work it out for every product you plan to advertise. Here is the arithmetic with made-up round numbers — replace every line with your own.

Two things fall out of that table. First, a product with thin profit — say under ₹150 an order — is a bad candidate for a first campaign, because a single wasted click cluster wipes it out. Bundle it with something, or pick another product. Second, 'six orders in a week' is your honest finish line for the test. Write it on a sticky note. If the campaign gets there, it is working. If it gets to three, it is close and worth iterating. If it gets to zero with money spent, something upstream (product, price, page) needs fixing before more ads.

Pick three products — your real best-sellers, not your favourites

Advertise what already sells. A best-seller has proven demand, it usually has reviews, you know the questions buyers ask about it, and you know it ships without drama. Your newest product or the one you are proudest of has none of that. Choose three, not one — the test needs a comparison — and not ten, because ₹300 cannot teach you anything about ten products.

1. Open your last 60 days of orders. Rank products by number of orders, not by revenue. Repeat orders from strangers (not friends and family) count double.
2. Cross out anything below your margin floor. Use the table above. A best-seller with ₹90 of profit is a best-seller you keep on the store, not one you pay to promote.
3. Check stock and shipping. Enough units for a week of orders, and a weight your courier rate is kind to. A 2 kg product that costs ₹180 to ship is a different business from a 300 g one.
4. Prefer a price a stranger pays without a phone call. Under about ₹2,000 for a first campaign. Above that, buyers want a conversation first — which is fine, but it is a different campaign (a WhatsApp one).
5. Prefer products that photograph clearly. The ad is a photo or a short video. A product that needs explaining loses on a two-second scroll.

Write five hook angles with AI (the prompt to copy)

The hook is the first line of the caption and the first two seconds of the creative. It is the only part most people see. You want five hooks for each product, each from a different angle, so the test tells you which angle your buyers respond to. AI is genuinely good at this job as long as you feed it real material — your reviews, your WhatsApp chats, the objection you hear most — and forbid it from inventing claims.

• Problem → relief: name the small annoyance your product removes. "Kurta that survives the washing machine."
• Price reveal: the price, plainly, early. Buyers who scroll past a ₹899 price were never going to buy — that is a good filter, not a lost customer.
• Occasion or gift: tie it to a moment — Diwali gifting, a wedding week, a first-day-of-college. Festival season makes this the strongest angle from September onwards.
• What's inside / how it's made: the material, the maker, the small detail. Works for handmade, food and beauty products.
• Objection answered: "Ships in 3 days. COD available." Boring, and often the best performer, because the objection is what stopped the last ten people.

Skip a social-proof hook unless it is true and specific — "rated 4.6 by 38 buyers" only if that is what your reviews page says. Meta and your buyers both punish invented numbers. If you have collected reviews on your store, the review text is the best raw material you own for every one of the five angles; there is a guide to collecting product reviews.

Creative formats that work on a phone budget

You do not need a studio. You need four kinds of creative, all of which a phone can make in an afternoon. Make one of each for your strongest product and two for the others, and let the hooks above become the on-image text and the first line of the caption.

Audiences for a beginner: broad first, retargeting second

The single biggest mistake with a small budget is splitting it. Five ad sets at ₹60 each teach you nothing, because none of them gets enough impressions to be a fair test. One campaign, one ad set, ₹300, three to five ads inside it. Meta picks the winners for you within the ad set.

1. Objective: Sales. Conversion location: Website. Conversion event: Purchase. This needs your Pixel connected before you start (next section).
2. One ad set at ₹300 a day. Location: India, or only the states you ship to comfortably. Age 18–65 unless the product says otherwise. Leave Advantage+ placements on so Meta can use Reels, Stories and feed.
3. Audience: broad, with one or two interests at most. For a first campaign, broad targeting plus at most one or two obvious interests ("ethnic wear", "home decor") is enough. Interest stacking is where beginners waste a week. Meta's Advantage+ audience option is fine to leave on.
4. Three to five ads, one per hook. Same product, different hook and creative in each. This is the actual test.
5. Do not create a second ad set for retargeting yet. Retargeting needs an audience of past visitors that you have not built. Give it two or three weeks of traffic first.

Retargeting is the second campaign, and the one most first-time sellers never get to. It shows ads only to people who already visited your store, looked at a product, or added something to the cart and left. Those people are far cheaper to convert than strangers. It works through your Pixel: Meta builds a Custom Audience from the events the Pixel fires — site visitors in the last 30 days, ViewContent, AddToCart in the last 7 days, excluding anyone who fired Purchase. Meta refuses to run ads to an audience that is too small, which is why you wait a couple of weeks of traffic first.

Measurement: make 'Purchase' mean a real, paid order

A browser pixel on its own misses sales. It fires from the thank-you page, so it misses the buyer whose page never loaded, the one on an ad blocker, and much of iOS. When it misses, Ads Manager under-reports and you turn off ads that were working. The fix Meta recommends is the Conversions API: your store's server tells Meta about the order directly from the order record, and the browser and server events carry the same event ID so Meta keeps only one. That last part is deduplication, and it is what stops one order being counted twice.

On a SitesPlaced store the authoritative Purchase is sent from the server with the order ID as the event ID. For prepaid orders through Razorpay, PhonePe, Cashfree or Stripe it is held until the payment is confirmed, so a buyer who opens the payment page and leaves is not counted as a sale; cash-on-delivery and WhatsApp-checkout orders count when they are placed. The browser fires a matching Purchase with the same ID and Meta drops the duplicate. If Meta is unreachable the event is queued and retried automatically. Customer details are hashed on the server before they leave.

1. Verify before launch. Set a test event code from Events Manager → Test Events, place a test order on your live store, and watch the browser and server events arrive paired. Remove the test code afterwards.
2. Add the right columns in Ads Manager. Amount spent, Purchases, Cost per purchase, Purchase ROAS, Adds to cart, Link CTR, CPM. Save it as a column preset so you look at the same thing every day.
3. Cross-check with your Orders tab. Ads Manager attribution and your order list will never match exactly — attribution windows and view-through counts differ. Your Orders tab is the truth; Ads Manager is the map.

The 7-day test plan, with kill and scale rules

Decide the rules before day one, because at 11 pm on day two with zero sales every seller wants to change everything. The plan below assumes one ad set, ₹300 a day, four or five ads.

• Kill rule 1 (day 3): the ad nobody clicks. Compare ads to each other, not to an internet benchmark.
• Kill rule 2 (day 6): the ad that has spent 2× your maximum cost per purchase with nothing to show. Spending ₹762 on a product whose profit is ₹381 and getting zero add-to-carts is enough evidence.
• Scale rule: if the week's cost per purchase is at or under your maximum, raise the daily budget by about 20% every two or three days — ₹300 → ₹360 → ₹430. Doubling it resets learning and usually raises your cost per purchase.
• Iterate rule: if you got clicks and add-to-carts but few purchases, the ad works and the product page or price does not. Fix the page (photos, delivery time, COD, reviews) before you spend more.
• Stop rule: zero add-to-carts across all five ads after ₹1,500 spent means the product, not the ad, is the problem. Swap in your next best-seller.

WhatsApp follow-up for people who clicked but did not buy

Most people who click an ad and reach your store leave without buying — that is true of every store on earth. A small store's advantage is that you can talk to them. Every person who messages you on WhatsApp after seeing the ad, every buyer who starts a WhatsApp checkout and goes quiet, is a warm lead you paid for. Follow them up the same day, personally, once. (On the AI Co-founder plan the campaign planner writes the WhatsApp broadcast and this follow-up message for the campaign it plans; the broadcast can go out from Broadcasts in the dashboard, but this personal follow-up is still you.)

Where the list comes from today on a SitesPlaced store: your WhatsApp chats (people who messaged you from the ad or from the store), WhatsApp-checkout conversations that stopped before confirmation, and — if you use Shiprocket's 1-click checkout on the Growth plan — the abandoned checkouts it captures, which you can follow up from there. That part is a manual habit. The automatic part is on the AI Co-founder plan: a buyer who typed an email or phone number at the cart and left gets a saved-cart email after 1 hour and again after 24 hours, and a WhatsApp message through a Meta-approved marketing template when your WhatsApp Cloud API is connected, with one tap to restore the cart and an opt-out on every message. The honest state of the whole topic is in abandoned cart recovery for small stores.

One rule: one message, then leave them alone. A second nudge a week later with something new to say (back in stock, festival dispatch cut-off) is fine. A third is how you get reported.

Festival timing: test in September, spend in October

From Navratri through Diwali (which falls in early November in 2026) and on to Christmas and the new year, Indian buyers spend more and big advertisers bid more. Both things are true at once: your ads convert better and each impression costs more. The way a small store wins that season is to arrive with a tested ad, not a guess.

• Run the ₹300 test now, four to six weeks before the festival you care about. You want a proven hook and a proven product before CPMs climb.
• Switch the winning hook to the occasion angle two to three weeks out: same product, same creative, caption rewritten around gifting and the dispatch cut-off.
• Say the last order date for guaranteed delivery on the ad and the product page. Nothing kills festival sales like a parcel that arrives after the festival — and nothing sells like a clear deadline.
• Plan stock, not just ads. An ad that works on a product that runs out on day four is worse than no ad.
• After the festival, keep the retargeting campaign running at a low budget. The visitors you paid for in October are the cheapest customers you will have in December.

FAQs