How Meta decides who sees your ad: the auction, total value, estimated action rates and why creative now does the targeting

Most Indian business owners meet Meta ads through the "Boost post" button: ₹500 here, ₹1,000 there, and no clear idea why one post reached 40,000 people and another reached 3,000. Before you spend seriously on Facebook and Instagram, you need a working picture of how Meta chooses which ad a person sees and what you pay for it. Once you understand that, you stop fighting the system and start giving it the signals that bring you buyers instead of casual likers.

What you need to know

Every impression is a small auction. When a shopper in Pune opens Instagram, Meta has a moment to decide which of thousands of eligible ads to show in her feed, Stories or Reels. It runs an auction for that single slot. The winner is not simply the advertiser who offered the most money.

The total value formula. Meta publicly describes the winner as the ad with the highest "total value", made of three parts:

• Advertiser bid — what you are willing to pay for the result you chose. On the default "Highest volume" bid strategy, Meta sets this for you within your budget.
• Estimated action rate — Meta's prediction of how likely this particular person is to take the action you optimised for (a purchase, a lead, a message) after seeing this particular ad.
• Ad quality — signals from people who saw the ad (hiding it, reporting it, engaging genuinely) and checks for low-quality features such as engagement bait ("Comment YES to win"), withheld information or sensational claims.

In simple terms: total value ≈ bid × estimated action rate + ad quality.

Why a cheaper bid can win. Here is a simplified illustration (the real system is more complex, so treat the numbers as teaching aids only). Two ads compete for the same person:

• Ad A: bid ₹300 per purchase × estimated action rate 0.5% = 1.5
• Ad B: bid ₹200 per purchase × estimated action rate 1.0% = 2.0

Ad B wins even though its bid is lower, because Meta expects the person to act on it twice as often. The lesson for a small business is powerful: a relevant ad for the right person can beat a big brand with a bigger budget.

You usually pay less than your maximum. The auction is designed so you pay roughly what is needed to win, not your full bid. Most campaigns are charged per impression (per 1,000 impressions, called CPM), even when you optimise for leads or purchases. Your cost per result is therefore a chain: what you pay for impressions, how many viewers click, and how many clickers act. Lesson 02 turns this chain into rupees.

You get what you optimise for. The optimisation event you pick tells Meta whom to look for. Optimise for link clicks and the system finds people who click often, including those who tap by mistake and bounce. Optimise for landing page views and it finds people who wait for pages to load. Optimise for purchases or qualified leads and it looks for people whose behaviour resembles buyers. Many owners who say "Meta gives junk traffic" had asked for exactly that.

Creative now does most of the targeting. Meta increasingly recommends broad audiences and lets its models decide who sees each ad. In that world, your creative is your strongest targeting tool. A Tamil-language Reel showing a mother packing school shoes into a bag self-selects parents in Chennai far better than any interest you could tick. The image, the first line, the language, the price shown and the people on screen all tell the system (and the viewer) who the ad is for.

Ad relevance diagnostics. Once an ad has had enough impressions, Ads Manager shows three rankings compared with ads competing for the same audience: quality ranking, engagement rate ranking and conversion rate ranking. "Below average" on quality often means the ad feels like clickbait or looks low-effort; below average on conversion rate usually points to the offer, landing page or form rather than the ad itself.

Budget pacing. Meta spreads your budget across the day and week to find the cheapest opportunities. Its help pages describe that a daily budget may be exceeded on high-opportunity days while staying within a weekly limit; confirm the current rule in the Meta Business Help Centre so an uneven daily spend does not alarm you.

Prices move with competition. CPMs rise when more advertisers chase the same people — typically in festive weeks such as Navratri to Diwali, around big sale events and at month-end or quarter-end. A common rule of thumb is to plan for costlier impressions in these periods and to launch tests in quieter weeks.

What you control and what Meta controls.

• You control: the offer, the creative, the optimisation event, the destination (website, form, WhatsApp), budget, the geography you serve, exclusions and the quality of follow-up.
• Meta controls: exactly who sees each ad, when, in which placement, and the price of each impression.

Step-by-step method

1. Open Ads Manager and list every active or recent campaign with its objective and its optimisation event (visible at ad set level under "Performance goal").
2. For each, write the business result you really wanted (orders, qualified enquiries, store visits) and mark any mismatch between that result and the optimisation event.
3. Add the three ad relevance diagnostics columns and note any ad marked below average.
4. Read each creative as a stranger would: in the first second, is it obvious who it is for and what is on offer?
5. Remove engagement-bait lines, misleading urgency or clickbait thumbnails that could lower ad quality.
6. Decide the single event that best predicts revenue for your business and that you can track reliably (lesson 04 covers tracking).
7. List two or three creative ideas that would self-select your ideal customer by language, city, problem or price point.
8. Write down which levers you will change this month and which you will leave to Meta.

Common mistakes

• Choosing "Traffic" or "Engagement" because the cost per result looks low, then wondering why no one buys.
• Judging an ad only by CPM; a cheap impression shown to the wrong person is expensive.
• Narrowing audiences to a few hundred thousand people with stacked interests, starving the system of room to find buyers.
• Using "Comment YES" or "Tag a friend" hooks that attract engagement but can hurt ad quality and bring non-buyers.
• Running one creative for everybody and expecting the system to find several very different customer types.
• Reading one day's results as a verdict; the auction fluctuates hour to hour.

Ask the AI Business Tutor

• "I sell [product or service] to [type of customer] in [city or region]. My last Meta campaigns used the [objective] objective, optimised for [event], spent ₹[amount] and produced [number] [results], which led to about [number] sales. Explain in plain English why the delivery system may have shown my ads to the wrong people, which optimisation event I should choose next and three creative ideas that would help the system find my real buyers."