- Ads
- Meta ads
- Small business
How much should a small business spend on Meta ads when starting out?
Set your first Facebook and Instagram ads budget from the cost of one result and the learning phase, with how the auction works and what to change or leave alone.
On this page
- Why "₹500 a day" is the wrong first question
- How Meta's auction actually works
- The learning phase, in plain words
- Work out your budget in three steps
- Where "cost per result" comes from
- When 50 results a week is out of reach
- Creative is now most of your targeting
- What to change, and what to leave alone
- How to scale what works
- A simple weekly routine
- Meta ads in PostAmico
- Sources
Start with the cost of one result, not with a round number. Work out roughly what one lead, booking or sale is likely to cost you, multiply it by about 50, and spread that over a week. That is the budget one ad set needs to give Meta enough to learn from.
That rule is the answer. The rest of this article explains why it works, so you can adapt it when your situation doesn't fit the rule, which it often won't.
Why "₹500 a day" is the wrong first question
Most people pick a budget that feels comfortable and then hope it works. The trouble is that Meta's delivery system needs data to work out who is most likely to respond. If your budget is too small to produce results, the system never learns, your costs stay high, and it's easy to conclude that "ads don't work" when the budget simply couldn't do the job.
To see why, you need to know a little about how Meta decides who sees your ad.
How Meta's auction actually works
Every time someone opens Instagram or Facebook, Meta runs an auction for the ad slots in front of them. The winner is not simply the highest bidder. Meta says the winning ad is the one with the highest total value, which combines three things:
- Your bid: what you're willing to pay, usually set automatically from your budget.
- Estimated action rate: how likely this person is to take the action you chose, such as messaging you or buying.
- Ad quality: signals about whether people find the ad useful or annoying, such as hiding it or reporting it.
Roughly: total value ≈ bid × estimated action rate + ad quality.
This has a practical consequence for small budgets. If Meta thinks your ad is twice as likely to get a response from someone, it can beat an ad bidding nearly twice as much. Relevance is the small advertiser's advantage. But Meta can only estimate your action rate from data, and that's where the learning phase comes in.
The learning phase, in plain words
When you launch a new ad set, or make a big change to one, Meta has little idea who will respond. It spends time showing the ad to different people and watching what happens. This is the learning phase. Results are usually less stable and more expensive while it lasts.
Meta's guidance has long been that an ad set needs about 50 optimisation events in 7 days to leave the learning phase. An optimisation event is whatever you told Meta to aim for: a lead form, a purchase, a message, a link click.
Why about 50? Think of it as a sample size. With five purchases, Meta can't tell whether the buyers have anything in common or got lucky. With fifty, patterns appear: these ages, these interests, these placements, this time of day. If an ad set can't reach that number, Ads Manager shows it as "Learning limited": delivery keeps going, but it isn't settling.
Work out your budget in three steps
Pick one goal
Choose the single result that matters most right now: messages, leads, bookings or purchases. One campaign, one goal.
Estimate the cost of one result
Use your own past results if you have them. If you don't, work it out from the funnel below, take a cautious guess, and plan to correct it after the first week.
Multiply by 50, divide by 7
Weekly budget = cost per result × 50. Daily budget = that number ÷ 7.
An example. A salon expects a booking enquiry to cost about ₹150.
- ₹150 × 50 = ₹7,500 a week
- ₹7,500 ÷ 7 ≈ ₹1,070 a day for one ad set
Budget calculator
Meta's long-standing guidance is about 50 results in 7 days per ad set to leave the learning phase.
Per week
₹7,500
Per day
₹1,071
₹150 × 50 results ÷ 7 days
Where "cost per result" comes from
If you've never run ads, the hardest part is guessing the cost of one result. You can build an estimate from three numbers instead of guessing it directly:
| Number | What it means | Example for an online store |
|---|---|---|
| CPM | Cost to show the ad 1,000 times | ₹120 |
| Click-through rate | Share of viewers who click | 1.2% |
| Conversion rate | Share of visitors who buy | 2% |
The arithmetic: ₹120 buys 1,000 views. At 1.2%, that's 12 clicks, so each click costs ₹10. At 2%, you need 50 clicks per sale, so a sale costs about ₹500.
These example figures are assumptions, not benchmarks; CPMs in India vary widely by audience, season and placement. The value is in seeing which lever moves the answer most. Double your website's conversion rate from 2% to 4% and the cost per sale halves to ₹250, which halves the budget you need to exit learning. Often the cheapest way to "fix" an ad budget is to fix the page the ad sends people to.
When 50 results a week is out of reach
At ₹500 a sale, the rule says ₹25,000 a week. Many small businesses can't do that, and that's fine. Don't spread a smaller budget across lots of ad sets. Do one of these instead:
- Aim for an earlier action. Optimise for messages, leads or add-to-carts instead of purchases. You'll reach the threshold sooner. The trade-off is real: Meta then finds people who message, not necessarily people who buy, so watch the quality of what comes in.
- Run fewer ad sets. One well-funded ad set learns; five starved ones don't.
- Accept slower learning. A smaller budget still works; it just takes longer and results will be less steady at first. Judge it over a month, not a week.
Creative is now most of your targeting
A few years ago, the craft of Meta ads was choosing detailed audiences. That has shifted. Since 2024, Meta has used a system it calls Andromeda to pick which ads are even considered for each person, and Meta has encouraged advertisers to give it a wider variety of creative to choose from.
In practice, your images and videos do much of the work interests and audiences used to do. A reel showing a bride's trial makeup finds different people from a before-and-after photo, even with identical targeting. For a small business, that means two things:
- Variety beats volume. Three genuinely different ideas (a demo, a customer story, a price-led offer) teach Meta more than ten small tweaks of one image.
- Broad audiences are less risky than they sound. Give Meta your location and the basics, and let the creative and the learning phase find the buyers.
What to change, and what to leave alone
During the first week, big edits restart learning. Changes Meta treats as significant include:
- changing the audience or locations,
- swapping the image, video or main text,
- changing the optimisation goal,
- large budget jumps at once,
- pausing an ad set for a week or more.
Small fixes, such as a typo, are fine. Otherwise, plan your changes for after the first week, and make them one at a time so you know what worked.
How to scale what works
When an ad set is out of learning and the cost per result is where you want it:
- Raise the budget gradually, by about a fifth at a time, every few days.
- Add new creative in a new ad set rather than replacing the winner.
- Keep an eye on frequency. If the same people see the ad again and again, costs rise. Fresh images or a new angle usually fix it.
Expect costs to rise a little as you scale. The first ₹1,000 a day reaches the people most likely to respond; the next ₹1,000 has to reach people slightly less likely to. Scaling is about finding the point where the extra spend still pays for itself, not about keeping the first week's cost per result forever.
A simple weekly routine
- Monday: check last week's spend, results and cost per result for each ad set.
- Decide one change per ad set: keep, raise the budget a little, refresh creative, or pause.
- Midweek: look for anything broken: a rejected ad, a link that doesn't work, a form with no responses.
- Friday: note what you learned, so next month's plan starts from facts.
Meta ads in PostAmico
PostAmico's Ads area shows your Meta and Google campaigns in one place, with a plain-language report on spend, results and what to change. New Meta campaigns that Veda creates start paused, so nothing spends until you are ready. Veda asks for your approval before it changes an ad budget, pauses or resumes ad spend, or deletes anything.
PostAmico is opening to its waitlist first. Join the waitlist to be among the first to try it.
Sources
- Meta Business Help Centre, About the learning phase
- Meta Business Help Centre, About ad auctions
- Meta Engineering, Meta Andromeda: supercharging Advantage+ automation, December 2024
- Jon Loomer, Meta Andromeda: what it means for your ad strategy
Questions
What is the smallest budget that works on Meta ads?
There is no single number. What matters is whether your budget can buy enough results for Meta to learn. A useful starting point is your expected cost per result multiplied by about 50, spread over a week, for one ad set.
Should I boost posts or use Ads Manager?
Boosting is quick and fine for getting a good post in front of more people. For leads or sales, a campaign built in Ads Manager gives you better control over the goal, the audience and what Meta optimises for.
How long should I wait before judging an ad?
Give a new ad set about a week, or until it has had the chance to collect around 50 results, before making big changes. Judging after a day or two usually means judging the learning phase, not the ad.
Why does the same budget get different results for different businesses?
Because Meta's auction doesn't only look at how much you bid. It also weighs how likely people are to take your chosen action and how good your ad is, so a relevant, well-made ad can win placements that a higher bidder loses.
Should I spend more on the ad that is working?
Yes, but in steps. Raising a budget by a large amount at once can send the ad set back into learning. Increase it gradually, by about a fifth at a time, and watch the cost per result.
Keep reading
All articles →- AI marketingIndiaSmall businessAI marketing for small businesses in India: what to automate first, and what to keepWhere AI saves real time in marketing for Indian founders and small teams, why it fails where it does, what to keep human, and how the DPDP Rules affect your plans. 8 min read
- SEOAI searchGEOThe death of search is a lie: why Google, Bing and Brave power the AI buying eraAI assistants still search the web before they answer. Why Google, Bing, Brave and DuckDuckGo decide which businesses AI recommends, and what to do about it. 8 min read
- BrandAI marketingContent planningHow to write a brand voice guide your team and your AI can actually followA one-page way to define your brand voice using four research-backed tone scales, with examples, words to use and avoid, and a template for people and AI tools. 7 min read
Put this guide to work.
Veda can do this for your business every week, in your voice. PostAmico is opening to its waitlist first.