Instagram advertising for business: how it works and how to launch it
In this article we break down how targeted advertising on Instagram really works: which ad account it runs through, on what principle the algorithm decides who to show your ads to, how to build audiences, which formats exist and how they differ, what it costs at the start — and, step by step, how to launch your first campaign properly rather than at random. With a real example from our practice — the Be Yoga case, where a two-week campaign on Instagram delivered a ROAS of 363%.
What targeted advertising on Instagram is and how it works technically
Instagram has no ad account of its own. All advertising — from Stories to Reels — is managed through Meta Ads Manager, the same tool that serves Facebook. Within that system Instagram is one of the placements, alongside the Facebook feed, Messenger and partner apps in the Audience Network.
Technically the launch looks like this: a Meta business account → a connected Instagram profile → Ads Manager, where the campaign, audience, budget and creative are configured. The «Promote» button inside the Instagram app itself is a simplified shell over the same account, but with a limited set of controls: a restricted choice of campaign objective, almost no control over the audience, no access to the pixel or conversion events, and no proper statistics in money terms.
Why this matters in practice. If your goal is not simply «more likes» but specific enquiries or sales, optimisation has to be pointed at that action. That is only possible through a full Ads Manager setup with the pixel installed (for a website) or with events configured inside the platform itself — Lead Ads, Direct messages, clicks on a phone number. Without this the algorithm optimises for what it can measure on its own: impressions and clicks, not money in your till. This is where the difference between «advertising that runs» and «advertising that sells» is established.
Business Manager vs a personal account. Another technical detail often skipped at the start: advertising should be run through Business Manager rather than through a personal ad account tied directly to a personal Facebook profile. A business account provides access for several people on the team, separation of roles (one person sees only statistics, another can edit campaigns), and — most importantly — more resilience against technical blocks than a personal account.
How the algorithm decides who to show your advertising to
At the heart of ad delivery is an auction. Every time Instagram opens a feed or a Story for someone, dozens of advertisers «compete» for that specific impression. The winner is not whoever pays most, but whoever has the highest product of three factors: bid, expected ad quality (will this person find this particular ad interesting) and expected action (will they do what is wanted after seeing it).
This means something simple but counter-intuitive: cheaper advertising with higher engagement often beats more expensive but uninteresting advertising in the auction. The algorithm has an interest in showing what people interact with — so a good creative literally lowers your cost per impression and cost per click. It is one reason two businesses with the same budget can end up with radically different costs per lead — the difference is not in the money but in how «interesting» the ad is to the audience it reaches.
The second key mechanism is the learning phase. When a campaign launches (or changes substantially — a new budget, a new audience, a new objective), the algorithm does not yet know who converts. It tests different audience segments, gathers data on clicks and purchases, and gradually narrows down to «people similar to those who have already bought». This is not instant: in Meta Ads, learning is considered complete after approximately 50 conversions on the target event per week. Until that point the results can fluctuate more than seems logical — and that is a normal part of how the algorithm works, not a reason to panic, switch everything off or change settings daily.
The consequence for a business owner: every abrupt change (a new creative replacing all the old ones at once, the budget doubling overnight, the audience completely rewritten) throws the campaign back into learning. This is not a fault in the system — it is how it was designed. Understanding this mechanism saves more money than any clever targeting setting.
Audiences: interests, broad targeting, look-alike
A media buyer’s skill used to be measured by their ability to build narrow interest-based audiences. Today the rules have changed: Meta has substantially simplified targeting and shifted the main work onto the algorithm, reducing the role of manual audience settings.
Interest-based audiences. The classic approach: choose categories such as «yoga», «healthy lifestyle», «online shopping» and show advertising to people whose behaviour on Meta matches. It works as a starting point when you have no purchase data yet, but it is often either too broad or too narrow depending on the niche.
Broad audience with automatic optimisation (Advantage+). Instead of manually selecting interests, you give the algorithm minimal constraints (age, gender, geography) and let it find buyers across Meta’s entire user base itself. In practice this often performs no worse — and sometimes better — than manual targeting, because the algorithm sees signals unavailable to a human: behaviour across hundreds of sites, purchase history, similarity to millions of other profiles.
Look-alike (LAL) — audiences similar to your buyers. The most precise tool once you have a base of customers or leads: the algorithm finds people with similar behaviour across the whole platform audience. LAL 1% is a narrow, highly accurate copy of your base; LAL 2–5% is broader and less precise but larger in volume. The quality of a LAL depends directly on the size of the base: 50 buyers produce a weak model, 500+ a far more accurate one. This is exactly why LAL simply does not work at the start of a project, when there is no buyer base yet — and that is normal; testing begins with interests or broad targeting.
Retargeting. Showing advertising to people who have already interacted with the profile, the site or previous ads but have not bought. Often the cheapest channel by cost per lead — but not always justified if the volume of that audience is small or the niche does not involve a period of «ripening». It is worth testing, but without blind faith that retargeting always rescues the result.
Ad formats on Instagram: where and what to show
Instagram delivers advertising across four main placements, each with its own logic of audience behaviour.
| Placement | Format | Characteristic |
|---|---|---|
| Feed | photo, carousel, video | the most «familiar» advertising format, suits every stage of the funnel |
| Stories | vertical full-screen photo/video | short interaction, strong for recognisable brands and simple offers |
| Reels | short vertical video | the largest organic reach among cold audiences, requires a native style |
| Explore / Profile | adapts to the ad format | additional reach with no separate configuration |
The practical conclusion: do not choose a single placement manually at the start. Automatic placements let the algorithm distribute budget itself towards wherever the ad gets the better response — and in practice that is almost always cheaper than manually restricting delivery to the feed. Restricting placements manually is only worth doing once you have statistics clearly showing that a specific placement is not producing conversions despite sufficient impressions.
Creative format also affects the result more than it seems at first glance. Reels currently receive the greatest organic priority in Instagram’s algorithm, but that only works for content shot in a native, «non-advertising» style — a long, obviously staged clip in Reels format loses to simple static. We examined these nuances in detail in «Creatives that sell: video, static and UGC in Meta Ads» — in short: video wins where the product or service has to be «shown in action», static wins when the offer and price speak for themselves, and UGC (content shot «like an ordinary person» would) often beats both with a cold audience, because it does not look like advertising.
Case: Be Yoga — from zero to 363% ROAS in two weeks
The best way to see the mechanics in action is a real example. Be Yoga sells yoga mats online through Instagram. The client’s brief was simple and honest: no rigidly defined KPIs, the main thing being a return on the money invested in advertising.
Before launching anything, we analysed what competitors already selling successfully in this niche were doing. The picture was clear: the winners bet on video creatives rather than static — logically so, since a yoga mat is a tactile product and the buyer needs to see the texture of the material and how it looks during practice. We went straight into shooting selling videos: the product in action, with quality and benefits visible.
A separate complication on this project: roughly half of the working page’s audience consisted of bots, which could distort the signals used for algorithmic optimisation. So interests for the campaign launch were selected with particular care, rather than relying solely on broad targeting or unconstrained Advantage+.
The budget was split across two jobs:
- Video creatives to a broad audience and to interest-based audiences — the main source of sales;
- A traffic campaign on a minimal budget ($5/day) to grow the number of relevant followers, raising trust in the account and building a base for future retargeting and look-alikes.
The result after two weeks: an ad budget of $262, revenue of $952, a ROAS of 363% and 69 leads. The first week alone produced 9 sales — full payback plus profit — followed by a further 13 sales at a cost per lead of $3.53. The traffic campaign added +57 new relevant followers at $0.35 each, and the trajectory allowed for further scaling.
An important nuance that often gets ignored: the Be Yoga result was not «targeting magic» but the consequence of competitor analysis before launch, careful audience selection given a bot-contaminated base, and the right choice of creative format for that specific product. Remove any one of those components and the result would have been different.
A step-by-step first campaign launch on Instagram
Here is the sequence we use to launch advertising for new clients — no superfluous steps, but none missed either.
Step 1. Prepare the technical base. Business Manager, a Meta business account, a connected Instagram profile, the pixel on the site (if you are driving traffic to a site) or configured Lead Ads / Direct if you sell through conversation. This is the foundation — without it optimisation for sales is physically impossible, because the algorithm has nothing to work from.
Step 2. Choose the campaign objective to match your funnel. For sales on a site, conversion (Purchase). For enquiries, lead generation (Lead) or messages. A «traffic» or «reach» objective is good only for brand awareness, not for direct sales — the algorithm will optimise for exactly what you specified, not for what you had in mind.
Step 3. Prepare 3–5 creative variants. At minimum one video and one static, with different messages and different hooks in the opening seconds. One creative is not a test, it is a guess: you cannot know in advance which message and visual will work on your audience. More on the approach to testing creatives in the article on formats in Meta Ads.
Step 4. Set the starting audience and budget. One or two audience hypotheses (interest-based or broad with automatic placements), with a starting budget of $5–15 a day per ad set. Do not immediately spread the budget across five or six audiences — the algorithm needs a volume of data for each one to find a pattern, and a scattered budget simply stretches that volume out over time.
Step 5. Let the campaign go through the learning phase without daily interference. The first 5–7 days are a data-gathering period. It is worth watching, but not «panic-optimising» by changing budgets or audiences every day — that resets the algorithm’s learning to zero and effectively restarts the test at your expense.
Step 6. Judge by money, not by likes. CTR and CPM are intermediate signals, useful for diagnosis but not goals in themselves. The main metrics are cost per enquiry or purchase, and ROAS. Cheap clicks with no sales behind them are not a campaign success but a signal that the problem lies either in the landing page or in the wrong campaign objective.
Step 7. Scale gradually, once the metric has been stable for 7–14 days. A sharp budget increase resets the learning phase just as abrupt changes at the start do. A safe pace is a rise of 20–30% every 2–3 days, no more. There is a separate article on this: «How to scale advertising without losing profitability».
What Instagram advertising costs at the start
The budget question is one of the most common, and the honest answer is this: the minimum budget for testing is $5–10 a day per ad set, which is roughly $150–300 a month on impressions alone. This is not «the price of a result», it is the cost of gathering the data the algorithm needs to start finding the right people and exit the learning phase.
On top of the media budget you should add three further costs that are often forgotten at the planning stage:
- The cost of creatives — from simple phone footage (almost free, but demanding time and a thought-through script) to a production with a camera operator, editing and voiceover;
- Time for setup and monitoring — the pixel, conversion events, campaign structure, weekly metric analysis and timely responses to deviations;
- The «cost of learning», if you are launching yourself for the first time — the first weeks almost always deliver worse results than were potentially available, while you work out the logic of the account, the difference between campaign objectives and the reasons metrics fluctuate.
The real budget needed for a stable result depends heavily on the niche, average order value and auction competition: physical products with an order value of a few hundred hryvnias and a B2B service with an order value in the thousands of dollars operate on different economics, even though the algorithm’s mechanics are the same. We examined the economics of a DIY launch versus a systematic approach in detail in «7 mistakes in a DIY ad launch that drain your budget» — which also covers why a «cheap test» most often ends up costing more than it appears to at the start.
VELAR’s position: Instagram advertising works by the laws of the algorithm, not intuition
Instagram creates an illusion of simplicity: a nice profile, a good photo, a «Promote» button — and there should be a result. In reality, behind that interface sits a complex system of auctions and machine learning that responds to signals, not to good intentions or the owner’s good taste.
Our position is straightforward: advertising on Instagram is not about «posting nice content and raising the budget». It is about technical preparation (pixel, conversion events, Business Manager), the right campaign objective for your business result, a deliberate choice of audience for the stage your project is at, testing several creatives rather than one, and patience while the algorithm learns. Those elements, rather than the size of the budget, determine the difference between a case like Be Yoga and advertising that simply «runs» without producing sales.
Our average ROAS across clients is 598%, and none of those results began with a random press of a button. Every one began with preparation, competitor analysis and systematic testing — not with intuition.
If you want to understand how targeted advertising on Instagram would work specifically for your business — without guesswork and without wasting budget on learning — request a free audit of your ad campaigns. We will examine your niche and your audience and show you a concrete launch plan.
Read also:
- «Creatives that sell: video, static and UGC in Meta Ads»
- «7 mistakes in a DIY ad launch that drain your budget»
- «How to scale advertising without losing profitability»
Frequently asked questions
How much does Instagram advertising cost for a small business?
The minimum ad budget for a test launch is roughly $5–10 a day per ad set, which is $150–300 a month on impressions alone. On top of that you should add the cost of creatives and the time to set everything up. For the first 2–4 weeks the budget largely goes towards training the algorithm rather than producing a flood of sales — that is a normal part of the process, not a mistake.
Do I need a separate Facebook account to advertise on Instagram?
Yes. Instagram advertising is technically launched through Meta Ads Manager with a linked Facebook account and Business page. The Instagram profile is connected to the Meta business account, and ads are then delivered into the feed, Stories and Reels. Without that connection, the «Promote» button inside the Instagram app offers far fewer settings than the full ad account.
How does targeted advertising on Instagram differ from boosting a post (the «Promote» button)?
The «Promote» button inside the app is a simplified version with minimal settings: almost no choice of campaign objective, audiences or formats. Ads Manager gives full control: optimisation for a purchase or a lead (rather than likes), the pixel and conversion events, several audiences at once, and detailed statistics on ROAS and cost per lead. For any business where advertising has to produce sales rather than reach, it is Ads Manager, not the button.
How long does it take to see the first result?
The first signals appear within 3–7 days, once enough data has accumulated to assess CTR and cost per click. But the campaign should be judged on sales only after it exits the learning phase — roughly 50 conversions on the target event per week, which at an average budget takes 1–3 weeks. In the Be Yoga case the first week already delivered full payback, but that is more an exception in a high-demand niche than a guarantee for any business.
Can I run Instagram advertising without a website?
Yes. Instead of a site you can drive traffic to a lead form inside Meta (Lead Ads), to Direct messages, or to the Instagram profile itself if sales happen through conversation. This is especially common in physical products and services at the start, before a full site exists. The key thing is that the landing point (profile, chatbot, form) is ready to process enquiries as fast as the traffic arrives.
