How to run Facebook ads: a beginner’s guide to your first campaign (2026)
Run your first Facebook ads campaign: pick the right objective, set a budget, build an audience, design the creative, add tracking, and read results properly.
Facebook ads look intimidating because Ads Manager throws every advanced control at you on the first screen, but the underlying job is simple: tell Meta what outcome you want, who you want it in front of, and what the ad should say, then give the machine enough budget and time to learn. Get those four right and the dozens of toggles around them mostly take care of themselves.
This is the whole thing in order, built around the three levels every Meta campaign has — the campaign (your objective and overall budget logic), the ad set (who you target, where it shows, and what you optimize for), and the ad (the actual creative people see). It is written for a first-timer with a product or service and a small test budget, and it is honest about the two mistakes that waste most beginner spend: choosing the wrong objective, and killing a campaign before Meta's delivery system has finished learning. If you want the organic-first groundwork before you pay for reach, start with [building a brand presence on Facebook](/how-to/build-a-brand-presence-on-facebook).
The steps
Set up the account, Page, and tracking before you build anything. You advertise from a Facebook Page inside a Business Portfolio (Business Manager), with a payment method attached in Ads Manager — sort those first so nothing stalls at launch. If the ad sends people to a website, install Meta's tracking now: the old Meta Pixel is now one source inside a unified Dataset (your former Pixel ID is the Dataset ID), and pairing it with the Conversions API is what lets Meta measure purchases and optimize toward them. No tracking means no conversion data, which caps what the algorithm can learn.
Pick the one objective that matches your real goal. Meta gives six campaign objectives under its ODAX framework — Awareness, Traffic, Engagement, Leads, App Promotion, and Sales — and the objective tells the delivery system which outcome to optimize for. It is brutally literal: choose Traffic and you get cheap clicks from people who never buy; choose Sales and Meta hunts for buyers. Pick for the action you actually want (a purchase, a lead form, a message), not the vanity metric. You cannot change the objective after the campaign launches, so this is the one decision to get right up front.
Decide daily vs lifetime budget — and keep the test small. Set your budget at either the campaign or the ad-set level; the campaign-level option (Advantage+ campaign budget) lets Meta shift money to whichever ad set performs best. Choose a daily budget for ongoing testing or a lifetime budget when the campaign has a fixed run and end date. Meta's minimum varies by what the ad set optimizes for — roughly $1/day for pure impression-based delivery, but around $5/day once you're optimizing for clicks, leads, or conversions (the case for most beginner campaigns) — and either floor starves the algorithm. Give a real test at least ~$10/day so Meta gathers enough signal. Budget for learning, not for instant sales; the first days are the system finding your audience.
Build your audience at the ad-set level. You have three audience types. Saved audiences are built from location, age, language, interests, and behaviors — good for a cold-traffic test. Custom audiences retarget people who already know you: website visitors (via your Dataset), past engagers, or an uploaded customer list. Lookalike audiences ask Meta to find new people who resemble a source audience, usually your best customers. Start broad-but-relevant rather than stacking narrow filters; Meta's delivery works better with room to optimize than boxed into a tiny segment.
Choose placements and let Advantage+ do the early lifting. Placements are where your ad appears — Facebook and Instagram feeds, Stories, Reels, Marketplace, the Audience Network, and more. For a first campaign, leave Advantage+ placements (formerly Automatic Placements) on: Meta spreads delivery across surfaces and finds the cheapest results, then the placement breakdown in reporting tells you where it actually worked. Manual placement control is a later optimization once you have data, not a day-one decision. Just make sure your creative is built to survive every placement's crop.
Set the performance goal (optimization event). Inside the ad set, the performance goal is the specific event Meta optimizes delivery toward — link clicks, landing-page views, leads, or conversions like purchases. Match it to the objective and to what your tracking can actually see: optimizing for purchases only works if your Dataset is firing purchase events. If you have no conversion signal yet, optimize for a shallower event (landing-page views or leads) until enough conversions accumulate. This setting, more than your targeting, is what shapes who Meta shows the ad to.
Build the ad creative to fit the format and the feed. At the ad level you pick the Facebook Page and Instagram account, the format, and the assets. Formats: single image or video, carousel (multiple swipeable cards, each with its own link), or collection (a cover plus a product grid that opens full-screen on mobile). Each format has its own dimensions, and getting them wrong is the most common reason an ad looks cropped or blurry — build vertical (roughly 4:5 or 9:16) so it fills a phone. Lead with a hook in the first second, keep on-image text minimal, and add captions to any video for sound-off viewing.
Add the destination, tracking, and publish. Set where the click goes — a website URL, an instant lead form, or a Messenger/WhatsApp thread — and make that page continue the promise the ad made, not a generic homepage. Add UTM parameters to the URL so your own analytics can attribute traffic, confirm the Dataset is connected for conversion tracking, then submit. The ad enters review (usually approved within a day) before it starts delivering. Send the click somewhere that matches the objective, or a great ad wastes its clicks on a page that can't convert them.
Let it learn, then read the results and iterate. Do not touch a new campaign for the first few days. Meta runs a learning phase while its delivery system finds your audience, and day-one numbers are noise — give it 3–7 days and enough conversions before judging anything. Then read cost-per-result against your objective (not likes), check the placement and audience breakdowns to see what worked, and iterate: run an A/B test on one variable at a time — image vs image, or audience vs audience — rather than changing five things at once and learning nothing. Scale the winners, cut the losers, keep the tests clean.
Common gotchas
Choosing Traffic when you want sales. Traffic optimizes for cheap clicks, not buyers — you'll get a low cost-per-click and no revenue. Match the objective to the real action; you can't change it after launch.
Editing or killing the campaign in the first 48 hours. Every meaningful change resets Meta's learning phase, and day-one performance is unreliable. Give it 3–7 days and enough conversions before judging or optimizing.
Running ads with no Dataset (Pixel) installed. Without conversion tracking, Meta can't optimize toward purchases and you can't measure what worked — you're flying blind on the metric that matters.
Stacking narrow interest filters. Over-tight targeting starves the algorithm of room to optimize and drives up cost. Start broad-but-relevant and let Meta's delivery find the buyers.
Uploading horizontal creative. A landscape image gets cropped or letterboxed in feeds, Stories, and Reels. Build vertical (4:5 or 9:16) and confirm it survives every placement.
Testing five things at once. If you change the image, audience, budget, and copy together and results move, you've learned nothing. A/B test one variable at a time.
Judging by likes and reach. Engagement metrics feel good but don't pay; read cost-per-result on your actual objective — leads, purchases, or link clicks that convert.
Legal note
Facebook ads must comply with Meta's Advertising Standards, and some categories — credit, employment, housing, social issues, elections, and politics — fall under Special Ad Categories with restricted targeting (you lose age, gender, ZIP, and detailed-interest options) and may require identity or authorization steps. If your ad uses AI-generated imagery or synthetic media, disclose it where the platform or your jurisdiction requires; see [how to disclose AI-generated ads](/how-to/disclose-ai-generated-ads). Any claims in the ad must be truthful and substantiated, and running conversion tracking involves handling user data under Meta's data-use terms and applicable privacy law.
Where Kompozy fits
Notice where a first Facebook ad campaign actually stalls. It is almost never the objective dropdown or the budget slider — those take a minute. It is step 7, and the iteration in step 9: you need a real set of on-brand creatives to launch with, and then a fresh set every week as the winners fatigue and the A/B tests demand new variants. Ad creative is the input the whole channel runs on, and Meta's own guidance is to test several against each intent, not ship one image and hope. Hand-producing that supply — vertical, format-correct, on-brand, in volume — is the wall beginners hit, and it is the exact job Kompozy is built for.
Kompozy is a full AI content generation and multi-platform publishing engine, not a repurposing tool, and for a Facebook ad campaign it turns one product or idea into a testable creative library. Photo Posts and Persona Photos give you scroll-stopping stills; Infographic Photos carry an offer legibly at thumbnail size; Carousel Posts, rendered pixel-exact to your brand by HyperFrames, map straight onto Facebook's multi-card carousel format; and for video ads there are Marketing Shorts (a hook plus demo footage), Persona Shorts, and Clipped Shorts — all generated vertical so step 7's cropping trap never bites. Because it produces several variants of every concept, 'build a set, not one asset' becomes picking winners in a review queue instead of a week in an editor, and a written Persona Brief keeps voice and look identical across every unit you test so the ad and the account describe the same business.
Be exact on the boundary: Kompozy does not touch Ads Manager. It does not set your budget, pick your audience, install your Dataset, place a bid, or read your cost-per-result — you run the campaign; it manufactures the creative the campaign spends its budget on. It also feeds the organic side the paid side rides on, publishing the same content across the eight social platforms plus blog and email so the Page your ads point back to isn't a ghost town. Creator ($49/mo for 2,500 credits) fits a solo advertiser building a first creative set and its weekly refreshes; Pro ($299/mo for 18,000 credits) sustains a marketer or agency feeding multiple campaigns and constant A/B variants across every platform; Enterprise is custom.
Frequently asked questions
How much do I need to spend to run Facebook ads?
Meta's technical minimum depends on what you're optimizing for — around $1 per day for pure-reach delivery, closer to $5 per day once you're optimizing for clicks, leads, or conversions — and either one is too little to learn from. For a meaningful test, budget at least ~$10 per day so Meta's delivery system gathers enough signal to optimize. Actual cost-per-click varies widely by industry and audience, so treat early spend as buying data, not sales, and scale only once a campaign is clearly working.
Which Facebook ad objective should a beginner choose?
Pick the objective that matches the action you actually want. For website purchases, choose Sales; for form-fill leads, choose Leads; for messages, choose Engagement (messaging); for pure clicks to a page, choose Traffic. Avoid defaulting to Traffic or Awareness because they feel safe — they optimize for clicks and reach, not customers. The objective can't be changed after launch, so decide by your real goal.
Do I need a Meta Pixel to run Facebook ads?
Not to run an ad, but you need it to optimize for and measure conversions. The Pixel is now one source inside a unified Dataset (your former Pixel ID is the Dataset ID), ideally paired with the Conversions API. Without it, Meta can't optimize toward purchases or leads on your site and you can't attribute results. Install it before running any conversion-objective campaign.
How long before I can tell if a Facebook ad is working?
Give it 3–7 days and enough conversions before judging. New campaigns run through a learning phase while Meta's delivery system finds your audience, and day-one numbers are noise. Editing the campaign during this window resets learning, so resist the urge to tinker. Read cost-per-result against your objective once the campaign has stabilized, not on the first afternoon.
What is Advantage+ and should I turn it on?
Advantage+ is Meta's suite of AI automations — Advantage+ audience (AI-expanded targeting), Advantage+ placements (auto-distributed delivery), and Advantage+ campaign budget (money shifted to the best-performing ad set). For a first campaign, leaving placements and audience automation on is sensible: it lets Meta find cheap results, and the reporting breakdown then tells you where it worked so you can take manual control later.