HOW TO RUN META (FACEBOOK & INSTAGRAM) ADS: 2026 GUİDE

Running Meta ads, meaning advertising on Facebook and Instagram, is a powerful performance channel that can produce measurable results even on small budgets when it is set up correctly. But the interface changes often, and most of the cost figures circulating online are inflated or undated. This guide covers the whole process, from Meta Ads Manager setup to campaign structure, Advantage+ automation, budgets, bidding, creative and measurement, based on verified sources and with costs given as sourced ranges. Meta advertising forms the performance layer of Google Ads and the broader digital marketing mix.

What Is Meta Ads Manager and How Do You Set It Up?

Meta Ads Manager is the single central panel where you create, manage and report on ads across all Meta surfaces, including Facebook, Instagram, Messenger, Threads and the Meta Audience Network. Before you start you need a Meta Business account, a business page, a defined payment method and a verified Pixel or Conversions API setup. The difference between a Boosted Post and Ads Manager matters: boosting is quick and simple but gives limited control over objective, placements and optimization, so serious and scalable performance should run through Ads Manager. Boosting is better suited to simple awareness or engagement, while performance marketing runs through Ads Manager.

Campaign Structure: Campaign, Ad Set and Ad (3 Levels)

Ads Manager still works at three levels in 2026: the campaign (objective and often the CBO budget), the ad set (audience, placements, schedule and optimization event) and the ad (creative, copy, headline and call to action). A key rule is that the objective is locked at campaign creation: to change it you cannot edit in place, you must duplicate or rebuild the campaign. ODAX (Outcome-Driven Ad Experiences) consolidated the old eleven objectives into six: Awareness, Traffic, Engagement, Leads, App Promotion and Sales. A common mistake is audience fragmentation: five or six or more ad sets for a single segment split the conversion signal and pit your ad sets against each other in the same auction, so a simple structure plus Advantage+ is preferable.

Placements and Ad Formats

Placements span four main surfaces: Facebook (Feed, Reels, Stories, in-stream video, Marketplace, right column), Instagram (Feed, Stories, Reels, Explore), Messenger and the Meta Audience Network (non-Meta third-party apps). Audience Network format types are native, banner, interstitial and rewarded video. The core ad formats are Single Image, Single Video, Carousel (two to ten swipeable cards), Collection and Reels or Stories vertical video. A Collection ad requires at least four products or images (the first becomes the cover), runs only on mobile and needs a connected catalog; the upper limit of products shown after a tap is about fifty (an estimate across sources, to be confirmed against Meta documentation). In 2026 only the interface changed, with the format now selected under Format Display Options within the ad creative, while the four-product minimum stayed the same. Threads and WhatsApp are expanding as ad placements but are not yet everywhere: Meta began rolling out Threads ads globally in January 2026 (gradually, with full rollout expected to take months), and on WhatsApp ads appear only in the Updates tab and were delayed in the European Union due to GDPR review.

Advantage+ Automation Layers: When to Trust the Automation

Advantage+ is not a single product but a suite of automation layers: Advantage+ Audience (targeting), Advantage+ Placements (placement), Advantage+ campaign budget (budget) and Advantage+ Sales and App campaigns. Advantage+ Placements, formerly automatic placements, distributes ads automatically across all Meta surfaces and is on by default; by Meta's own data, ad sets using it reported about 11.7% better cost per result (CPA) on average. That figure is a Meta internal-test average; because the methodology is not shared, it varies by account, industry and period, and it is not a guarantee. In Advantage+ Audience the interests you enter are treated not as hard filters but as suggestions to the algorithm, which can go beyond them. Around June 23, 2025, Meta merged many detailed-targeting categories and removed detailed-targeting exclusions for many campaigns; the alternative is to manage exclusions through custom audiences and CAPI signals. Advantage+ is not universally superior: manual targeting can win for accounts without enough signal (roughly under fifty conversions per week), for very niche B2B, and for accounts with strong lookalike strategies, so treat it as a strong default to test, not a guarantee.

Audiences: Custom Audiences, Lookalikes and Broad Targeting

Custom Audiences and Lookalike Audiences are still available in 2026. The retention windows are: website custom audiences up to 180 days (Purchase events can extend to about 730 days), Facebook Page and Instagram engagement up to 365 days, and lead-form engagement up to 90 days; uploaded customer-list audiences persist until you manually remove them. In prospecting (cold-audience) campaigns, Advantage+ can over-concentrate on easy-to-convert existing customers, so excluding recent purchasers and current customers protects new-customer acquisition. The 2026 trend is to move from manual interest targeting to broad targeting and to feed the algorithm strong first-party signal; the mechanism behind it is Meta's new AI delivery system, Andromeda. In practice, by agency sources, Advantage+ Audience needs about fifty conversions per week across the account and roughly seven days of consistent data to run stably; for small, local or very narrowly defined demographic accounts, manual targeting can perform better.

Budget Management: CBO vs ABO, Learning Phase and the 20% Rule

ABO (ad set budget optimization) gives each ad set a fixed budget and full control, so it is recommended for the testing phase. CBO (campaign budget optimization, now called Advantage+ campaign budget) keeps the budget at the campaign level and lets the algorithm distribute it, and is recommended when scaling proven winners; the 2026 consensus is that neither is universally better and the choice depends on the campaign stage. The learning phase needs about fifty optimization events per ad set within roughly seven days for stable delivery; a "Learning Limited" status does not resolve on its own, so its cause (insufficient budget, too-narrow audience, sparse events or a fragmented structure) must be diagnosed and fixed. The 20% rule is this: while a campaign is in the learning phase, raising the budget by more than 20% at once can reset learning, whereas once data is stable an increase of about 20% every forty-eight to seventy-two hours is safer. Other changes that reset learning are targeting, bid-strategy or optimization-event changes; a temporary cost fluctuation is expected after a reset, although the exact penalty percentage rests on single-source estimates and is not a universal fact.

Bidding Strategies: Lowest Cost, Cost Cap, Bid Cap and ROAS Goal

Meta's main bid strategies are Lowest Cost (Highest Volume, the default), Cost Cap, Bid Cap, and Minimum ROAS or ROAS Goal. The typical progression is to test with Lowest Cost, scale with Cost Cap, and optimize with advanced strategies. Lowest Cost is ideal for new, low-data campaigns but has no cost control, so day-to-day fluctuation is typical. Set Cost Cap roughly 10 to 20% above your target cost; Bid Cap gives absolute control but severely limits delivery if set too low. Wait for at least fifty conversions before evaluating any bid strategy, because intervening early disrupts learning. Break-even ROAS follows a simple formula: one divided by your profit margin. For example, if the profit margin is 25%, break-even ROAS is four times, and your ROAS target should be based on your real business margin rather than an arbitrary number.

Advantage+ Sales (Formerly ASC): Deprecated APIs and Ecommerce Setup

Meta merged the Advantage+ Shopping Campaigns (ASC) structure into the Sales objective; ASC is no longer a separate campaign type in Ads Manager but a default Advantage+ toggle inside a Sales campaign. The legacy ASC and Advantage App (AAC) programmatic APIs are also being phased out: by industry sources, Marketing API v24.0 (October 8, 2025) blocked new ASC and AAC creation on that version, and v25.0 (Q1 2026) ends new creation across all versions; for exact dates you should rely on Meta's official Marketing API changelog. The existing customer budget cap limits how much of the budget is spent on existing customers; without a cap, the algorithm concentrates on easy-to-convert existing customers, inflating the attributed ROAS figure while suppressing new-customer acquisition. The key to scaling Advantage+ Sales is creative volume and quality rather than budget, so run the campaign for seven to fourteen days before evaluating it.

Is Advantage+ Really Better? Meta's Claims vs Independent Evidence

Meta's own reported figures are optimistic: an average of about $4.52 ROAS for Advantage+ users (which Meta frames as roughly 22% higher ROAS than business as usual) and, in a study of fifteen A/B tests, about 12% lower cost per purchase for Advantage+ Shopping. Those values are averages from Meta's internal A/B testing, not independently audited, and they hide wide variance, so they should be read as a best-case ceiling. Independent evidence complicates the picture: by attribution vendor Wicked Reports' analysis of 55,661 campaigns, Advantage+ new-customer acquisition cost more than doubled in a year (from about $257 to about $528), while manual campaigns acquired new customers at least as cheaply as the prior year. On the incrementality side, Haus's 640 lift experiments found that 58% of brands saw higher incremental ROAS on manual campaigns; with an important correction, the same study found Meta's in-platform attribution under-reports incrementality by about 15% on average, and that Advantage+ drove a larger share of its impact on new customers (about 70% versus 65%). The issue is not that "platform ROAS is always inflated" but a mismatch between attributed and true incremental ROAS, which is why, as measurement specialists warn, you should validate with holdout or geo lift tests rather than trust high in-platform ROAS. For adoption context: by Tinuiti's Q2 2025 panel, about 35% of US retail advertisers' Meta spend went to Advantage+ Shopping (this is share of Meta budget within the panel, not total retail ad spend), and Meta reported over four million advertisers using its generative AI ad tools in the same period.

Creative Strategy: The First 3-Second Hook, Mobile-First and UGC

The mobile-first default format is 9:16 vertical (1080x1920): the large majority of Meta inventory is vertical and almost all users are on mobile, so produce creative vertically first. The first one to three seconds decide whether someone keeps watching or scrolls, so test at least two different hooks against the same body; the best structures are a problem-first open, a pattern-breaking visual and a direct benefit promise. Design for silent viewing: by industry sources the large majority of Facebook videos are watched with sound off (the commonly cited 85% figure rests on older data and the exact rate changes over time), so burned-in captions should be treated as effectively mandatory. User-generated content (UGC) tends to drive higher engagement than polished creative with cold audiences because it resembles organic content, so use polished brand content in retargeting and the trust stage. Creative diversity is the number one performance lever of 2026 because Andromeda prioritizes creative variety over targeting, so about fifteen to twenty creatives per campaign spanning different hooks and formats are recommended, since creative fatigue at high frequency can appear within one to two weeks.

Measurement: Meta Pixel + Conversions API (CAPI), EMQ, AEM and iOS Privacy

The Meta Pixel alone is not enough: iOS privacy restrictions, ad blockers and consent banners cause Pixel-only setups to miss a large share of conversions. The Conversions API (CAPI) sends events server-side directly to Meta and recovers part of that lost signal; Meta recommends Pixel and CAPI together for every advertiser in 2026, because thin first-party signal is the main condition under which Advantage+ automation underperforms. When running both, deduplication is critical: events sharing the same event_id and the same event_name are counted once, and the naming must match exactly, including case. Event Match Quality (EMQ) is a score from one to ten measuring signal quality; the target is seven or higher, so send as many valid identifiers as possible with each event, such as email, phone, the fbp and fbc cookies, external_id, first and last name, and address. Aggregated Event Measurement (AEM) is the privacy-preserving layer for iOS users who opt out of tracking under Apple's ATT; Meta removed the eight-event limit in June 2025 but this did not fix iOS tracking, since the data remains aggregated, delayed (up to about seventy-two hours for privacy thresholds) and partly modeled, and Meta also removed the seven-day-view and twenty-eight-day-view attribution windows in January 2026. Giving a single ATT opt-in percentage is misleading because the denominator changes the result: across the whole user base it was roughly 25% in 2025, while among only those who saw and answered the prompt AppsFlyer reports about 50% and Adjust about 35%; the old 89% opt-out figure is outdated and should not be used.

Account Security, Policy and Special Ad Categories

Ensuring account security and policy compliance before running ads is the step new advertisers most often skip. Turn on two-factor authentication in Business Manager, define spending limits and set up the payment method under the business name; if an account is disabled, the appeal process runs through the Account Quality panel. Meta's Special Ad Categories restrict targeting on certain topics: credit, employment and housing, along with social issues, elections and politics, fall under this scope, and in these categories targeting options such as age, gender and detailed location are limited. Financial products and cryptocurrency assets are subject to additional approval and restrictions in many regions, so an advertiser must confirm local eligibility from Meta's current policy page. In addition, phrases like guaranteed results, risk-free or guaranteed high returns violate Meta's Unrealistic Outcomes policy and lead to ad rejection; no legitimate ad offers guaranteed returns.

The Turkish Market: Cost Ranges, Localization and VAT

Turkey is a large market (by agency estimates, roughly 38 million Facebook and over 52 million Instagram users). Cost figures for Turkey are agency estimates and change rapidly with industry, season, creative quality and the lira exchange rate, so any single exact number is unreliable; common planning ranges are cost per click (CPC) roughly 1 to 5 TL, cost per thousand impressions (CPM) roughly 15 to 60 TL, and ecommerce cost per acquisition (CPA) from about 80 TL to 500 TL and above depending on the product. Global comparisons also argue against trusting a single number; sources disagree widely, with CPC roughly $0.50 to $3.50 and CPM $5 to $15, and single decimal "precise" numbers give a false sense of accuracy. Cost varies greatly by industry: the lowest CPC is in categories like fashion and food, the highest CPC in services like law and insurance, and CPM values rise in the fourth quarter and discount weeks. On billing, Meta issues the ad invoice in US dollars; if you correctly enter your tax ID (VKN) in the payment settings, Meta does not apply VAT and the declaration responsibility shifts to the business, otherwise a bank or intermediary may collect VAT, so confirm the current situation from Meta's billing screen and with your accountant. It is essential to determine your own numbers by testing with a small budget in your social media campaigns.