The Complete Guide to Meta's Advertising Changes: Targeting, Costs and What to Do Now

Meta has fundamentally changed how its advertising platform works.

This complete guide covers every major targeting change, why costs are rising, and exactly what businesses need to do to adapt and protect their results.

If you manage advertising on Facebook or Instagram or you pay someone to do it for you, 2026 has been a year of significant, sometimes unsettling change and we’re only 3 months in!

Targeting options have been removed. Costs have increased. The algorithm now behaves differently to how it did even eighteen months ago.

Some of these changes were well communicated. Most weren't. And for businesses running campaigns without specialist support, the result has often been a quiet erosion of performance: the same budget producing fewer leads, higher costs per click, and a growing sense that something isn't working the way it used to.

This guide covers everything that has changed on the Meta advertising platform since early 2025, why it happened, what it means for your ad spend, and, most importantly, what to do about it.

Part 1: The targeting changes explained

The removal of detailed targeting exclusions

The most immediately disruptive change for many advertisers came in March 2025, when Meta removed the ability to exclude audiences based on interests in new ad sets.

Previously, exclusions were a standard part of campaign setup. If you were running a campaign for a premium product, you might exclude audiences associated with discount shopping. If you were targeting business owners, you might exclude students. If you were running a local campaign, you might exclude interests that skewed toward audiences outside your geography.

That level of control is now gone.

The timeline of how this was rolled out:

• March 31, 2025: Detailed targeting exclusions removed from new ad sets in Ads Manager
• June 10, 2025: Exclusions removed from boosted posts on Facebook and Instagram
• June 23, 2025: Meta began consolidating interest categories into broader groupings
• January 15, 2026: All existing ad sets using discontinued targeting options stopped delivering

If you're still running campaigns created before these dates and haven't updated them, there's a chance your ads are either underperforming or have stopped delivering altogether. Warning banners in Ads Manager will flag affected campaigns, but these are easy to miss if you're not checking regularly.

Meta's own justification for removing exclusions was performance-based: internal testing reportedly showed a 22.6% lower median cost per conversion when exclusions were not used. The argument is that exclusions, while feeling like precision, were actually limiting the algorithm's ability to find genuinely interested users in unexpected audience segments.

Whether you accept that argument or not, the change is permanent.

The consolidation of interest categories

Alongside the removal of exclusions, Meta has been steadily merging and retiring hundreds of granular interest categories. Categories tied to specific sports, music genres, food preferences, car models and lifestyle behaviours have been absorbed into broader groupings.

When advertisers try to use a discontinued interest, Ads Manager now suggests alternatives, but the alternatives are often significantly broader. What was once a precise targeting signal becomes a much larger, less defined audience. For advertisers who built their strategies around specific niche interests, this has been a meaningful loss of control.

The direction of travel here is clear: Meta is pushing advertisers away from interest-based audience definition and toward AI-driven audience discovery.

What's still available

It's worth being clear that not all targeting options have been removed. You can still:

• Target by location (country, region, city, or radius)
• Target by age and gender
• Use Custom Audiences based on your own data (customer lists, website visitors, app activity)
• Use Lookalike Audiences based on your customer data
• Use Advantage+ audience tools, which blend your inputs with Meta's AI

The shift is specifically away from interest and behaviour-based manual targeting, and toward first-party data and algorithmic audience finding.

Part 2: The Andromeda engine and what it means for your creative

What Andromeda is

In late 2024, Meta introduced a new ad delivery infrastructure called Andromeda, powered by NVIDIA's GH200 chip.

It processes advertising decisions roughly 100 times faster than the previous system and can handle approximately 10,000 times more ad variants simultaneously.

The practical consequence of this is a fundamental shift in how ads are matched to audiences.

Under the old system, you defined your audience, and Meta found the best ad to show them. Under Andromeda, your creative content acts as the primary targeting signal. The algorithm reads your ad, its visual content, copy length and tone, format, and the audience signals it contains, and uses this to determine who is most likely to engage with it.

In other words: creative is now targeting.

What this means in practice

This shift has significant implications for how you approach campaign strategy.

An ad that looks and sounds like it's aimed at small business owners in professional services will find small business owners in professional services, not because you've filtered for them, but because the algorithm has learned that this type of content resonates with that audience.

An ad that uses lifestyle imagery associated with a particular demographic will surface to that demographic, again without you manually specifying it.

The quality, variety, and relevance of your creative is now doing the job that audience segmentation used to do.

Meta's own data supports this. Accounts using diverse creative libraries, multiple formats, multiple angles, regular refreshes, tend to scale faster and maintain lower CPMs than accounts running a small number of static creatives for extended periods.

What good creative looks like under the new system

Format variety matters. Your campaign should include static images, short-form video, carousels, and where appropriate, founder or team-facing content.

Static images still account for approximately 60–70% of Meta conversions, so video-only strategies are not the answer, but variety across formats keeps CPMs stable and gives the algorithm more to work with.

Creative fatigue is more damaging than before. If Meta's algorithm perceives that it is showing the same visual repeatedly, it may reduce delivery or increase CPMs as a result. Running the same ad for more than four weeks without refreshing is increasingly risky.

Authenticity signals perform strongly. Organic-looking content, founder videos, behind-the-scenes footage, real customer testimonials, often outperforms polished production content. This reflects broader changes in how people engage with content on these platforms.

Psychological variety. Different ads should address different motivations: pain points, aspirational outcomes, social proof, direct questions, curiosity. Each angle will find different users, giving the algorithm a richer picture of who responds to your brand.

Part 3: The cost picture

Why prices are rising

Meta ad costs have increased meaningfully and consistently over the past year. Understanding why helps you respond strategically rather than simply accepting higher costs as an inevitability.

More competition in the auction.

More businesses are running Meta ads than ever before, and they are increasingly using sophisticated, AI-assisted tools to bid more effectively. The result is an inflated auction environment: more advertisers bidding for the same impressions drives prices upward.

Meta's infrastructure investment.

Meta has reportedly invested between $64–72 billion in AI infrastructure and data centres in 2025 alone. This investment is ultimately reflected in the pricing of its ad products. Advertisers are, in effect, partly funding the platform that manages their campaigns.

Signal loss from iOS tracking restrictions.

Apple's iOS privacy changes have removed a significant proportion of tracking signals from Meta's attribution system, estimated at 20–30% of measurable conversions. This creates an invisible cost: cheaper conversions that go unmeasured inflate your apparent CPC and CPM figures, making performance look worse than it sometimes is.

Reduced targeting precision.

Counterintuitively, the removal of granular targeting has contributed to higher initial costs for some advertisers. Without the ability to precisely define audiences, the algorithm requires a learning period during which it spends budget finding the right people. This initial inefficiency has a cost.

The numbers

To give context to the scale of these changes:

• CPM increased by over 20% across all industries on Meta in 2025, with no industry exempt from rising costs
• Tier 1 markets (including the UK) experienced a 12% year-on-year cost increase through late 2025
• UK CPM for eCommerce campaigns typically sits between £5–£10; for lead generation it runs higher
• Cost per lead for Facebook lead campaigns rose approximately 20% year-on-year in 2025
• During the 2025 holiday period, CPMs spiked by as much as 66% in competitive e-commerce categories
• Instagram CPM surpassed Facebook CPM in Q2 2025, reflecting growing competition for Instagram placements

For context, this cost inflation is not unique to Meta, Google Ads CPC rose approximately 12.88% across 87% of industries in the same period.

Digital advertising has become a more expensive discipline broadly, and Meta's changes have simply made this more visible to advertisers who previously relied on precise targeting to keep costs down.

Part 4: How to adapt your strategy

1. Prioritise first-party data above everything else

In the new Meta environment, your own customer data is the most valuable asset you have. Customer email lists, website visitor data, past purchasers and enquirers, these form the foundation of effective campaigns.

Upload your customer list to create a Custom Audience. Build Lookalike Audiences from your best customers. Use your CRM data to create audience segments based on behaviour rather than purchased interest data.

First-party data gives the algorithm something reliable to learn from, which shortens the learning phase, reduces wasted spend, and improves the quality of algorithmic audience expansion.

2. Implement the Conversions API (CAPI)

If you are still relying solely on the Meta Pixel for conversion tracking, you are operating with incomplete data. Meta's own analysis suggests that over 50% of browser-side conversions now go untracked due to iOS privacy restrictions and cookie consent requirements.

The Conversions API (CAPI) sends conversion data directly from your server to Meta, bypassing browser-level restrictions. Implementing CAPI alongside the Pixel can recover up to 15% of lost attribution signal, which means a more accurate picture of performance, faster algorithm learning, and typically lower costs.

This is no longer optional for any business running consistent Meta campaigns.

3. Build a diverse creative library

Given that creative is now the primary targeting signal, your investment in creative quality and variety has a direct impact on campaign performance.

Practically, this means:

• Refreshing your creative at least every two to four weeks
• Testing multiple formats simultaneously (static, video, carousel)
• Testing multiple angles (pain-point led, benefit led, testimonial, direct question)
• Exploring authentic, lower-production content alongside polished assets
• Using Meta's Dynamic Creative or Flexible Creative tools to let the algorithm identify the best-performing combinations

4. Embrace broader targeting with good inputs

For many businesses, moving to broader targeting feels uncomfortable. The impulse is to narrow down, to feel like you're being precise. But Meta's own data and independent testing consistently show that broad targeting combined with strong first-party signals outperforms narrow manual targeting.

Meta's testing showed a 32% reduction in cost per acquisition when advertisers moved from narrow to broad targeting, provided the algorithm had good first-party data to learn from.

The key shift in mindset: instead of asking "who do I want to reach?", ask "what do I need to give the algorithm so it can find my best customers?"

5. Audit your existing campaigns

If you haven't reviewed your campaigns since mid-2025, there are likely issues to address.

Specifically:

• Check for warning banners relating to discontinued targeting options
• Remove or replace any interest exclusions from saved audiences and draft ad sets
• Review whether any ad sets have stopped delivering unexpectedly
• Check your CPM trends a steady upward drift in CPM often signals creative fatigue

6. Rethink how you measure performance

Meta introduced incremental attribution in 2025, which measures only conversions that would not have occurred without your ad.

This is a more honest, and often less flattering, measure of ad performance than last-click attribution.

Engaged-view attribution thresholds also changed, dropping from 10-second views to 5-second views. More conversions are now credited to video ad views, which can inflate reported results for video campaigns.

Understanding which attribution model your reports are using, and whether it reflects true incremental value, is increasingly important for making good decisions about where to spend your budget.

Part 5: The bigger picture

It would be easy to read all of this as bad news.

Costs are up, control is down, the rules have changed.

But there's a more useful way to look at it.

Meta is still, for most SMEs, one of the most cost-effective advertising channels available. Even with a 20% cost per lead increase, Facebook lead generation still costs significantly less than equivalent Google Ads campaigns, and a fraction of LinkedIn's typical B2B rates. The reach, currently totalling 55.9 million UK users, is unmatched on any other single platform.

What has changed is what it takes to do it well. The era of set-and-forget campaigns with tight interest targeting is over. The businesses that will perform well on Meta in 2026 and beyond are those that invest in creative quality, maintain their first-party data, understand the algorithm they're working with, and treat the platform as a system to feed rather than a machine to control.

That requires more strategic thinking than simply setting up a campaign and checking it monthly. But for businesses that make that investment, the returns are still very much there.