How to Lower Your Meta Ads CPM

by Damien Hurwitz
on October 23, 2026

Meta advertising costs can change considerably from one campaign to another, even when two advertisers are targeting similar audiences.

Tinuiti’s Q4 2025 digital advertising benchmark report found that impressions on Meta properties increased 17% year over year while CPM fell 7%. At the same time, 44% of the advertisers in its analysis saw their CPM increase year over year. In other words, there is no single CPM that applies to every Meta campaign. Audience, creative, placements, objectives, competition, and timing all play a role.

CPM, or cost per mille, measures how much you pay to generate 1,000 ad impressions.

The basic formula is:

CPM = (Ad Spend ÷ Impressions) × 1,000

A lower CPM means you are buying impressions at a lower cost. That can help stretch a media budget, but CPM should be reviewed alongside click-through rate, cost per click, conversions, cost per acquisition, revenue, and return on ad spend.

A $7 CPM is not necessarily better than a $15 CPM if the $15 CPM campaign generates significantly more customers. So how can you lower your Meta Ads CPM without sacrificing the quality of your traffic?

1. Start With the Right Baseline

Before changing your campaigns, figure out whether your CPM is actually high for the specific campaign you are running.

A prospecting campaign targeting a broad audience should not be compared directly with a retargeting campaign aimed at people who already visited your website. A lead-generation campaign should not necessarily have the same CPM as a campaign optimized for reach.

The same applies to geography and audience size.

A campaign targeting a large U.S. audience can face a very different auction environment from one targeting a small local market. Industry and season can also affect competition for available impressions.

For a directional 2026 reference, Biddyco’s Meta Ads statistics reported a median CPM of $18.73 for June–August 2026 among U.S. advertisers spending 100,000-2.3 million per month.

That figure should be treated as a benchmark for comparison rather than a target, since actual CPMs vary based on audience, industry, campaign objectives, and budget.

Your own historical performance is usually more useful. Compare your current CPM with previous periods that have similar objectives, audiences, placements, and seasonal conditions.

2. Broaden Your Audience

A very narrow audience can make it more expensive to win impressions.

This is partly because you are competing for a smaller pool of people, and other advertisers may be targeting many of the same users. Broadening an audience gives Meta more potential impressions to evaluate.

This does not mean removing every audience restriction.

Start by reviewing whether your targeting is narrower than it needs to be. Consider whether detailed interests, stacked exclusions, narrow age ranges, or other restrictions are limiting delivery without producing better customers.

Meta has continued moving toward AI-assisted audience selection. Its Advantage+ audience tools use advertiser-provided audience suggestions and can expand delivery based on Meta’s models when the system expects that doing so can improve results.

For many campaigns, giving Meta more room to find potential customers can be more effective than manually defining an extremely specific audience.

3. Test Advantage+ Placements

Another way to give Meta more inventory to work with is to use Advantage+ placements.

Rather than restricting a campaign to a single placement such as Instagram Feed, Advantage+ placements can distribute ads across eligible Facebook, Instagram, Messenger, and Audience Network placements.

Meta recommends Advantage+ placements as a way to give its delivery system more opportunities to use the campaign budget.

Placement mix can have a meaningful effect on CPM.

Tinuiti’s Q4 2025 data found that Meta impressions increased while CPM declined, with the growth of newer, lower-cost inventory such as Reels contributing to the shift.

That does not mean every campaign should automatically use every placement without review. Creative still needs to make sense in each environment.

The better approach is to give Meta enough placement flexibility, then monitor the quality and business results coming from each area.

4. Optimize Creative for More Than One Placement

A single square image may technically work in multiple locations, but that does not mean it is the best asset for each one.

Meta’s Reels guidance recommends vertical 9:16 video, audio, and keeping important elements within the safe zone. In Meta’s analysis of 15 split tests, Reels campaigns using this type of creative had a 34.5% lower cost per result than campaigns using image ads in the tested setup.

That statistic is about cost per result, not CPM, so it should not be presented as evidence that vertical video always lowers CPM.

The larger takeaway is that creative should fit the placement.

For example:

Instagram Feed:
A strong product image or short video can work well.

Stories and Reels:
Vertical video with a quick opening message is more natural.

Facebook Feed:
A testimonial, product demonstration, or educational format may work well.

Giving Meta assets that fit its available placements can make it easier for the system to deliver campaigns across more inventory.

5. Improve Your Ad’s Initial Hook

Creative quality can influence auction performance.

Meta says its ad ranking systems use AI models to assess which ads are likely to resonate with different audiences. In January 2026, Meta reported that improvements to its ad-ranking systems drove a 3.5% increase in ad clicks on Facebook and more than a 1% increase in conversions on Instagram during Q4 2025. Meta also reported a 12% increase in ad quality on Facebook after combining improvements to its ranking models and backend systems.

For advertisers, that makes the creative itself an important part of the campaign setup.

The first few seconds or first visual should make the message clear.

A clothing brand might lead with the product being worn rather than a logo.

A home services company could open with the problem it solves.

An electronics brand might show the product in use immediately rather than beginning with a long introduction.

The objective is simple: make the ad relevant enough that the person has a reason to keep watching or reading.

6. Refresh Creative Before Performance Falls Too Far

An ad can perform well at launch and become less effective after the same audience has seen it repeatedly.

As people become familiar with the same image, video, headline, or offer, engagement can change. Frequency can rise, and the campaign may have fewer fresh ways to capture attention.

Watch CPM alongside:

  • Frequency
  • CTR
  • Cost per click
  • Conversion rate
  • Cost per acquisition
  • Comments and other feedback

A rising CPM paired with declining CTR can be a sign that the creative needs attention.

Do not wait until the campaign completely stops working before testing another concept.

Instead, build a steady creative testing process. For an apparel client, that might mean rotating product photography, creator-style videos, customer testimonials, lifestyle images, and different promotional messages rather than producing several minor variations of the same ad.

7. Avoid Running Too Many Small Ad Sets

Account structure can affect how much useful data Meta has available.

A common mistake is splitting one audience into many small ad sets because each segment seems important.

For example, a business might create separate ad sets for:

  • Women ages 25–34
  • Women ages 35–44
  • Instagram shoppers
  • Beauty interests
  • Online shoppers
  • Lookalike audiences

Some of those groups may overlap heavily.

Instead of giving Meta one larger pool of potential customers to work with, the budget and conversion signals become distributed across several smaller groups.

Consolidating similar audiences can give the delivery system more data within each ad set and reduce unnecessary competition between your own campaigns.

The goal is not to eliminate segmentation altogether. Segmentation should have a reason behind it.

8. Give Meta Better Conversion Signals

Meta’s advertising system has become increasingly automated.

In 2026, Meta is investing heavily in AI systems that determine which ads to show to which people. Meta says its newer ad-ranking models use longer sequences of user behavior and additional engagement signals to improve predictions about which ads will resonate.

That makes accurate conversion data especially important.

If your campaign is optimized for purchases, Meta needs reliable purchase information. If it is optimized for qualified leads, your tracking setup should help distinguish meaningful leads from low-value submissions.

Make sure your:

  • Meta Pixel is installed correctly
  • Conversions API is set up where appropriate
  • Purchase or lead events are firing correctly
  • Event values are accurate
  • Campaign optimization event matches the business goal

Better data does not guarantee a lower CPM, but it gives Meta stronger information to use in delivery.

9. Don’t Chase the Cheapest CPM With the Wrong Objective

This is one of the most important points in the article.

An advertiser can sometimes lower CPM by changing the campaign objective or optimization event, but that does not automatically improve the campaign.

A campaign optimized for reach is asking Meta to find people who are likely to be reached. A sales campaign is asking the system to find people who are more likely to purchase.

Those are different jobs.

A sales campaign can have a higher CPM than an awareness campaign because Meta is competing for impressions among people its system expects are more likely to produce the desired action.

That higher CPM may still be worthwhile if the campaign produces better customers.

Instead of asking:

“How do I get the lowest CPM?”

ask:

“What CPM can this campaign afford while still producing profitable results?”

That is a much more useful metric for budget decisions.

10. Test Different Creative Formats

Meta gives advertisers several creative formats to work with, including image, video, carousel, and Reels. Testing different formats can help you determine what your audience responds to and which inventory your ads can use effectively.

For example, an ecommerce brand could test:

Product photo:
A clean image with a clear product benefit.

Product demonstration:
A short video showing how the product works.

Customer testimonial:
A customer explaining why they bought it.

Creator-style video:
A more informal presentation designed for social feeds.

Carousel:
Several products or features presented in one ad.

Do not test only visual format. Test the message too.

A product-focused ad, a discount-led ad, and a problem-solution ad may attract very different responses from the same audience.

 

11. Watch Frequency and Audience Saturation

A low CPM is harder to maintain when the same people keep seeing your ads.

Frequency measures how often the average person in your audience sees the campaign during a given period.

A campaign targeting a small remarketing audience may have a naturally higher frequency than a large prospecting campaign. That does not automatically mean something is wrong. It does mean you should look at frequency alongside CPM and other performance metrics.

If frequency keeps rising while CTR falls and CPM increases, the audience may need more creative variety or more room to find new people. Expanding the audience, adding new creative, or adjusting the campaign structure can help address saturation.

12. Account for Seasonality

Meta auction prices can change as advertiser demand changes.

Retail advertisers, for example, may experience a different competitive environment around major shopping periods than during quieter parts of the year.

That means you should avoid treating a temporary CPM increase as proof that something has gone wrong with the account.

Look at your historical data.

Compare:

September 2026 vs. August 2026

and, when useful:

September 2026 vs. September 2025

A month-over-month increase may look concerning until you compare it with the same period from the prior year.

This is another reason your own campaign history is more useful than a universal CPM benchmark.

13. Make Changes Carefully During Learning

New ad sets need time to gather delivery and conversion information. Significant changes can send an ad set back into learning, which can make performance less stable. WordStream notes that major changes to budget, targeting, placements, bidding, or creative can trigger another learning period.

That does not mean you should leave a broken campaign running. It means you should distinguish between a genuine problem and normal early-stage volatility.

Constantly changing the audience, budget, creative, and optimization event at the same time makes it difficult to determine what actually changed performance.

A better process is to make deliberate changes, document them, and give the campaign enough time to produce useful data before drawing conclusions.

What Causes a High Meta Ads CPM?

If your CPM suddenly increases, check these factors first:

Audience:
Is the audience unusually narrow or overlapping with another campaign?

Placement:
Are you limiting the campaign to a small number of placements?

Creative:
Has the same ad been running for too long?

Frequency:
Are the same people seeing the campaign repeatedly?

Seasonality:
Are more advertisers competing for the same inventory?

Objective:
Are you optimizing for a high-value action that naturally requires more selective delivery?

Tracking:
Are your conversion events accurate and firing as expected?

Campaign structure:
Have you split a potentially unified audience across too many ad sets?

Looking at the full picture is more useful than changing one setting simply because the CPM looks high.

Should You Always Try to Lower Your CPM?

No.

A campaign with a $10 CPM and a $100 cost per purchase may be less useful than a campaign with a $20 CPM and a $50 cost per purchase.

The first campaign buys cheaper impressions. The second may buy more valuable impressions.

That distinction matters because Meta’s systems are increasingly built around predicting actions, not simply delivering the largest number of impressions at the lowest possible price. Meta says its advertising ranking models consider signals about how likely people are to engage with ads and the quality of the ad experience.

For that reason, CPM should be part of the performance conversation, not the entire conversation.The most useful reporting framework connects CPM to the metrics that follow it:

CPM → CTR → CPC → Conversion Rate → CPA → Revenue/ROAS

If CPM rises but CTR and conversion rate improve enough to keep acquisition costs in line, the campaign may still be working well. If CPM falls while the quality of traffic declines, the lower number may not be an improvement.

Final Thoughts

Lowering your Meta Ads CPM is less about finding a single setting and more about improving the conditions under which your ads compete.

Start by checking whether your CPM is actually high for the objective, audience, geography, and time period. Then look at the areas you can control: audience breadth, placement availability, creative quality, conversion signals, account structure, and testing.

Meta’s advertising system is also becoming more automated in 2026. Its AI models have an increasingly large role in audience selection, ad ranking, creative delivery, and campaign optimization. That makes strong creative and reliable conversion data more important, not less.

Most importantly, do not optimize toward the cheapest CPM at the expense of the outcome the campaign was built to produce.

A healthy Meta campaign is one where the media cost makes sense for the value being generated.

Sandhill Digital helps businesses manage paid social campaigns across Meta and other platforms, with strategy, creative testing, audience development, conversion tracking, and performance reporting built around measurable business goals.

Sources

Book A Free Strategy Consultation

Consultation Contact Form
Name
Name

This is a success preview text.

This is a error preview text.