Meta Ads Audit: Evaluation of Ad Sets

The audit of a Meta advertising account includes several layers you should consider. One is specific to the history of what this client does within their ad sets.

This is the third post in a series about what to look for when running a Meta ads audit. This is not a simple topic, so I am breaking it down into multiple posts. You can read the first two here:

The ad set is critical because so much can be controlled there which has the potential to impact results. As discussed in prior posts, this isn’t about labeling something “right” or “wrong” in most cases. It’s about observing how a client has structured their advertising in the past to provide context for their results. This will help highlight potential improvements.

Let’s begin…

Ad Set Volume

Does this client have a complicated campaign structure or a simple one? Maybe they’ve found a complicated system that works. But if performance is an issue, causes tend to start here.

There are several problems that can be sourced to a high volume of ad sets…

1. Auction Overlap.

When you create multiple ad sets, you will likely create an environment where you can reach the same person with multiple ads. This is especially possible in this era of algorithmic targeting and audience expansion where the algorithm is going to reach the people it wants to reach.

Auction Overlap prevents you from having two ads entering the same auction so that you don’t drive up costs by competing with yourself. In the process, it prevents your ad sets from performing optimally.

2. Audience Fragmentation.

One of the reasons that advertisers create multiple ad sets is to segment the audience. While Auction Overlap can happen when you create multiple ad sets that have the ability to reach the same people, some advertisers will go the opposite route and create multiple ad sets to break up a larger audience.

An example could be creating different ad sets by state or country, gender, or age group. While there can be a benefit for doing so, a large budget would be required. Otherwise, fragmenting your audience in this way will make delivery less efficient.

3. Watered Down Budget.

Let’s assume a hypothetical situation where a client spends $100 per day on ads. If one ad set is used, the entire $100 can be dedicated to that one ad set to make sure it runs optimally. But if they complicate things by creating 10 ad sets, each one will spend $10 per day, on average.

Even if Auction Overlap and Audience Fragmentation aren’t an issue, breaking up your budget can and will hurt performance. You are unlikely to exit the learning phase, and your results are sure to be unstable and unpredictable.

When evaluating a client, make note of the number of ad sets they use overall and the average number of ad sets per campaign. If they are struggling to get good results, it may be helpful to consolidate some of their ad sets.

Conversion Locations

Does this client tend to send people to their website? Instant forms? Messenger? Somewhere else? Make note of these strategies.

It is generally recommended that you prioritize sending people to your website, but that can come with unique challenges. If they do prioritize sending people to their website, do they have tracking issues? Reporting inconsistencies?

If running leads campaigns, do they use instant forms or send people to their website? Both have advantages and disadvantages. But the challenges they face may be reflected in this choice.

When prioritizing sending people to your website, other elements will become critical (we’ll get into this more in a later post):

1. Is the web page well designed?
2. Does the website load slowly?
3. Are the pixel and API installed and are events tracking properly?

These challenges are also why using instant forms can be preferred in some cases.

Performance Goals

The performance goals that are used by an ad account will say a lot about the strategies that they employ. Do they go straight for the conversion, or do they focus on top-of-funnel actions?

Some of the more common performance goals:

• Maximize number of conversions
• Maximize value of conversions
• Maximize number of leads (instant forms)
• Maximize number of landing page views
• Maximize ThruPlay views

The performance goal is a critical element because it defines success, which is what guides the algorithm. Since the algorithm is literal, this can lead to lots of cheap, low-quality results when optimizing for anything other than a conversion — and even for leads.

That doesn’t mean that there’s never room for top-of-funnel optimization, but it needs to be done very carefully. Optimization for conversions should also take priority, using the majority of your budget. Top-of-funnel optimization, for example, should not steal budget from campaigns optimized for conversions.

The higher the overall budget, the more freedom you have to spend money on awareness and engagement campaigns. But for the vast majority of brands, this could be a huge waste of money.

Make note how this client is using performance goals and what percentage of their budget is assigned to each.

Conversion Events

When using performance goals associated with conversions, the conversion event sets the specific action that defines success. The algorithm isn’t looking for just any conversion. Instead, depending on what you select, it will optimize for more purchases, leads, or something else.

There are three primary groups of possibilities…

1. Standard Events.

Standard events are those that any advertiser can use. Examples include purchase, lead, and complete registration. These are recommended since Meta will have data from other advertisers to better understand exactly what you’re looking for.

While you can optimize for the purchase or lead standard event, some advertisers also choose something further up the funnel, like view content, initiate checkout, or add to cart. This is often done as a last resort when purchases aren’t consistently achievable.

2. Custom Events.

These are events that can’t be defined with a standard event and are unique to a specific brand or publisher’s business. Sometimes advertisers use custom events when a suitable standard event was available, usually due to a misunderstanding of when these are necessary.

I’ve used custom events to better define deep website engagement (time spent, scroll depth, and more). This provides the option of optimizing for quality engagement on a blog post, for example, rather than optimizing for a link click or landing page view.

3. Custom Conversions.

Custom conversions help categorize a standard or custom event. For example, let’s assume you’re optimizing for the purchase standard event. Your results column will be for all attributed purchases. But that doesn’t require that the purchases be for the specific product that you are promoting. Someone could click on your ad and purchase something else. Custom conversions allow you to segment reporting but also optimize for a more refined event.

All three approaches have a place. I’d generally recommend standard events whenever a standard event is suitable for the action you want. Advertisers can otherwise overthink and overcomplicate this by going with custom events or custom conversions when they don’t need to.

Make note of the events that your client optimizes for. Do they go straight for the bottom of the funnel or do they hang around the top or middle? Do they prioritize standard events or do they consistently overcomplicate things?

These decisions can have a profound impact on results. Make note of them.

Attribution Settings

When using the Website conversion location and using a performance goal to maximize the number or value of conversions, an attribution setting is used. By default, it’s 7-day click, 1-day view, and 1-day engaged-view.

This is how Meta gives credit to ads for a conversion. In other words, a conversion will be credited if someone is served your ad and converts within seven days of clicking your ad or one day of viewing it without clicking. This impacts both optimization for delivery and your default reporting.

Most advertisers use the default 7-day click and 1-day view in all cases. They shouldn’t. Here’s what I recommend…

1. 7-day Click and 1-Day View.

Use this when optimizing for the purchase event. Not all purchases are immediate or in a straight line. Your ads can contribute in many ways, and you’ll want to get that credit (which you’ll need to contextualize).

2. 7-Day Click.

If the client uses remarketing strategies to promote the purchase of a product, I’d advise removing 1-day view. Otherwise, your results can be significantly inflated by purchases that would have happened without your ad.

3. 1-Day Click.

If you’re optimizing for a non-purchase event, you should almost always use 1-day click only. Assuming it’s a free thing that requires little thought, it’s debatable whether anything beyond 1-day click actually contributed to your results. Leaving the attribution setting at the default might both inflate your numbers and confuse the algorithm, encouraging it to focus on flimsy conversions.

Make a note of the attribution settings your client uses and when. My recommendation is to be hands-off and use the defaults when optimizing for a purchase, but remarketing and non-purchase events should be treated differently. The chosen approach will have an impact on both the reported results and the quality of those results.

Bidding

There was a time when manual bidding was a big part of an advertiser’s toolbox. These days, it’s generally recommended that you allow the algorithm to bid for you. There are exceptions, but they are rare.

The default bid strategy in most cases is Highest Volume. In other words, the algorithm will focus on getting you the highest volume of optimized actions within your budget.

But there are some other options:

1. Cost Per Result Goal.

Meta will aim to keep the average cost per purchase, lead, or other optimized action to the amount you set. While this sounds reasonable, it can also lead to low-quality results when not optimizing for a purchase. The algorithm also no longer cares about spending your budget, and you may struggle to deliver your ads if the goal can’t be maintained.

2. Highest Value.

This is the default when “Maximize value of conversions” is selected. Instead of focusing on volume, the algorithm will prioritize the overall Return on Ad Spend. This is especially useful when selling products at very different price points.

3. ROAS Goal.

When maximizing value, a ROAS Goal would restrict the algorithm in an effort to maintain a specific return. This doesn’t guarantee that the ROAS that you set is possible or realistic. If it can’t be reached, your ads will likely stop delivering.

4. Bid Cap.

The Bid Cap is the most you’ll spend in the auction, which is different from a Cost Per Result Goal. A low bid doesn’t guarantee a low Cost Per Result and a high bid doesn’t guarantee a high Cost Per Result.

The bottom line is that you should rarely use a manual bid strategy. My recommendation is to only use it as a last resort when efforts haven’t been profitable and you’re trying to control costs. Since manual bidding will negatively impact volume, this would also only be recommended for the highest budgets so that optimal volume can be maintained.

Only advertisers who know what they’re doing and have a very clear approach should use manual bid strategies. More often than not, advertisers will use them in an attempt to show how smart and necessary they are. But they rarely lead to better results.

Make a note of the bid strategies that are used and when. This will be something to potentially adjust based on results.

Budgeting

Budgeting is one of the most critical elements of ad performance. You need to spend enough to get optimal results.

Of course, there are two elements of your budget…

1. Ad Account Budget.

How much is spent overall every day? Every month? This matters quite a bit. The overall available budget impacts the options that are available to you. A high ad account budget gives you the freedom to add more complexity. You can run advertising for different stages of the funnel (sales, leads, traffic, awareness). You can add complexity and get creative.

Small budgets don’t have this freedom. They should consolidate their budget as much as possible, focusing on the stage or stages of the funnel that are most important. They should limit their number of ad sets and avoid extra complications that only make their results worse.

2. Ad Set Budget.

This is closely related to both ad account budget and ad set volume, which was the first topic covered in this post. If your ad account budget is $20 per day, there are limited things you can do with that. And the advertiser has their hands tied.

But the advertiser can also take a reasonable budget and misuse it by creating a high volume of ad sets, which dilute the budget and make it difficult for any of the ad sets to perform optimally.

Make note of what the ad account budget and ad set budgets are. These have important roles if performance has been an issue.

Run Time

How long does the client typically allow ad sets to run? Do they continue for months at a time or do they have a short life-span?

It’s possible that this is driven by the client’s industry or promotional calendar. Not all businesses have the ability to run a single campaign for months at a time because what they need to promote is constantly changing.

But this also impacts performance. It’s much more difficult to get stable and predictable results from short-run advertising. The algorithm often needs time to learn and your results can improve with time.

There’s also a flip side to this. Some advertisers can become too attached to advertising that worked in the past. Are they still using the same ad sets that ran a year or two ago, even though the results aren’t there?

Even if ad sets run for months at a time, how often are they edited? This, too, can impact performance. Constant tweaks may restart the learning phase, and the ads may struggle to ever get going.

Make note of these behaviors as they will impact results.

Exclusions

Advertisers are rather limited on what can be excluded these days. Meta removed the ability to exclude detailed targeting (interests and behaviors), which is a good thing in this era of algorithmic targeting.

Exclusions now are limited to custom audiences. And there are certainly times when this would make sense. You might exclude the people who already bought the product you are promoting, particularly if it can only be purchased once.

Now that Meta has eliminated the Existing Customer Budget Cap, you might also use exclusions to control how much is spent on your existing customers by creating two ad sets: One for remarketing and one for prospecting.

But advertisers can also overuse exclusions. I’ve seen examples where advertisers are so focused on only attracting new customers that they’ll exclude virtually everyone: Their email list, prior customers, and even website visitors.

While unintentional budget allocation spent on remarketing can inflate your results if you aren’t careful, advertisers also shouldn’t overreact by completely neglecting those who are most likely to act. There’s a balance.

Part of this is about making sure you continue to reach your warm audience. Additionally, you want to avoid exclusions when they aren’t necessary as this is often central to segmenting audiences and fragmentation.

So some exclusions are good, but we want a healthy balance. Document how the client is using exclusions since an underuse may inflate results while an overuse may hurt them.

Targeting Decisions

The recommended approach to targeting has changed significantly during the past few years. Many advertisers have failed to adjust.

It’s also a moving target. Some targeting strategies are no longer possible now. Certain restrictions are only possible with some performance goals. Advertisers often have no idea the impact, or lack of impact, their inputs have on ad delivery.

This isn’t to say that there’s a universal “right” and “wrong” way to approach targeting in all situations. That’s not the case with targeting, and it’s not for much of what we’ve discussed in this post. Results will be your guiding light. If the results aren’t there, targeting could be one of the reasons why.

When evaluating targeting decisions, the focus should be on the following…

1. Demographic Restrictions.

When using Advantage+ Audience, advertisers can set a true age minimum. Otherwise, age minimum, maximum, and gender are suggestions. The algorithm will go beyond those suggestions if it means getting you better results, as defined by your performance goal.

There are ways to fully restrict age and gender, and there are times when this is necessary, particularly when the algorithm can be fooled by low-quality results. But otherwise, these restrictions can hurt results, particularly when optimizing for purchases.

2. Detailed Targeting and Lookalike Audience Restrictions.

If you’re optimizing for conversions, any detailed targeting and lookalike audience inputs are essentially used as suggestions. It’s not good or bad to use them. They may have little impact.

There are times when you can restrict targeting to detailed targeting and lookalike audiences, assuming the performance goal allows for it. In fact, it may even be preferred in those cases.

3. Algorithmic, Broad, and Expanded Targeting.

Does the client rely heavily on algorithmic targeting? This could mean going broad or giving limited audience suggestions. It could also mean using older forms of targeting that are automatically expanded (whether they know it or not).

A heavy reliance on algorithmic targeting is generally recommended for an e-commerce focus. Such an approach can lead to low-quality and misleading results when optimizing for other actions.

4. Remarketing.

Remarketing happens naturally when using algorithmic targeting, which is proven by using Audience Segments with sales campaigns. Because of this, I don’t recommend using general remarketing strategies like we did years ago to isolate all of our website visitors, email subscribers, and people who engage with our posts. Restricting the algorithm this way is usually counterproductive, and it’s not scalable.

The targeting approaches an advertiser takes will tell a story. Are they applying the same strategies they may have used five or 10 years ago? Does their approach change, depending on the optimization?

Whereas decisions made in other areas of the ad set have the danger of hurting results, targeting decisions can lead to another unintended consequence: Results that aren’t what they seem. Advertisers can fall in love with inflated results, rather than using strategies that are more scalable.

Make note of the various strategies this client is taking. Some may have minimal impact, whether they know it or not. Others could be detrimental.

Placements

Back in the day, advertisers were quick to manually remove placements to optimize their results. The thought was that by removing the bad performers, they could make ads more profitable.

While I wouldn’t say you should never remove placements now, the reasons you would aren’t about improving performance. You’d do it because of misleading results that could confuse the algorithm.

If you’re optimizing for any type of conversion, you should use Advantage+ Placements. This allows the algorithm to dynamically show your ads in different placements to get you the best possible results. This is making the most of your budget to balance variation in costs and conversion rates from different placements. You’ll generally see that the placements that don’t lead to the result you want will be used less often.

The problem is when the action you want happens cheaply and in high volume in a placement, but the algorithm doesn’t realize it’s low-quality. This can happen when optimizing for link clicks, landing page views, or ThruPlay views. The algorithm will leverage certain placements to get you what appear to be great results, until you dig deeper.

So, this really comes down to a couple of scenarios that you should monitor…

1. Optimizing for Conversions.

When optimizing for any type of conversion, Advantage+ Placements should be used. If the client has been manually removing placements in these situations, they are likely driving up costs.

2. Optimizing for Non-Conversions.

When optimizing for non-conversions, it is often a good idea to remove certain problematic placements that lead to low-quality results. If you optimize for link clicks or landing page views, the primary placement to remove is Audience Network. When optimizing for ThruPlay views, you should remove Audience Network Rewarded Video. But you shouldn’t universally remove these for other types of optimization.

Does your client manually remove placements? Make note of when they do or don’t. Removing placements can hurt performance, but not removing them might inflate certain results.

Your Turn

We’re getting closer to the end of this series on Meta ads audits. There is so much that can be evaluated, it’s important not to leave out any of the critical elements.

Anything that you’d like to add? Let me know in the comments below!