Optimizing Ad Performance: The Ultimate Guide

February 15, 2023 -
By Scott Konopasek

This guide has 3 sections:

  • The 3 Levels of Optimization (where we’ll introduce a new framework for approaching optimizations)
  • Performance Improvement and The Cost of Doing Nothing (where we cover the formula for performance improvement and measurement)
  • A Simple Process for Optimizing Ad Spend (where we break down how to approach making changes)

Let’s Dive in!

The 3 Levels of Optimization

You won’t be optimizing everything on a campaign all the time. There are actually three levels of optimizations you can make to a campaign and each level differs based on how frequently you’ll make each type of optimization.

Level 1: In-Platform

In-Platform optimizations are the most common changes advertisers make to campaigns because they’re the most obvious and the types of improvements you make most frequently. Level 1 optimizations are typically done multiple times a week inside of your ad platforms. These optimizations are things like adjusting audience targeting, testing new creatives, adjusting bids.

The data you need to inform Level 1 adjustments will be inside of the ad platforms you’re using, not 3rd party tools like Google Analytics. You need the level of insight that can help you make granular adjustments.

When you talk to a marketer about optimization, they’ll normally only think about these In-Platform optimizations. They might also consider Level 3 optimizations, but Level 2 almost never comes to mind.

Level 2: Delivery Parameters

Optimizing Delivery Parameters is the least common type of change that marketers deploy. This is normally because the data necessary to drive these insights is unavailable or difficult to find. Level 2 optimizations are done cross-platform, so quality cross-platform measurement is a necessity.

Delivery parameters will normally be optimized every 2-4 weeks. To generate these insights, you might ask questions like:

  • How often are users seeing our ads on multiple channels?
  • How often are media mixes driving conversions versus a single channel by itself?
  • Which specific media mixes have the highest conversion means?

When you’re working at this level, you’re trying to understand how your campaign is delivering across channels and how you can adjust delivery parameters to be more efficient? You’ll keep an eye out for things like time of day, day of week, which channels are reaching or converting a lot of incremental users.

You might find a channel that’s reaching a lot of new people, but it’s struggling to convert them. That’s when you can pair that channel with another to increase conversion rates.

Level 2 boils down to trying to adjust how your campaign is delivering in the short term or medium term to drive better longer term results. These optimizations will look like small budget changes across channels to reduce waste and improve efficiency or scale.

Level 3: Big-Picture Media Mix

Regularly doing Level 1 and Level 2 analysis gives you the data you need to execute Level 3. These Big-Picture Media Mix Optimizations are the least frequent, done 2-4 times per year.

You start this analysis by stepping back to look across the quarter and asking:

  • What consistently has done well this quarter?
  • What consistently struggled?
  • What drove overall success?
  • Where are you less efficient?

This isn’t exactly the same as a postmortem on your campaign, but you are taking a similar high-level view. If you do this well, you’ll produce insights like, “video played a really key role in the media mix. It really struggled to convert incremental new users, but does a great job of supporting conversion rates. Based on this we allocated a little bit too much money to video based on what we need to drive optimal cross-channel delivery.”

When you decide to reduce overall video investment for the quarter by 10% and reallocate that to remarketing and native, the data that supports these changes comes from the past 3 months of analysis you’ve been doing at Levels 1 and 2.

These types of optimization might sound similar to Level 2 and the difference is that you’re looking at trends that become clear only after several months that aren’t noticeable from month to month. When you’re analyzing every week or month, performance can be impacted by factors outside your control – things like the economy or the weather, or anything else that might change how people buy. When you analyze 3-6 months’ worth of data, you normalize for these variants and can get a better understanding of each tactic’s consistency.

When you’re analyzing you’re big-picture media mix, you should also evaluate your user journey. You need to understand like how long does it take to sell your product. How many ads, over what time period drives the best conversion rates and highest quality conversions?

Level 3 is all about taking a longer perspective, reviewing the user journey, understanding what’s consistently working and what’s consistently delivering results. Then based on the insights you find here, you’ll be making bigger budget changes and allocations.

The Key to Unlocking Uncommon Performance

Most marketers are already operating at Level 1 and Level 3 – these are very common. The key to unlocking uncommon results is executing Level 2 optimizations to improve your delivery parameters.

You can only optimize Facebook or your DSP so much. Once you’ve got the right copy or the right creative, there’s only so much more you can do to improve. If you operate in-platform effectively, you might improve conversion rates to an extent, but the next level of results requires you to go cross-platform.

The way to unlock the next level of performance is to start adjusting delivery parameters to, for example, build cross-channel frequency that can get you beyond what’s possible in platforms alone.

In our analysis, we see two meta trends around the opportunity to improve delivery parameters: first, most users only see a single impression; and second, conversion rates are almost always higher at a 2+ frequency. The 2+ frequency users normally will convert at a higher rate if those impressions happen in multiple channels as well.

This means that advertisers have a huge opportunity to improve conversion rates by increasing frequency and building cross-channel frequency.

When we analyze campaigns, we see that normally, less than 10% of conversions saw ads in more than one channel. It’s not uncommon for us to see conversion rates increase 3-10x when these Level 2 optimizations are made.

Make Sure You Spend Time Working At Each Level

Each of the three levels of data analysis are important for consistent performance improvement. Every level is a different way of looking at your data and a way to find opportunities for optimization. 

  • Level 1 is where you’re working in platform every week to make improvements. 
  • Level 2 is where you adjust delivery parameters each month to unlock higher conversion rates
  • Level 3 makes larger optimizations to your overall media mix and budget allocations.

Spending time operating at each level is crucial if you want to get the most out of your campaigns.

Performance Improvement and The Cost of Doing Nothing

In every marketing campaign, there’s room for improvement. A marketing campaign is never perfect, but the more you can improve upon it, the better you’ll be able to reach your goals. You need to consistently optimize your marketing efforts for maximum impact.

Unfortunately, too many marketing teams do nothing. This inaction normally falls into 2 main categories: doing nothing to improve data quality and not optimizing frequently. These two variables are the heart of performance improvement.

The Performance Improvement Formula

In the long run, performance improvement comes down to a simple formula. The level of improvement you’re able to achieve is a simple function of your data quality that informs optimizations, and the frequency at which optimizations are shipped.

Data Quality x Optimization Frequency = Performance Improvement

Data Quality

Data quality is the first piece teams need to put in place. Let’s imagine two scenarios: one where you have no marketing analytics and one where you have an omniscient understanding of attribution.

With zero marketing data, you would have no idea what was working and what wasn’t in your campaigns. If forced to make optimizations, you’d be just about the same as my drunk uncle trying to hit a piñata blindfolded. Any changes you made to the campaign would have a 50/50 chance of making things better or worse.

With an omniscient view of attribution, you would be able to understand how every customer journey goes and how valuable each touchpoint on that journey is. When optimizing, you would know exactly what needed to be done and you would be perfectly confident making changes to the campaign.

All this is to say that data quality matters. With zero marketing data or poor-quality data, you can’t optimize. With perfect data, you would know exactly what optimizations to make to maximize response.


Once data quality is solved for, the next piece to figure out is optimization frequency. If your data quality allows you to make an average improvement of 2% with every optimization, the frequency at which you optimize now becomes the driving force behind how much you can improve performance. Advertising optimizations don’t add together, they compound.

Remember: Performance Improvement = Data Quality x Optimization Frequency.

Now we’ll dive into how you can improve data quality and optimization frequency to improve performance as quickly as possible.


The way you measure ad performance will determine the quality of your analytics data for improvement. Without proper measurement, you’ll find yourself with very little data making low-confidence optimizations to campaigns. While perfect measurement isn’t possible, getting the best possible data is worthwhile.

Prioritizing proper measurement is not as common as you might think. Roughly two thirds of marketers don’t have an attribution or measurement tool in place to get quality data. Historically, this has been because measurement tech was so expensive. Lucky for us, there are now a number of more affordable options. Other barriers to the adoption of measurement tech are low-trust in attribution methods, low technical sophistication to fully use these tools, and reliance on free tools like Google Analytics.

Levels of Measurement

Here are 5 different levels of measurement that you should be familiar with in your pursuit of quality marketing analytics:

Free Reporting is the lowest-quality level of campaign data. Free reporting will come from sources like ad platforms of Google Analytics. These tools are normally designed to earn more of your ad budget, not tell the absolute truth. If the reporting is free, your ad budget is the product.

Adservers are the next level up of measurement technology and give more cross-channel information than free reporting. With an adserver, you should expect to make better optimizations than with free reporting.

Media Mix Models do more heavy lifting than adservers to give cross-channel insights. While these reports can be very general, there are plenty of startups working to deliver the statistical estimations of MMM’s faster and using less data. With MMM, you get more insight into how channels work together that unlocks new potential optimizations.

Multi-Touch Attribution is akin to media mix models in that MTA estimates how channels work together to produce results. Similarly, MTA will let you make better optimizations than free reporting or an ad server because you’re getting more granular performance data.

Incrementality testing will uncover a different set of insights that MMM/MTA miss out on: how channels work alone to produce results. While Incrementality can’t give terribly specific data, it allows you to see the value of ad channels in a way that free reporting, ad servers, or attribution tech simply can’t provide.

Bi-Modal Attribution is the newest way to get high-quality analytics. BMA lets you combine the unique insights from MMM/MTA and Incrementality to fully understand how ad spend turns into results. With BMA, you can make the best optimizations because it provides the most granular and highest-quality data.

Tactical Examples

Let’s assume that you are going to make an optimization to your campaign every month. Here’s what performance improvement would look like with 3 different levels of measurement:

Measurement LevelMonthly Improvement1 Year Total Improvement
Google Analytics2.5%34%
Ad Server3.5%51%
Bi-Modal Attribution5%80%

As you can see, over the course of a year, having higher quality data for optimization makes a big difference.

Optimization Frequency

The quality of data informing optimizations combines with optimization frequency to determine your overall performance improvement. Even if you’re stuck with low-quality data, you can make significant improvements by making positive changes every 2-3 weeks.

There are too many teams that have a ‘set it and forget it’ mindset when running campaigns. Trying to optimize quarterly or not at all is a great way to miss out on a lot of potential performance. Even if your data is low-quality, you can still make small changes to experiment with different ideas and find the ones that work best for your business. The key is to be consistent with your optimizations so that you don’t waste time waiting for months before trying something new

Marginal Gains → Exponential Improvements

The key to consistently improving performance over time is focusing on marginal improvements.

You’re probably not going to look at your campaign and optimize 25% of it all at once. It wouldn’t be realistic to reallocate 25% of your spend and then immediately get a 25% performance improvement.

The marginal gains approach is to optimize one part of your campaign at a time by just a small amount. Start with 1% here, and 2-3% improvement there.

If you do that consistently, that 1%-3% gain every 2-4 weeks compounds over time. You improve your results by 2%, and now you’re growing that expanded result by an additional 2% every time you optimize.

Examples of Frequency

If you’re deciding between making an optimization to your campaign every 2 weeks or every month, here’s how performance improvement would differ:

Optimization FrequencyOptimization Amount1 Year Total Improvement
Every 2 Weeks3%103%

In the long run, optimizing more frequently will make a greater impact even if you’re making smaller optimizations more often.

The Cost of Doing Nothing

Performance improvement is a function of how good your data is and how often you’re shipping optimizations.

If you were to truly do nothing over the course of a year, performance would remain at benchmark.

If you use the highest quality data available (BMA) and optimize every 2 weeks, you would more than double your results in a year with 203% of your initial performance.

Bimodal Attribution

Bi-Modal Attribution is the choice data source for media buyers because it gives the highest quality data and you get fresh campaign insights every 2 weeks to optimize performance as frequently as possible.

Wrapping Up

If you’re looking to get the most out of your campaigns, the first step is to get the right measurement technology in place to improve your data. Then, work to optimize frequently using the Marginal Gains Approach (don’t try to be a hero).

Don’t forget:

Performance improvement = Data Quality x Optimization Frequency

A Simple Process for Optimizing Ad Spend

When media planners are running a campaign there’s a classic dilemma: do you optimize for a higher conversion rate or for scale of results? While these may seem to be the same, they have distinct use cases and will impact your campaign in different ways. 

In this article we explore the differences between optimizing for conversion rate or scale, and then walk through a simple process you can use each week to optimize your ad spend.

Efficiency or Scale?

Any optimization will improve your campaign by making your ads more efficient (higher conversion rates) or better scaled (converting more users). In other words, optimizing for conversion rate should help you get more results from each dollar spent on ads across a media mix. 

Optimizing for scale, on the other hand, will help you reach and convert more people with your ad spend. It’s important to understand which goal is most relevant to your campaign objectives and then optimize accordingly.


Optimizing ad spend for efficiency means adjusting your ad spend and delivery to produce higher conversion rates. This means building frequency across creatives, ad sets, and channels. Generally, the more ads a user sees, the more likely they are to convert. There’s certainly diminishing returns, but our data across clients and verticals shows that most users need to receive more impressions.

Optimizing for higher conversion rates is the ideal strategy when you don’t have new budget but still need to drive more conversions. Building frequency is one of the biggest opportunities for advertisers that’s hiding in plain sight.

A great example of efficiency optimization is delivering retargeting ads to high-intent site visitors. Improving the frequency and diversity of these ads will improve conversion rates and capture users most likely to purchase.

Another example would be seeing attribution data that shows when users see ads in Search and Social, they’re 5x more likely to convert than in either channel alone. The opportunity is to build cross-channel frequency by showing a Search ad to the Social audience and vice versa.


Optimizing ad spend for scale means deploying additional budget to channels that reach and convert unique audiences. These channels have high incremental value and work well without support from other channels. 

You should focus on optimizing for scale when you have big growth goals, net new ad budgets each month/quarter, and channels that are converting users at high rates with lower reach.

An example of a scale optimization would be seeing that Google ads convert at a 0.5% and receives 90% of your Search budget, while Bing converts at 3% while only receiving 10% of your Search budget. Optimizing for scale in this case would be adding more spend to Bing until Bing’s conversion rate matches Google’s, or until the platform can’t spend anymore.

3 Steps to Optimizing Spend

There are three simple steps that every advertiser should follow when optimizing ad spend:

  1. Cut Waste
  2. Exploit Efficient Channels
  3. Optimize Scaled Channels

Before jumping into optimizations, you need to understand what you’re trying to accomplish with your optimizations. Get more specific than “increase ROAS”. 

  • What are the strategic goals the business is trying to achieve? 
  • Will ad budgets be increased, decreased, or kept constant over the next months? 
  • What results does leadership care most about from marketing?

With these answers in mind, you’re ready to begin.

Cut Waste

Because there’s wasted ad spend in every campaign, there’s always an opportunity to take the lowest-performing spend and reallocate it to improve returns. The first step of optimizing ad spend is to cut from the bottom.

Cutting waste begins with identifying your worst-performing channels. Which channels are producing the lowest return? Where are ads delivering inefficiently? (e.g. converters have 4 frequency on average, but a channel has 1.1 average frequency)

You need to evaluate each channel on 2 factors: it’s incrementality and how the channel lifts conversion rate in the media mix. Attribution models like last-click/touch don’t tell the whole story, so be careful not to jump to conclusions.

Once you’ve identified the lowest-performing channels, we recommend cutting between 5% and 10% of your overall ad spend so it can be put to better use.

Exploit Efficient Channels

After you’ve identified and cut your lowest performing ad spend, it’s time to reallocate and get more results! The first category of channels that deserve more spend are your efficient but unscaled tactics.

As you compare channel performance, you’ll have some channels or media mixes that have lower than average CPAs, but don’t have a high conversion volume. This is the best place to optimize first.

Once you’ve identified your high-efficiency/low-scale channels, add budget here to see the most results. You can keep doing this until the channels or tactics you picked start to hit diminishing returns and CPAs rise to the average.

You should plan on adding spend to these channels every 2-4 weeks until they no longer have the lowest CPAs.

Optimize Scaled Channels

After exploiting all of your high-efficiency channels, the next strategy is to optimize your scaled channels. This is the next place to deploy your reallocated dollars.

Your ‘Scaled Channels’ will be delivering a high volume of conversion, potentially with low conversion rates. The opportunity with ‘Scaled Channels’ is to increase efficiency by adding frequency either within the channel or across other channels.

On many campaigns we observe, a vast majority of users only see one ad – what a waste! By increasing frequency to users, you can make scaled channels more efficient. This means building a custom audience that you can serve ads to across your media mix. Understanding exactly how each of your ad channels works together to produce results will make optimizing your scaled channels much easier.

Seeing Things Clearly

When you compare channels to understand each’s efficiency and scale, it’s important to see how many conversions in each channel are shared vs incremental. Once you can report on both ways a channel performs, your best and worst channels become clear and you don’t have to do any guesswork.

By seeing each channels’ Bi-Modal contributions, you can tell how often channels work together and what the iCPA is for each. This makes optimizing your channels much easier, and it will help you avoid spending on unprofitable channels. You can also see how often a channel provides incremental revenue that would not have been made without its contribution.

The best way to get this data is with Bi-Modal Attribution.

How to Pick Between Efficiency and Scale

When you have a choice between optimizing for efficiency, or optimizing for scale, it’s important to know what your goals are and what’s happening with your marketing budgets.

We almost always recommend starting by exploiting efficient channels. This will mean adding budget until CPAs increase to your baseline. Starting here is often best because this is where you can get more conversions at the lowest cost.

The best part of pushing new optimizations every month is that performance improvements don’t add up, they compound. This means that frequent, small improvements combine over time to make massive impacts. No matter how big or small your ad budgets, you should spend time every 2-3 weeks cutting from the bottom and reallocating to the top.