Cracking the Code: The Truth Behind Ad Attribution

Who would've guessed that the most dry and technical aspects of advertising are the most crucial? I'm being a bit sarcastic, of course. There isn't a magical button that prints money for you. And if you don't understand how ad platforms are billing you or claiming credit for results, they're going to be far less effective. This is especially true if you're running ads across multiple platforms or if you have a stream of organic revenue that isn't coming from ads. By definition, ad attribution is how ad platforms assign credit for the results "driven by the ads." They use this data not just for reporting to advertisers but also to fine-tune future ad delivery. It's called attribution, not actual results, because it's about pinpointing the interactions that lead to desired outcomes, like a sale or a conversion... Notice how the terminology can be interpreted differently than intended.

But it's simpler than they make it sound. Ad attribution happens when an ad platform notes a conversion or sale and links that back to an ad. This means the ad "drove" the interaction. However, this doesn't necessarily mean the user purchased because of the ad; it merely indicates the user interacted with the ad before converting or purchasing, and that's why it's reported to you. If you're advertising on more than one platform and a user interacts with ads on multiple platforms, the conversion could be counted twice, with each platform reporting the same conversion. This is why it's crucial to distinguish between what's driven by an ad and what's directly caused by it. And here lies the core idea of the issue, what many advertisers miss. At a glance, they take whatever the ad platform reports at face value, based on surface-level metrics, without digging deeper. While it's possible that all sales could be driven by ads, and all reported conversions are a result of these ads, for most businesses, the opposite is often true, especially for those advertising on multiple platforms.

Before we dive deeper, think about what motivates you to launch campaigns, increase budgets, and spend more on a platform. Which metric makes you scale your ad spend? Conversions, ROAS (Return On Ad Spend), and revenue reported by the platform are designed to convince you that your investment is paying off. That's the rationale behind the creation of attribution models—they're meant to make you feel like you're maximizing your ad spend. Still confused? Let's simplify these concepts and definitions to better understand and discuss how you can set up your ads in a way that ensures your attribution settings align with your actual business goals.

It all began with impressions, likes, and clicks. Initially, attribution wasn't in the picture. It only came into play when advertising platforms started to track results that occurred off their platform, on other websites and apps. They aimed to capture the real goals of advertisers by monitoring what happened after users landed on a website and completed an action, like making a purchase or signing up. The reasoning was straightforward: advertisers aren't just interested in clicks or likes; they're after sales and conversions. Therefore, it was logical to measure the actions users took on their websites after clicking on an ad and use this data to refine ad targeting, aiming to reach more users likely to take similar actions. This shift in strategy proved to be a game-changer, enhancing ad targeting and effectiveness by focusing on whether users were likely to fulfill the advertiser's main goal (a conversion or sale), rather than just clicking on an ad. However, this also gave ad platforms an opening to claim credit for conversions and sales they hadn't directly generated, encouraging advertisers to increase their spending.

Despite the platform, most advertising companies bill based on ad space usage (CPM - cost per 1000 ad impressions), meaning they charge you for simply displaying your ads to users, regardless of the users' actions. Even PPC (Pay-Per-Click) platforms like Google, which offer click-based pricing, ultimately base their charges on the market rate for impressions. Thus, ad impressions incur charges regardless, but the tangible actions users take with your ads—clicks, views, engagements—are measurable and clear outcomes. If you spend $1 on ads and your ad gets two clicks to your website, those clicks aren't part of any attribution model. The ad platform reports the direct engagement users had with your ad, including the number of impressions, how many times the ad was watched or clicked, likes, etc. However, attributing the activities on your website after a user clicks on an ad to get there, or in some cases, just views the ad and later visits your site, becomes particularly complex.

Meta has consistently been at the forefront of ad attribution, with many platforms adopting its methodologies. Let's examine the attribution models Meta provides and their default settings.

Meta (Facebook & Instagram)

Tiktok Ads: 

Google Ads:

Snapchat Ads: 

If you're feeling a bit lost, let's simplify it. The controversy around ad attribution has led to a push for platforms to shorten their attribution windows. This change aims to create a tighter link between when a user interacts with an ad and when they make a purchase, addressing concerns that previous methods were unfair and unreliable. Ad platforms are now limiting credit for conversions that occur well after the initial ad interaction, a shift driven by the need to preserve their credibility. Advertisers are increasingly savvy about these models, and overly broad attribution windows were undermining confidence and diminishing the effectiveness of ad algorithms by attributing unrelated outcomes to ads.

Meta traditionally allowed for attribution up to 28 days after a click or even just an ad view, taking credit for conversions within this window. Google, even more ambitiously, tracks conversions up to 90 days after a click and 30 days after a video or ad view. This practice raises questions about the validity of such extended attribution periods.

Consider this scenario:

You run a website selling running shoes and advertise on Meta, TikTok, & Google. A potential customer, browsing on Nike or Adidas, encounters your Facebook ad. Whether or not they click, they later search for your brand on Google and click on a sponsored ad, but don't make a purchase immediately. Days later, they see a TikTok retargeting ad, influenced by their online behavior and your site's pixel data, but again, they don't click. Eventually, they decide to buy and return to your site to complete the purchase. The next day, all three platforms claim a conversion, reporting three conversions for a single purchase. This illustrates that a user doesn't need to click on an ad for it to be counted as a conversion, with conversions potentially attributed days or weeks after the initial ad interaction.

Attribution is designed to maximize ad spending by suggesting ads are highly effective. Understanding its impact on marketing budgets and ad efficiency is crucial, as inaccurate attribution can lead platforms to misallocate future ad impressions. Moreover, platforms take credit for "view-through" conversions, which don't require a user interaction with the ad. This complex issue merits discussion. You can read more about the issue of “view-through conversions” here.  While advertising platforms may not always align with advertisers' interests, they have made significant changes in recent years, narrowing attribution windows from 28 days for both clicks and views to now typically allowing up to 7 days for clicks and 1 day for views. Always verify the attribution settings when creating a campaign. Advertising platforms tend to like to hide the attribution information so by design, advertisers will be more likely to select a larger attribution model that includes 7 days of clicks and at least 1 day of view.

Here is how it looks on Meta’s ads manager:

Here is how it looks on Tiktok:

While on most platforms you select the attribution model for each campaign, on Google you have to edit the settings of the conversion on the conversions page and it will apply to all campaigns automatically. 

It must be meaningful conversions. If you go based on what the advertising platforms set by default, you open yourself to the world of view-through conversions. This is especially damaging because when a conversion is tied to a click, a user has to click on it for it to count. With view-through conversions, ad platforms can detect which users are close to the finish line in terms of making a purchase and throw a few ad impressions right before it happens and take credit for it, which could potentially make you spend more money on that ad platform. To avoid this, make sure that moving forward, every conversion that will be reported will only be a click-through conversion that requires a click to count. This will disable the meaningless reporting of conversions that had nothing to do with the ad platform and will ensure the algorithm is directed towards delivering ads to users who are likely to convert after a click and not just position ads that will result in conversions that came after a view or an ad impression.

A single attribution model. Can you compare apples and oranges? If you are using a different attribution model for each ad platform, how do you know how much each conversion is worth? Should you have the same targets if, for example, on Google, it attributes click conversions for up to 30 days and on Facebook it only tracks based on 7 days? If that’s the case, how do you define how much each of these conversions is worth to you?

And why would you define them differently? The people who work at Google will claim that the cycle between discovery to purchase is longer than other platforms and it takes users a longer time from the first click until they actually purchase than it would take them on other platforms like Meta. To me, I haven’t seen any factual data to support this and conceptually I don’t see how that would make any sense. The same users are on both platforms and to me, it just seems like Google is better at attributing in longer time frames and they are using it as an argument for better ad performance. But regardless of whether or not they are right, if the ad platform is effective at driving actual conversions, the majority of the conversions will happen at a closer time period to when they clicked on the ad. If it doesn’t, it means the ad has done little to impact the user and you should probably reconsider anyway. Additionally, even if you “lose some of the data” by shortening the conversion windows, you are still able, to a greater extent, measure the conversions and performance against each other as each platform will measure conversions in the same time window. Measuring the conversions in the same way will allow you to have a little more confidence in their value, thus it will allow you to scale and increase the budgets appropriately across all platforms in a more unified way.

Measure the time difference for each conversion. Selecting the same attribution model for all platforms is the first step, but it shouldn’t be the only step. You do want to investigate to better understand how long typically most of the conversions take from the time that they last interacted or clicked on the ad. For example, it could be that users who click on an ad on Instagram mostly convert within one day and users who come from TikTok might convert mostly after 2-3 days. Knowing this information is crucial to better understand the relationship between the ad platform and the conversions and how much of a direct impact it has. It’s hard to say what’s more valuable as some advertisers can argue that platforms that drive conversions right after an ad click are positioned along the funnel right before a user is likely to checkout. For example: When you find a shoe that you want to purchase through a Facebook ad and the next day you want to return to the website through a Google search. If the user interacted with both an ad on Facebook and an ad on Google search, you are likely to expect that Google search will always have the shortest time lapse between a click and a conversion as Google ads are positioned much closer to when users are ready to purchase compared to social media platforms where they are generally targeting users in discovery mode. Thus, when it comes to Google shopping or Google search, you would expect a very short time from click to purchase. That being said, having this data will allow you to compare across similar ad platforms like social media ad platforms, for example.

Let’s say that you are measuring the conversion time between Instagram, Facebook, TikTok, & Snapchat. You realize that the conversion time between Instagram, Facebook & TikTok is similar and that most people convert within the first day of clicking on an ad. Then, you look at Snapchat ads and you realize that the majority of conversions came after 4-5 days. One assumption that you can make is that Snapchat played a smaller role in that conversion as the conversion only happened days after seeing or clicking on an ad compared to having a more direct connection after clicking on an ad through the other platforms. If you compare all social media platforms to each other, you can assume that the furthest apart a conversion happens after a click, the less impact that specific click had on the user and it’s likely to be associated with a different factor or a different ad from a different platform. This will also allow you to prioritize your budgets more effectively across ad platforms and will allow you to theorize and make educated optimizations on which platforms are actually driving the conversions and which ones are “stealing them”.

How to measure and compare?

On Meta, you can show the breakdown based on different attribution windows:

You can break down the results based on the different time periods and can see what percentage of conversions come from each model.

TikTok offers attribution analytics to give you a deep dive into all your attribution models.

Google also offers a rich attribution breakdown report:

As you can see, you can see the number of conversions broken down based on the number of days.

On Snapchat - it’s intentionally more complex to make it a little bit harder to compare.

You do have to manually check the total number of conversions for each attribution model separately and calculate the numbers and percentages and make a summary.

Set up your attribution strategy for success.

If you made it through this entire article you are already on the way to making a big difference for your business as most people don’t get beyond the first 500 words, so great for you. Now that you understand the concept of measuring and setting your attribution windows properly ask yourself what it means for your ads and your business. It’s important to note that there isn’t one model that works for all, but there are common models that I generally recommend sticking to. Regardless of what the advertising platform might recommend, I find it most effective to stick to one attribution model across all ad platforms. Whether it’s Google search or social media ads, if you compare metrics like cost per purchase or “ROAS,” then it must be on the same attribution windows.

Ad platforms’ main objective is for you to spend more money, not generate the most profit. That’s why they are notorious for mastering the game of attribution, and they have extremely sophisticated systems to predict when a conversion might happen and they will know exactly when to position an ad before that conversion happens, so not always the platform that has the closest time between click and conversion is the best platform, there are obviously other factors involved like is this ad platform attributing more warm traffic or repeat customers than others and where in the purchase process do users more likely interact with it. My recommendation is to take the largest spend ad platform that you have that you have a higher confidence rate knowing that it’s the main driver of your revenue and conversions and use that as a model for all other smaller spending platforms to follow.

For example: if you are spending $1000 per day on Meta and most conversions happen within one day. You would expect that if you spend $100 on TikTok to also expect conversions within 1 day after a click and so on. There is no perfect formula here that you can use but the point here is to start tracking how the different ad platforms are attributing results of your ad and to start measuring and comparing them. Understanding the data will allow you to craft an attribution strategy that brings you the best return on your marketing spend across all channels and helps you optimize towards higher efficiency levels. It could also help you remove or reduce spend on less effective advertising channels that you might have thought were effective.

Attribution is a complex game by design and there isn’t a third-party tool that I could verify and trust that is better than just working directly through the ad platforms and creating the most effective strategy that is based on data you can cross-examine and action based upon.

In summary, mastering ad attribution is essential, not just for understanding the complexities of advertising but for ensuring your marketing efforts align closely with your business objectives. The journey through ad attribution—from understanding "view-through" conversions to navigating varying attribution windows—demands a proactive and analytical approach from advertisers. By embracing a consistent attribution model, carefully analyzing conversion timings, and focusing on significant engagements, companies can transform their ad spend from a mere expense to a strategic investment. This strategy not only enhances the efficiency of advertising dollars but also promotes a more transparent and accountable advertising landscape. Ultimately, the essence of ad attribution lies in leveraging these insights to refine marketing strategies, turning complex data into a distinct competitive advantage, and ensuring the highest return in terms of actual sales on your website, rather than relying solely on the inflated ROAS figures often reported by ad platforms.