Just like in sports, the only thing that matters is winning. In advertising, the ROI is the holy grail, and the only thing that matters. Since advertising platforms started pixelating websites and tracking every event that happens on and off their platforms, ROAS became the most important metric in every advertising dashboard and the most important goal for advertisers. Things like cost per click, cost per like, or cost per sale became irrelevant or secondary because what difference does it make how much a click or sale costs you, all that matters is your total profit. The thing is, ROI and ROAS are two different metrics. ROI measures your total profits, ROAS measures the total revenue the advertising platform “generated”. ROAS (return on advertising spend) measures the “return” on every dollar spent on ads. ROI measures your total profits after deducting all associated costs.
I am not writing this post to explain the differences between ROI and ROAS as there are endless amounts of blogs that did that before me. In this blog post I would like to discuss how advertising platforms are using the ROAS metric to attribute (claim) conversions that they might not be directly or even partly responsible for. If you are still not following, let me take a second to explain. Most advertisers (including me), rely heavily on the information reported by the advertising platform (Facebook, Google, Snapchat, etc.) to better understand how our ads are performing and how much revenue they generated. Advertising platforms use the metric ROAS to report how much revenue you generated based on how much you spent. They calculate it based on an X ratio based on your ad spend. If you spent $1 and generated $2 in revenue, then that’s a 2X ROAS, it’s that simple. For the most part, if you are using one advertising platform and you know for a fact that all of your sales are coming from that traffic source, then you don’t have to worry much. However, if you have a lot of existing sales or you are using more than one advertising platform to run ads (Facebook and Google, for example), then you might have an ROAS overlapping issue which means both platforms are trying to claim the credit (attribution) for the same revenue.
Think of it as having a clothing store with 2 sales associates. When a customer purchases a product, he is asked who helped him complete his order. He mentions two different sales associates and they both want to get a commission for the sale. Who gets the credit since they both claim they drove the purchase? The same thing happens with advertising platforms. Advertising platforms understand the importance of showing advertisers positive results and will do everything in their power to make advertisers think that their ads are successful. Although this might be news to some advertisers, larger advertisers are well aware of this and run lift tests to measure the exact value of each platform, it’s not them that I worry about. It’s the small businesses who in many cases, don’t understand how these metrics are calculated and attributed.
It used to be more simple. Back in the early days of PPC ads (Adwords, Yahoo Ads, etc.) you would create advertising campaigns based on specific keywords and would bid and manage your campaigns based on the cost of a click and would calculate the revenue on your own without relying on advertising platforms to tell you how much money your campaigns generated. When pixelated advertising was introduced by Facebook, the rules of the games changed. Advertising platforms quickly realized that advertisers will achieve better results and spend more money if they start optimizing their ads towards whatever brings them the most sales, instead of the lowest cost per click or visitor. The landscape of ads changed and ad platforms like Facebook and Google started reporting directly on the advertising dashboards how much revenue is being generated for every dollar spent on ads. Aside from changing the way we advertise, advertisers started relying heavily on what advertising platforms reported for ROAS as their main performance indicator. The main issue here is the conflict of interest. If the advertising platform’s main objective is for you to spend as much money as possible, then is it wise to rely on their reporting on how good the campaigns are performing? I mean, can you think of another scenario where you are paying a company to promote your business but you are also relying on their reports to rate their own work and performance? Who in their right mind has thought that taking their word for how “revenue” they generated for us is a good idea? It’s like asking a chef in the restaurant to report how good his dishes were tonight. It’s like asking businesses to rate their own customer service satisfaction without checking other sources. The ones who disagree with me will claim that the advertising platforms back their assertions up with all the proof and data and nothing is made up or maliciously reported, which I completely agree with. The issue is not whether or not these sales or conversions happened, it’s the question of who should get credit for them. What if the ads are being shown to people who already know your brand or people who already have an intention to purchase, will that still be considered as a return on ad spend, since the platform didn’t actually generate the purchase?
Think of it from the point of advertising platforms like Facebook Or Google. They build an advertising platform machine with the primary goal of making advertisers successful so they can spend as much money as possible on the platform. They were not made to support or help businesses be as successful as possible, they were built to make themselves successful. They build extremely complex algorithms with a simple goal, help advertisers achieve their objectives for the lowest cost possible. This means that algorithms don’t care if it’s an existing customer, a website visitor from last week, or an Instagram follower. Think of a game that your main goal is to get as many players from one side of the field to the other. You have a limited time and a limited amount of moves. Your best bet would be to find the players who are closest to the other side to save time and resources right? Algorithms use the same logic. They evaluate users based on a large pool of audience. They examine which users are most likely to respond and meet the advertiser’s objective, and if these users happen to already have the intentions to make a purchase or a conversion then so be it. It doesn’t matter. What matters is getting results, even if it means getting results that would happen regardless. Now advertising platforms are aware of the critics and that’s why they give you tools to exclude existing customers and website visitors, but most advertisers are not aware. They think that if they don’t include their customers or website visitors in a campaign, and just use broad or prospecting targeting, that will resolve it, but it doesn’t. Advertising platforms commonly offer advertisers the option to choose between remarketing and prospecting audiences to ensure advertisers that they are bringing new users and are not double-dipping in the same bowl, but even that doesn’t mean that you will only be targeting new users. How do you define a prospecting audience anyway? What if it’s someone who has visited your website a month ago or is an active shopper who just doesn’t follow you on social media, is he considered remarketing or prospecting? What if prospecting the majority of users indeed have never heard about your business by some have, and the ones who have are the ones who are converting and delivering the results. From a logical perspective, a user who knows a brand or business is far more likely to make a conversion than someone who is unfamiliar, so it would take fewer ads served to generate a purchase for someone who knows your business. This is why algorithms will shift some delivery towards users who are likely to have some intent.
Everything being pixelized and advertising platforms tracking every step makes this whole process easier. They use statistical sampling to predict what’s the timeframe a user is likely to make a meaningful action from the first time he responded or clicked on an ad (for example, how long does it take a user to make a purchase from the first time he has interacted with an ad), then they can estimate when to show him another ad before that action happens and take credit for that conversion or sale. For example, if I know that an ordinary user on my website completes a purchase in an average of 24 hours after visiting my website, then an algorithm can show an ad to him within that time frame and claim that conversion. Even if the algorithm is mostly wrong (which is not likely), it’s a numbers game. If out of 100 ad impressions, it was able to generate a conversion, then that’s still much better than targeting cold traffic as cold traffic tends to need much more impressions to generate a sale. If this is confusing to some, then just think of the example of a person handing out promotional flyers at the entrance of a store and whoever ends up making a purchase, he claims credit for the sale. Even if only one out of every 100 people ends up purchasing something, then he will still have much better performance than if he would give out the flyers somewhere on the street.
It all started with view-through conversions.
Attribution and accurate reporting weren’t as big of an issue as it is now, thanks to the fact that advertising platforms count views through attributions in their calculation for conversions and ROAS. If you are not familiar yet, view-through conversions happen when a user sees your ads, and then makes a conversion (purchase/sign up or whatever your goal is). While the most common practice used to attribute a conversion after a click, view-through conversions have emerged as a new way for advertising platforms to say, “Hey, we are taking credit for this sale.” The issue is with view-through conversions as I explained above, it’s really easy for advertising to predict when users are likely to make a conversion which makes it easy for them to show users ads right before they purchase. You can argue when a user sees two different ads on Google and Facebook and clicks on them, which platform should get credit, but when one platform only shows an ad to a user without him clicking on it, I think it’s easier to downgrade its value or contribution. This might not sound like an issue, but imagine if you are selling a product that normally takes a few days from the point a user sees the product until they make the purchase. Think of how many ads they might see on Google, Instagram, Facebook, or even Tiktok, and how all of these platforms will try to take credit for the conversion and claim to be responsible for driving it.
Assisted vs originated the conversion.
All of this lengthy introduction brings me to the most important point, there should be a difference between platforms that originated the conversion or assisted the conversion and that information should be displayed separately. If seeing an ad first on Google is what made me purchase, and a Facebook ad reminded or enforced my actions to purchase, then they should be credited differently. Like in basketball, one player gets credit for the points he scored and another player gets credit for the assist. The same idea should be applied here. It’s not a return on ad spend if that platform generated the purchase, it’s an assisted return on ad spend, which has a different value. I mean, advertising platforms can easily identify which users never visited your website or profile after seeing an ad, but who are we kidding, their mission is to make more money, so why would they ever do that.
The challenges of advertising on two platforms.
If you have had success using paid advertising for your business and feel like it’s time to expand to another advertising platform, then it’s important to be aware of double attribution in your ads reporting. Double attribution is when both platforms take credit for the same conversions and revenue. You can easily identify this by combining the ROAS both platforms are claiming and seeing if it makes any sense or if it exceeds your total revenue. In many cases, it does. Large advertisers put little trust in these reports and run their own test lifts in which they do statistical a/b test with a test and control group and carefully identify the true ROAS or added value of using the secondary advertising platform. As a small business, I recommend being as careful and possible and consider running smaller tests that at least will provide some level of assurance as to the true impact of adding a second platform and what the true ROAS is. Although it’s not a valid test by the books, you can still try to run some tests that are separated by gender/location or age groups, etc.
ROAS is an important metric to follow in paid advertising. But if you use more than one advertising platform, then beware of double attribution. Each platform may try to take credit for the same conversion which may skew your ROAS results. Try combining the ROAS results from each advertisement platform. If the combined revenues exceed your total revenue, then you know that both platforms are attempting to take credit for the same conversions.
Keep in mind that advertising platforms have completely different goals than your business’ goals. Platforms will attempt to take credit for conversions regardless of their actual role in influencing the customer. This is an attempt to make their algorithm appear more successful, to ultimately influence you into spending more money on their platform.