The Facebook ROAS Death Spiral: Fix it Now — or Lose Your Shopify Store

tl;dr

  • Facebook Ads reporting on ROAS is broken.
  • You’re likely pulling back on Facebook ads as a result, and hurting your business even more.
  • Switching to a different metric based on the two cold-hard facts of your business (money leaving your bank and money going into your bank) can get you back on track and scaling profitably again.

Introduction

Let’s think of Facebook ROAS (Return on Ad Spend) like gambling.

Here’s the situation. You walk into a bar alone (the Priest and the Rabbi don’t join you this time). You turn on the Vikings vs. Packers game and put down $1,000 on 2:1 odds for the Vikings to win.

A miracle happens, and the Vikes actually win. SKOAL!

The bookie pays you $5,000 — a 5:1 return and says, “There you go. $2,000 for a 2:1 return”.

You’re a little confused, but you’ve got big plans for an epic tailgating party, so you don’t say anything. 

Next week, you wear your lucky, unwashed Dalvin Cook jersey and put down $10,000 for the Vikings to win again at a 2:1. 

You swear you see Roma Downey carry Adam Thielen “like an angel” into the endzone for a game-winning touchdown. 

The bookie pays you $50,000 — a 5:1 return again, and says, “There you go. $20,000 for a 2:1 return.”

Aside from the fact that you’re likely going to get mugged on your way home, you’re excited because you realize that your bookie says that he’s paying you 2:1, but he’s actually giving you 5:1

If the bookie continues to give you 5:1, do you care that he says he’s giving you a 2:1? Aren’t you more focused on the actual exchange of money?

What if I told you that I’m almost certain you’re doing the exact opposite with your Shopify store right now?

Instead of going to the bookie that says he’s paying you 2:1 but actually gives you 5:1 in cold hard cash, you’re going to the bookie that says he’s paying you 4:1, but he’s actually only giving you 1:2 in cash — you’re losing money on every deal.

Shopify store owners have a huge problem.

And it’s such a big problem that I’m willing to publish some radical ideas about advertising that I swore we would keep hidden away, tucked into our pants like a midwestern Dad’s mowing shirt. Why? Because our advertising agency, Elumynt, couldn’t possibly help everyone out. 

Hey, Facebook ROAS, what’sa matta with you?

We’ve been auditing (and taking on) new account after new account that’s “infected” with the same “disease” — The Facebook Ads ROAS Death Spiral.

You used to get a much better ROAS on Facebook ads, but something just isn’t “working” lately. 

You’re blaming the agency: 

“Let’s go, guys; you need to figure out how to get us back to the same ROAS we had last year. Have you tested this audience? What about this one? What if you target people that are having a birthday this month — we need to keep testing a bunch of random things that I heard from a guru on his pre-recorded webinar from 2014.”

Your agency is blaming the iOS 14 or “the creative”:

“We need you to send us more creative — more, more, more, more. That’s not good enough; it has to be better.”

And all the while, it’s causing both of you, client and agency, to make a series of truly scary decisions putting both of your businesses in jeopardy.

The client’s random test ideas from the “List of Bad Facebook Ad Tactics That Ignore How the Algorithm Works” creates unnecessary time spent and budget wasted on tactics that won’t drive incremental revenue, reducing the time and budget spent on things that the data actually supports doing (although this data isn’t being tracked or understood by most agencies right now — that’s what we’ll get to shortly). 

Your agency is spinning their wheels on the wrong things instead of scaling stuff that’s already working — because they’re unaware of the real problem.

And these bad decisions are leading to even lower performance from Facebook ads, which leads to the final bad decision in this cascade of misfortune: businesses are scaling back their ads week after week, month after month, because they’re afraid they are wasting money and they want to stop the bleeding as much as possible until someone can figure out what the heck is going on. 

And that decision — an understandable decision, but one that’s based in fear, uncertainty, and a lack of understanding about attribution —  is causing many businesses to dig a hole that’s even further from being profitable than if they had left things alone. 

Mayday! My Facebook ads are in a ROAS death spiral!

Toni Wall Jaudin at The Atlantic tells about the harrowing tales of the death spiral, “In the early decades of flight, aviators were bedeviled by bad weather. Those who encountered poor visibility mid-flight told harrowing tales of disorientation and confusion. Surrounded on all sides by milk-white fog or hazy darkness, pilots entered a world where nothing behaved as it should. When they observed the plane slipping into a gentle descent, they corrected to gain altitude, only to find the plane diving downward faster. Or, when they were certain the plane was flying level, the turn indicator would register a turn to the right. What the instrument registered as level, meanwhile, felt like a turn to the left.

Under these conditions, bailing out often became the best option. Those who didn’t often joined their plane as it crashed into the ground.”

What good is a gauge if you can’t trust what it’s reporting? 

I see a lot of people doing the same thing with their Facebook ads right now — they’re attempting to use their gauges, but they don’t understand how those gauges have changed, so they’re “overcorrecting,” and it’s putting their business in jeopardy. 

I’m sure you’re well aware of iOS14.5 because a lot of people like to talk about it as if they have a clue what it actually means. 

A lot fewer of those people were talking about ITP 2.3 when that came out last year. 

The truth is, I’m convinced that very few people truly understand what happened and how it’s affecting their gauges (metrics) — especially ROAS.

I think what makes this so hard for many people to understand is its name — ROAS (Return on Ad Spend). The problem is that ROAS is not actually measuring the return on your ad spend — it’s a misnomer. 

True ROAS would be calculated as “the revenue that you collected from the advertising you did” divided by “the ad spend itself.” That’s what people think the metric called “ROAS” in Facebook ads is telling them, but it’s not. The metric that’s mislabeled in Facebook ads as “ROAS” is calculated as the REPORTED revenue that you collected from the advertising you did divided by the ad spend itself.

ROAS on Facebook ads is not actually ROAS

That’s a big difference. 

Remember our bookie? He was REPORTING a 2:1 but paying out a 5:1 — but you keep going to the bookie that’s telling you he’ll give you a 4:1 (sounds better than 2:1), but he’s only paying you a 1:2.

By not understanding how the “reported ROAS” in Facebook ads has truly changed, you’re on a dangerous path towards killing your business — you’re on the Facebook Ads ROAS Death Spiral.

I’m sorry, officer… my ROAS gauge must be broken

Let’s think of ROAS as a gauge. It helps you understand something visually. 

For a moment, I want to move away from the ROAS “gauge” and talk about another gauge that you probably understand a little better — the speedometer in your car.

In very simplified terms, your speedometer measures the speed of your car by getting 1:1 real-time feedback from the rotation of your wheels (if you’re a car nut, I know it’s more complicated, but I’m trying to make this really easy to understand).

How a car speedometer works

(Source: https://standexelectronics.com/applications-markets/speedometer/)

What if the software for the speedometer of your car was changed so that it could only count a few of the rotations, but it wasn’t sure what percentage of the rotations it was counting (10%, 30%, 50%), and it wasn’t getting that feedback right away — it was getting that feedback 3 minutes after you reached that speed?

So now, you’re on your way to work, you accelerate the car out of the driveway, but it still says zero. By the time you get to the stop sign, the speedometer now registers 5 mph even though you’re stopped. Then you hop on the highway at what feels like 60 mph (while the speedometer is reading 7 mph) and 3 minutes later the speedometer eventually reaches its “full speed,” but it only says you’re going 15mph. 

How would that change your driving? 

If you don’t adjust your understanding of the new reported speed, you might think you need to speed up.

You might even try it. 

You might be driving at 120 mph, but your speedometer would tell you that you were only going about 30 mph. No matter how hard you tried, you could never again get your car to register (report) a speed of 60 mph, potentially leaving you to think your car is broken and “just not performing as well as it used to.”

Facebook ads ROAS gauge is broken

But what else could you do? What if you got smart and realized that you could still look outside and notice that things seem to be passing you by “visually” at the same speed they used to. You could notice that you still “feel” like you’re going 60 mph. You could keep that speed up — 60 mph but registering as 15 mph in your car — and drive past a police officer, to have them measure it, in a 25 mph school zone as the radar reads your speed at 60 mph and you get a rearview mirror full of pretty flashing lights. 

Of course, the car manufacturer could do you a favor by simply renaming the metric that your speedometer was using so that you didn’t think it was actually measuring the speed of your car anymore. It’s registering a number that sort of helps you understand your speed relative to itself, but not the actual speed. 

So 15 mph in your car would still be relatively faster than 10 mph in your car, but you don’t know if it’s 5 mph faster, or 10 mph faster, or only 2 mph faster… but you do know that it’s faster. So the relativity of it is still a helpful metric to track; you just can’t think of it as the actual speed of your car anymore.

That’s what Facebook Ads ROAS is like post iOS14.5 (and, well… other conversion metrics, too). It’s still measuring SOME conversions — just not all of them. And it’s not getting that data back in real-time. 

Facebook ads ROAS GIF

After iOS14.5 it’s only measuring a few of the conversions (but it’s not sure what percentage that is). That doesn’t mean that it’s not causing the same amount of conversions as it was before, just like pressing the gas pedal in your car is still moving the car 60 mph even if the speedometer is now only showing 15 mph. 

The cause and effect are still the same; you’re just getting a very different number reported to you, and you have to adjust your thinking or you’ll risk your life trying to get your car to register 60 mph again “like it used to” the same way that you’ll risk your business by trying to get your Facebook Ads to register the same ROAS that you got last year.

A 2x “ROAS” in Facebook this year (or pROAS) might be the same thing as a 5x ROAS last year — 2 might equal 5 — at least in terms of how much actual revenue they are driving for your business. 

What people see in Facebook for ROAS

“OMG — my Facebook ROAS isn’t doing as well as it did last year so let’s make a bunch of changes that will really screw up our account!”

What people would see in their bank account for advertising on Facebook

But since people are so used to looking at their speedometer (ROAS in Facebook), they’re missing the whole point — these two scenarios are identical for their business. They spent $20,000 on ads, and they brought in an incremental $100,000 into their bank account. 

The fact that Facebook isn’t reporting all of the revenue shouldn’t matter.

ROAS vs MER on Facebook Ads

Is that frustrating? Yes.

Is the relative platform ROAS (pROAS) helpful to decide if things are doing better or worse? Yes, but… only if you compare the pROAS this week with last week, or other more recent periods where it’s measuring the same thing. But comparing the pROAS from this week with last year is the same as the speedometer issue above — this year, a reported car speed of 15 mph is the same car speed as last year’s 60 mph.

Directional change of ROAS and MER on Facebook Ads

So, the best thing to do in this situation is to recognize that ROAS in Facebook is a misnomer. It’s not actually measuring the return on your ad spend. It’s measuring some return on your ad spend, and if that new number goes up, you’re very likely doing better, and if it goes down, you’re very likely doing worse (incrementality aside) — so you can still use it to figure out if what you’re doing is improving or not.

At Elumynt, we’ve been calling this number pROAS (platform Return on Ad Spend) as a way to make sure we recognize that it’s not measuring the actual effectiveness of the ad spend anymore, but it’s still helpful for determining what’s working better relatively.

The truth, the whole truth, and nothing but the truth

OK, you get it now — at least sorta. So how can you figure out what the correct pROAS for your business is? How do you know if a 2x pROAS is the same as a pre-iOS 14.5 4x, 8x, or 3x?

Well, let’s go back to our speedometer analogy. How do you know if 15 mph is the same as 60 mph, or the same as 80 mph, or the same as 40 mph? We could take our eyes off the gauge (platform metrics) and look at our surroundings. We can also use measurements that are outside of our car (i.e., the police radar). 

Square root of ROAS meme

For measuring the effectiveness of our ads, we can take our eyes off of pROAS and come back to the core of what we are trying to do. In a car, we push the pedal, and the car moves forward (assuming you don’t have it in reverse). In advertising, we should be able to spend money and see more sales converted.

In advertising, there are two undeniable facts:

  1. How much money you spend. You can literally look at your bank account or credit card and see how much money you spent on ads. That’s perfect 1:1 data, still.
  2. How much revenue you collected. Again, you can look at the transactional revenue in Shopify and see how much money you collected (also still a perfect 1:1 data point). 

Only two facts in advertising - MER

Is your goal to increase how much money Facebook told you it drove to your business (a reported metric that isn’t based on 1:1 data), or is your goal to increase how much money actually went into your bank account?

MER (pronounced em-ee-ar)

That brings us to a metric that is outside of the ads platform (and therefore not plagued by any issues with reporting, attribution, loss of data, etc.) 

The “outside” metric that compares those two facts is called MER — Media Efficiency Rate.

Although, the industry hasn’t done a great job of agreeing on a name for that metric. It can also be called Total ROAS, Biz ROAS, Blended ROAS, Enterprise ROAS, or some other similar variation. Alternatively, some people prefer to use tACoS (total Advertising Cost of Sales), which is essentially just the inverse ratio of those same two undeniable facts — Total ad spend / Total revenue.

I personally don’t love calling it anything with ROAS because it’s not necessarily measuring the return on your ad spend. “Return” makes it feel like it was wholly dependent on the advertising. You could maybe argue the same thing about Media Efficiency Rate, but I like that it’s removed from the ambiguity of “ROAS”.

The Formula for MER is: MER = Total revenue / Total ad spend

how to calculate MER

When I calculate this number, I prefer to use Shopify gross sales minus Refunds. I’ll share more about why that is later, but it has to do with Returns and keeping a more consistent metric.

The nice thing about this number is that it eliminates attribution — it’s only looking at two facts and can’t be manipulated by reporting windows, spending more on “retargeting” or other less incremental tactics, etc.

So using the MER approach, you could simply increase your spend on Facebook and see what happens. Do you remember proofs in math class? This is about as basic of an “if-then” statement as you can get with advertising.

Hypothesis: IF I increase my spend on Facebook ads, THEN my revenue will increase. 

Then test it. 

Did your revenue go up? How did that affect your MER?

If you spent $1,000 more and you collected $5,000 more revenue, then you had an incremental MER of 5x ($5,000/$1,000) from your Facebook ads.

If your pROAS (the ROAS reported in the Facebook ads platform) is a 2x, then you could assume that a 2x in platform actually produces a 5x in real money going into your bank account.

Reported ROAS vs. incremental MER

Or, to put it another way, if you spend $1,000 on Facebook ads, Facebook will drive a very real $5,000 into your bank account but only tell you that it drove $2,000. 

Now, one time could be a fluke. So it’s a good idea to repeat the test. Go ahead and increase your spending again. And then again. And then again. Do you keep getting the same results? Still not convinced, go ahead and reduce your spend and see if revenue comes back down. 

Note: It’s worth calling out that you might not always get the exact same results. Some weeks might be better than others, so you may get an MER that ranges from those incremental dollars — especially if you’re continuing to scale up. 

There are several things that aren’t perfect with this test (for instance, if you have a product that typically takes people longer to decide if they want to buy it or not — a considered purchase), but it’s the easiest way to test this out without getting too bogged down with complicated metrics. And if you get consistent results from this test week after week after week, you can start to assume it’s more than just a correlation — it’s likely causation.

Note: All Facebook Ad Accounts aren’t created equal. If you scale a poorly set up Facebook Ad account, it’s simply accelerating your inefficiencies. You still need to spend on the right audience with the right ads.

Bill Gates Quote about scaling inefficiencies

MER’s Kryptonite

Obviously, the downside to MER is that it doesn’t tell you which source drove the revenue. Did that sale come from Facebook ads, or did it come from Snapchat, Google, Organic search, Direct, Email, influencers — the list goes on. Here’s where we enter another stage of frustration for many people.

If your business has a couple of channels that drive significant revenue, the method above likely isn’t specific enough for you. We’ve got some clients that are spending nearly $1M/month on influencers, and while that’s a great channel, the revenue from it isn’t nearly as predictable or stable, so it can really muddy up the data of a generic test as described above. The same thing is true for another client of ours that’s spending over $500k per week on infomercials. Heck, even if you have a lot of ad spend in another ad platform like Google, Snapchat, or TikTok, it can make it hard to do a simple scale test on a single channel if you’re only using MER to measure against.

So what can you do in those situations?

Go back to being scientists. There really is more to Facebook Ads than just “more creative.”

more facebook ads creative meme

 

Too many people have gone away from that in advertising lately. “Gurus” are out there chasing metrics that they don’t understand. It’s as if we’ve given up on being strategists and allowed the platforms to think on our behalf.

You have to remember that data isn’t inherently 3-dimensional. It’s a 2-dimensional image of a 3-d world, much like an X-ray.

3-dimensional data (Facebook ROAS plus Google Analytics plus MER plus…)

I broke a lot of bones growing up, mostly because I loved taking risks to do something that others couldn’t do — for no other reason than to prove that it could be done.

Callout: I still love taking risks now, but I take much more strategic risks 🙂 

One time I was practicing my pogo stick (go ahead, I can hear you chortling over me being such a nerd), and I was bored of doing it no-handed, 360’s, etc., so I decided to try doing a backflip… off a loading dock. 

It went about as well as you could imagine, and I broke my wrist. When the ER took X-rays, they didn’t just take one image. They took a few. I don’t remember exactly how many, but I’m sure they took at least an AP (anteroposterior, or simply – front-to-back) and a lateral (side) view. 

AP and lateral xray view of wrist

Why?

Because they can’t see the full break from looking at just a single directional view. But if they combine that with a different view in another direction, they can visualize a more 3-dimensional understanding of the break and how to treat it. 

The same is true when looking at data. The platform data is one view. MER is another view. Google Analytics provides yet another view. All of these views are correct, just like each of the single views of the X-ray were correct — they’re simply incomplete. You have to combine all of these views together to correctly understand your business. Anytime you focus on ONE view as your “source of truth,” you enter a 1-dimensional view of your business, and your business will suffer from incomplete data leading to incorrect recommendations for changes to make with your advertising.

Independently, these three images make absolutely no sense whatsoever. But when you combine them all together, like combining Google Analytics, with MER, with Facebook Ads, etc., the message becomes a lot clearer. 

3-d data for digital ads attribution

Getting another view of your advertising by looking at channel data in Google Analytics

Direct Sales

Question: “Hey William, when I look at Shopify or Google Analytics, it says that the increase in revenue didn’t come from Facebook, it came from Direct.”

Answer: Ok, but let’s understand what “direct” means.

Google Analytics states that, “A session is processed as direct traffic when no information about the referral source is available, or when the referring source or search term has been configured to be ignored.

But to make that a hint clearer, MonsterInsights clarifies that, “Google Analytics defines direct traffic as website visits that arrived on your site either by typing your website URL into a browser or through browser bookmarks. In addition, if Google Analytics can’t recognize the traffic source of a visit, it will also be categorized as Direct in your Analytics report.

First, let’s consult our intuition. We were moving along without a big increase in revenue from people just happening to type in our website, then we increase our spend on Facebook, and we got more revenue from “direct.” Then we did that again, and again, and again, getting the same results (more revenue from direct when we scale our Facebook ads) — doesn’t it make sense that people aren’t likely to just type in your website in higher volumes unless they were prompted to do so… like from seeing your ads?

Second, it’s important to understand how the data is being reported in Google Analytics (and Shopify) for traffic source. Remember, these are not necessarily reporting undeniable facts about which traffic source was responsible. They are giving credit to the channel that was the last touch.

Someone clicks on your ad, lands on your website, leaves the tab open, but doesn’t buy (session 1). This session gets credited to Facebook. 

Then they come back to that tab later (session 2) — the platform is going to call that session “direct.”

Now, in the old days, before ITP 2.3 (yes, this started before you heard about iOS 14.5), Google Analytics would still give the credit for that to the last paid source, not direct. “Source precedence—A direct-traffic visit that follows a paid-referred visit will never override an existing paid campaign. Whatever is the latest paid campaign visit is listed as the referral for the visit.”

While that’s technically still true today, Google Analytics is also plagued with similar “attribution” issues that Facebook is experiencing as a result of Apple’s changes to tracking. That means that a significantly higher amount of revenue that came from a paid channel is now being counted in Google Analytics (and Shopify) as “Direct.” You might say that Google Analytics and Shopify Analytics, like Facebook Ads, suffer from a bit of “amnesia” and can’t “remember” where that traffic originally came from even 24 hours earlier for many people. 

In summary, you didn’t magically have more people coming directly to your website, there was a cause, and that cause was the increase in your Facebook ads (or other top-of-funnel advertising platforms like YouTube, Snapchat, Tiktok, etc.)

Organic Sales

A similar trend happens when you look at “organic” traffic. Organic traffic is when someone goes to a search engine, searches for something, and then clicks on a search engine result that takes them to your website.

The next question you should ask yourself — what search terms were sending more of that traffic to my site this week?

Google Analytics is terrible at exposing that, but you can still get a ton of valuable search term information from Google Search Console. When we do SEO audits, on most Shopify websites the client is always surprised to find that often 70 – 90% of their “organic traffic” is actually just variations of their brand name. 

That’s not always the case. There are some stores that do an amazing job getting truly organic search traffic, but I can’t think of the last website that we audited that was even getting 50% of its organic traffic from something that wasn’t a branded search term.

And these aren’t small sites. These are Shopify Plus websites that have been around for years doing a few million dollars in sales every month — often with closer to 80% of their organic traffic coming from their own brand terms.

Again, let’s consult our intuition. If we put money into a machine (Facebook Ads), and that machine reaches our ideal audience (awareness), leads to an increase in traffic to our website, and we also see an increase in branded search traffic that’s registering as “organic” — what’s the likelihood that people are searching for our brand terms more for no reason vs. because they saw our ads? And if you see that week after week after week during a scaling test, it can begin to eliminate doubt that people were just “all of a sudden” flocking to your store to buy stuff for no good reason.

Facebook spend and organic traffic revenue

Paid Search & Shopping

You probably get where I’m going with this, but what are the odds that those increases in branded search terms are only resulting in an increase in organic branded search and not causing an increase in clicks on your branded search ads? 

When you scale up your top-of-funnel tactics on channels like Facebook Ads, TikTok Ads, Pinterest Ads, Snapchat Ads, YouTube Ads, etc., you will see an increase in clicks on your paid branded search ads as well — and you will see an increase in clicks (and efficiency) for your Smart Shopping campaigns in Google Ads, too. 

Wait — why?

Because Smart Shopping, by default, also uses your branded search terms.

Facebook spend and paid search revenue

Oh, and from a measurement point of view, Google Ads are a lot less affected by the iOS14.5 stuff. They’re still affected, but not as much. When someone clicks your search ad, that means they’re already looking for this product — they’re further down the funnel and more likely to buy soon. That means that a higher percentage of purchases will happen within the window that Apple will allow the platforms to track.

Callout: This is the point where people will spend on Facebook ads, not get the attribution that they want to see in Facebook because of iOS14.5 (and ITP 2.3, etc.), but they will see an increase in efficiency and “demand” in Google Ads that was created by the Facebook Ads, and instead of spending more on Facebook to keep creating that increased demand, they will reduce their ad spend on Facebook and increase their ad spend on Google. This obviously doesn’t impact the bank account and MER the way they would have hoped, so then they scale back here, and the vicious cycle of improperly diagnosed attribution from 1-dimensional data continues to cause pain to many companies.

Email (and SMS)

This one is my favorite. 

Question: I don’t think it’s from Facebook ads; I think our email is just a lot more effective. Look, my email platform is taking credit for ALL of this, and Google Analytics is giving it a lot of credit as well — so that’s two platforms saying email (and SMS) are “creating” a lot of revenue for us, right?

Answer: Email and SMS are amazing channels. Seriously, keep doing a great job here. You probably are actually still leaving a lot of money on the table and need someone to help you scale that, too. 

But for the sake of attribution, we need to look at a few things. 

First, why is your email list so effective? Would it be reasonable to assume it has something to do with the quality of your email list? It’s made up of a bunch of people that have already qualified themselves as an ideal customer of yours by literally purchasing from you before.

Email list for ecommerce store from mostly purchasers

What happens to that amazing list if you use too many tactics to drive email signups from people that have not qualified themselves (i.e., free giveaway) causing the % of qualified users to drop? It can hurt your email deliverability, click-through-rate, conversion rate, etc. By all means, build that email list, but manage it thoughtfully and strategically. 

But that’s another article for another day.

When someone sees your ad (or even clicks on it), do you think it makes them more likely to open your email?

Hint: it does.

When someone sees your ad, do you think it makes them more likely to click the CTA in your email? Do you think it makes them more likely to convert on your website after clicking your CTA? 

Again, the answer is yes, and yes. 

Advertising helps improve your email effectiveness (and email improves your advertising effectiveness — your customers don’t live in a single channel).

But here’s the kicker — if they both play a role in that sale, which platform do you think is going to get the credit for the purchase, Facebook Ads or Email? 

last click attribution in google analytics meme

In platform data: Email is often going to lead to a purchase almost immediately, compared with Facebook Ads, which can be delayed a bit. Facebook might not take credit for that purchase in its own platform (because of everything we talked about above), but email will almost certainly be able to take credit for it (because a significantly high percentage of those sales will happen within the window that Apple will allow them to track data for). 

Google Analytics and Shopify data: Which action do you think is likely the last click, the Facebook Ad or Email? Overwhelmingly that’s email. Let’s just assume that if both Facebook Ads and Email play a role in 100 purchases, maybe 70% of those are going to click last on the email vs. maybe only 30% that clicked the email first and then later clicked a Facebook Ad.

I’m generalizing here, as every store is going to be different, but the trend will hold true for most stores. That’s not a scientific number, just an intuitive guess directionally. 

That means that Google Analytics and Shopify are both going to give credit to email (last click) and not to Facebook Ads.

What’s even more, is that Facebook created more sales than just the ones that “clicked,” as evidenced by the increase in Direct, Organic Search, and Paid Search/Shopping.

Do you see how understanding a more 3-dimensional view of the data will allow you to have a better understanding of what’s actually happening?

“It’s not all about 1-channel, it’s about an ecosystem, and right now brands are getting too focused on just one thing not working and being reactive rather than taking a step back to see the whole picture.” — Grace Pietsch, Sr. Ecommerce Growth Strategist at Elumynt

Last Click, which one should it pick?

OK, so clearly this last-click stuff can really get in the way of understanding your data if you aren’t careful. 

Last-click is very important, but let’s pull back for a minute. 

I like to view last-click as “harvesting”.

Imagine you run a small tomato garden, and you hire 4 employees. One employee plants the tomato seeds for the garden. All 4 employees water the garden. And then only 3 employees harvest the vegetables — the 3 that don’t plant the seeds.

last click attribution tomato garden

Employee One plants 10 tomato plants. A bird brought a seed over from another garden and planted one more tomato plant. And 2 more tomato plants grow up on their own because a few tomatoes fell last year, and those seeds were dormant in the soil. 

So now you have 13 tomato plants, and everyone waters (nurtures, ya know… like a nurturing campaign) them. 

A short while later, you have a lot of wonderful tomatoes that are ready to be picked, and the remaining three employees go out and harvest the tomatoes. 

When you return from vacation, you ask all four of your employees to write down how many tomatoes they each collected. 

Employee 1 – 0

Employee 2 – 50

Employee 3 – 65

Employee 4 – 35

TOTAL: 150 tomatoes

Ads attribution by last click

Based on this data, you incorrectly deduce that paying for employee 1 is “costing” you too much money, so you cut their hours to part-time for next season. 

Next season comes around, and this time Employee 1 only has enough time (resources…hint: ad spend) to plant 5 tomato plants. The bird brings another seed over and plants another one again. And 2 tomato plants grow up on their own again from seeds that fell last year and were dormant in the soil. 

This season when everyone hands you their tally of tomatoes, it looks like this:

Employee 1 – 0

Employee 2 – 30 (down from 50 the previous year)

Employee 3 – 30 (down from 65 the previous year)

Employee 4 – 20 (down from 35 the previous year)

TOTAL: 80 tomatoes, down from 150 tomatoes the previous year.

ads attribution by last click the following year

Those last-touch harvests can’t happen if those tomato plants aren’t planted in the first place. The same is true for advertising, and ignoring that data by only looking at the “last-touch” attribution will cause you to make decisions that are detrimental to the growth and profitability of your business.

[Case Study] Facebook Ads come back to life

Scaling back Facebook Ad Spend was hurting the business

When iOS 14.5 hit, a few clients struggled to grasp what it meant for pROAS, MER, and how to optimize. As a result, those clients were spending less and less on Facebook, which naturally correlated with less total revenue actually going into their bank account. 

Facebook ad spend vs. Shopify revenue

Pulling back spend was understandable in May/June because the effects of iOS14.5 were fresh, and no one really had a firm grasp on exactly what it meant. The ROAS in platform (pROAS) was dropping from a lack of attribution, so pulling back ad spend was reasonable until we could convince them of optimizing around MER instead. 

At Elumynt, we’ve been tracking MER for a long time and optimizing around it, but it wasn’t something we were pushing aggressively with our clients to adopt. That said, when iOS14.5 came out, we weren’t blindsided, and for most of our clients, we were able to keep right along with ad spend and scaling without missing a beat. 

But for a few clients, this change in reporting/optimization goals was tough to understand. Haley Nixon, our Senior Paid Social Specialist, put together an amazing article where we attempted to bridge the gap for those that were still stuck on “in-platform ROAS” by helping them look at aROAS (Adjusted ROAS). I won’t go too deep into that, but if you want to read more about it, you can check out her article on our blog, “Facebook is Lying to You About Revenue: Implications of iOS14.5 and ROAS”. 

For this client, lowering the ad spend did increase the efficiency (based on MER), but when you factor in overhead, it meant that this client was making less profit each month they pulled back. That’s important to remember — higher efficiency of ad spend at reduced ad spends often results in less profit, not more. 

Scaling up Facebook ad spend was helping the business

For several clients, we were given the green light to go ahead and scale as a test against MER (ignoring Facebook ROAS for the time being). This chart is an example of what that looked like for one of those clients — specifically a client that had scaled back a ton on Facebook, and that is in their off-season, so that it can’t be said that this was from seasonality (more on that later).

The correlation between increasing Facebook spend each week with increasing revenue week after week after week made it easy to move from correlated to likely causative.

How did this impact the MER for this client?

This client likes to be at about an MER of 4. You can see that Week 1, Facebook was barely spending anything, and their MER was a 6. Maybe that sounds like a good thing, but if you’re not spending enough money, it doesn’t matter if you have a 20x MER; at a certain level, you’re just not bringing in enough sales to cover your overhead, which was the problem this client had. 

Their MER was “great,” but they had scaled back on FB ads so much because of the iOS14.5 issues that they weren’t actually making any profit — in fact, they were losing money at an MER of 6

Week 2, we begin to increase spend on Facebook ads (first week of the scaling test), and MER naturally takes a nosedive that first week (as we predicted it would) — we’ve brought in more people to the site, but they weren’t ready to convert just yet.

Week 3 rolls around with more spend on Facebook, and the MER improved. Then it happened again on week 4, and then on week 5, the client decided to pull back on Facebook ads just a bit. If you look at the revenue chart above, revenue was mostly flat, but the drop in spend helped the MER go up more that week. But remember, that’s like turning off your oven and being amazed that it’s still warm. If you dial back your FB ad spend (heat source), you’ll still harvest the sales from that the following week (for most brands).

So then, in Week 6-8, we scale aggressively because the client wants to aim for a pretty far away top-line revenue goal to see if we can hit it. It did drop the MER a bit each week, but again, remember the oven simile, getting aggressive paid off, as the ad spend plateaued, the MER continued to climb (along with revenue), and the client hit their top-line revenue goal for the month at a 4.7 MER (above their MER goal). 

Oh, and in case you’re wondering if this is somehow connected to seasonality — this client sells something that you ONLY typically buy in November/December, so scaling like this in their off-season, with an MER above goal, paints a very clear picture of cause and effect from scaling up their Facebook Ads.

We have case study after case study of this exact trend happening for most of our other clients — at least the ones that aren’t still trying to get their speedometer to register 60 mph in a 15 mph world. Or, you know… trying to get Facebook Ads to report the same thing it did before iOS14.5.

These are clients that are willing to look beyond the incomplete data that Facebook is telling them and look outside the platform — at their actual bank account — to decide if they are growing or not. 

Time of death is… wait, nevermind, it’s alive again!

I do want to be very direct here, that if you don’t change the way you understand the reporting of ROAS in-platform on Facebook, Snapchat, TikTok, YouTube, etc., that you will very likely stall your growth or worse, go backwards. You will NEVER get the same reported ROAS that you were getting before — that should only be your goal if you have no interest in the facts of the money you are spending and the money going into your bank account.

For everyone else that wants their business to succeed, let’s do a quick recap:

  1. You’re not getting the same ROAS in Facebook that you got last year (or even earlier this year), so you’ve been pulling back on ad spend, and your business is suffering.
  2. Now you understand that what Facebook is reporting as ROAS isn’t actually the return on your ad spend, it’s not a fact — the gauge is broken. 
  3. So instead, you’re going to use MER (a fact) to evaluate how effective your ads are and run a scaling test on Facebook to see how it affects your business. 
  4. You understand that the scaling test needs to be done correctly (correct ad account setup) and that it will have a lot of the attribution go to other channels like Direct, Organic, Paid Search, etc., because of how attribution works in Shopify and Google Analytics.

If your store has been suffering lately, you’re not alone. 

The steps I’ve outlined above are imperative to revive your business — but they’re also just the beginning. There are several ways that we build upon MER to really help your business grow profitably, but this article is long enough already, so that will have to be in a follow-up article. 

Now, if you’ve “already tried” scaling FB and it didn’t work, it’s very likely that the account isn’t set up correctly.

I’d be curious to know a few other things:

  1. Did you measure it correctly (i.e., profit, not pROAS)?
  2. Did your team set up the scale test correctly (i.e., there’s a huge difference between scaling something good and scaling something bad)?
  3. It could truly just be seasonality for you (or something else entirely, like a new competitor that entered your market and is dominating, etc.)

There are many ways to figure out what the issue is if the scale test didn’t help you increase actual dollars going into your bank account. If you’re stuck, we’re happy to help.

I also want to make sure you understand that these ideas aren’t only my own. Our entire team has worked on this — iterating, testing, validating, diagnosing — these ideas are the work of so many brilliant people. I’m just gathering them together in a single document for your sake so that you can regain control of your business objectives. 

 

Image of the author
William Harris
As the founder and CEO of the rapid-growth ecommerce advertising agency, Elumynt, William Harris has been critical to driving growth for multiple startups in ecommerce and SaaS with 6 recent exits, the largest selling for almost $800M. His contributions have been featured on Forbes, Entrepreneur, Fast Company, Shopify, The Next Web, and many more.