Has your Facebook advertising cost increased in the last 3-6 months?
If your answer is “yes”, then you’ll want to read the rest of this mega post.
It will take you 8 to 10 minutes, but by the end, you’re going to get a clearer idea of how to run your Facebook advertising campaigns this year.
How this experiment is going to help you
Below is a live Facebook advertising experiment we’re running right now.
As you scroll down, you’re going to see 3 things:
- The ads we are running.
- The results we get and what we think.
- The ideas we have and how we put them into action.
This gives you an insider’s look into our process without having to pay us $5,000/month and you’ll learn to implement at least some parts of it in your business.
How we will gauge the success of this experiment
For this experiment, we have 3 goals.
Goal 1: Get cost per click (CPC) down to $0.20.
Quick backstory (and a bit of math)…
In Oct and Dec last year, we ran 2 campaigns.
The first campaign was to generate leads for a new $297 course. The CPC was $3.62.
The second campaign was to drive traffic to a few old but popular articles. The CPC was $4.69.
Assuming our conversion rate is 40% (optimistic for cold traffic), we would pay between $9 and $12 for a lead.
At $9-12, we will have to convert at least 3.3% of all the leads to break even.
That’s not even counting the cost of editing the videos, using the email marketing and course softwares, spending my time, and everything else.
So to make all these work worthwhile, we need to bring the CPC down to somewhere between $0.40 and $0.50.
For the purpose of this experiment, we set our sights further and decided on a stretch goal of $0.20, which increases the chances of making this experiment profitable.
Goal 2: Get leads under $3.
In the first campaign, we drove cold traffic to a squeeze page for a week where we gave away a whole year’s worth of AdEspresso.
Nearly 1 in 5 visitors converted and each lead cost us $20, which is unprofitable for us at this point, given that we don’t plan to upsell any of them to a higher-value product or service.
But why was the cost per lead (CPL) so high?
There were 2 reasons.
Reason #1: Sky-high cost per 1,000 impressions (CPMs).
The online marketing niche, as you may know, is one of the most competitive niches on the Facebook advertising platform.
But do you know just how competitive it is?
Take a look at our CPMs below.
Yes, you are not mistaken…our CPMs were $47.69 to $88.90!
This means that Facebook charged us at least $47 just to show our ads 1,000 times (crazy!)
For good measure, AdEspresso published that the average CPM in Q3 2016 was just a mere $7.19.
Granted, we could’ve done a few things to make it cheaper, like:
- Testing the ads to get higher relevance scores. Most of the ads we ran only scored between 3 and 5.
- Testing different ad formats, like videos, which had lower reported CPMs.
- …and if you’ve been running the same ad for some time, refreshing the ad helps too.
But think about it.
If Facebook charges you $47 for every 1,000 impressions, you only have 2 choices:
- Accept the hefty cost and wait till you get 1,000+ impressions (or a big enough sample size) before tweaking the ad, which means you would have burnt at least $47 before you decide if an ad is worth it!
- Risk making a wrong decision by tweaking the ad way before it reaches 1,000+ impressions.
Both choices are far from ideal.
And that was when I realized that there was a bigger, less obvious problem.
Reason #2 (the real problem): We were doing it wrong (I admit, this one’s a bit hard to swallow).
Sure, we may have generated 600K+ leads for an American insurer for less than $2 each just by driving cold traffic to a bunch of squeeze pages.
But that wasn’t our business.
Unlike our client, we didn’t regularly create and publish content in 2016. People didn’t know us or the people behind J Social.
We didn’t have a list of raving fans who were dying to hear from us. We had a handful, but not enough to build a business.
We didn’t have a ready list of 100+ customers which we could use to create a lookalike audience. We had to find the right audiences from scratch.
We didn’t have CPMs that ranged from $3 to $10. If anything, ours were 4-20x as much.
So driving cold traffic to squeeze pages directly didn’t work for us.
It got us leads, but they were too expensive.
So this time, we are doing it differently.
Here’s our new plan:
- We’re going to drive cold traffic to a few old and new articles (including this one).
- For every new article we put out, we’ll include a button to sign up for more updates or offer a content upgrade at the end.
- We’ll test a bunch of different headlines, copies, and images to see if any of them sticks. If anything, we want to test more quickly with smaller sample size size to find what works quickly.
- We’ll choose the article with the most traffic and/or leads and create a free offer around that topic.
- We’ll retarget all our website visitors in the last 30 days to the free offer.
Our goal is to get leads under $3 (stretch goal).
Our revised lead generation plan
To do this, we will need 3 things to happen:
- Each visit to the article to cost between $0.40 and $0.50. Our stretch goal is $0.20.
- Each retargeted visit to the squeeze page to cost between $0.40 and $0.50.
- 1 in 3 visits to the squeeze page to convert.
The combination above will give us a CPL between $2.40 and $3.
Goal 3: Spend up to $2,000.
At the end of this live experiment, our goal is to get 100+ of you to say “yes, it’s possible to make Facebook Ads work for me and my clients in a profitable way” and start running your campaigns.
We know that there are plenty of marketing consultants and small business owners who don’t have a big budget.
And most other articles you have read so far are the complete opposite.
They either have a sizable budget (at least a few thousand dollars a month), a brand name that people know about, or no case study to show.
What about the rest of you who are just starting out or have failed several times and only have $500 a month to spare? What if you are a marketing consultant who has clients just like that?
If that’s how you feel, then we just want to let you know that this article is written for you.
That’s why we’re only spending $10-$20/day in this experiment.
Our goal is to show you that it’s possible to get results even with a small budget.
How we set up the experiment
To make it easy for you to follow what we are doing, we broke out our campaigns into a few sections below.
1. Number of campaigns.
Because we are driving cold traffic to our articles, the first thing we wanted to know is whether the Engagement or Traffic objective will bring cheaper link clicks.
We can never be sure about these things.
For example we got more traffic from running an engagement campaign for a client and less traffic for another client than when we ran a traffic campaign.
So to be sure, we created 2 campaigns – one for engagement and another for traffic.
2. The ads we are running.
Engagement and traffic ads look different.
To make the test as fair as possible, we kept the same image and copy for every piece of content in both campaigns.
We also promoted the same 3-5 articles so that we can be sure the difference isn’t due to a different mix of content.
Here’s an example of the engagement ad we ran.
Here’s an example of the traffic ad we ran for the same piece of content.
3. The people we are targeting.
Last Nov, we did a small launch and throughout the launch, 400+ people opened and clicked on our emails, read our posts, and visited the sales page.
We consider this group of people the most relevant, even though not all of them bought our product.
So we uploaded this list on Facebook and created a lookalike audience.
Since 1% lookalikes in USA generates over 1m+ people, we wanted to segment this group into 2.
After some consideration, we decided to split the group into 2 age groups:
The younger people are usually in their first career. From our email list, these are mostly marketing consultants or young marketing minds in business. We were curious to know if they would respond differently from the older group.
There are potentially many ways we could have segmented the group. The same may be said for you. You should look at your business to decide what makes the most sense.
More importantly, treat this as a temporary segmentation. If the data tells you that there is no meaningful difference, you can try another segmentation later.
4. Where we are running the ads.
Placement wasn’t something we wanted to play around with in this experiment. So we chose the safest placements – Desktop and Mobile Newsfeeds (including Instant Articles) – for both campaigns.
5. How much we are spending on each ad set.
Traffic campaigns require a minimum spend of $5/day/ad set.
On the other hand, engagement campaigns only require a minimum spend of $1/day/ad set.
We decided to spend $5/day/ad set for the traffic campaign and $2/day/ad set for the engagement campaign.
We did this to minimize our initial spend/day while seeing if the engagement campaign had the potential of getting cheaper clicks than the traffic campaign.
If the entire engagement campaign got cheaper clicks at a lower spend, then it has the potential of competing with the traffic campaign.
On hindsight, we should have spent $5/day/ad set just to be consistent and eliminate the possibility of budget as a factor that could have affected the results.
Our results (new updates every week)
This is the first time we’re doing a live experiment, so we are going to try a few ways of presenting our results below. If you have a better idea of how we can present our data than what you see below, we would like to hear from you in the comments section.
Update 1: Dec 30 2016 (Day 4)
We started this experiment on Dec 27. Here are some stats of our experiment so far:
- Total ad spend: $56.90.
- Reach: 2,602.
- Total ad clicks: 48 (i.e.1.74% CTR).
- Total traffic: 16 (i.e. 0.58% link CTR).
- CPM: $20.62.
From the stats above, you can see that the sample size is small. But there were 2 telling differences between the traffic and engagement campaigns:
Difference 1: Engagement campaigns cost us more money
Although we spent 1x more money on the Traffic campaign than the Engagement campaign, we reached 4x more people in the Traffic campaign than the Engagement campaign.
How did this happen?
The answer is in the next row of the table labelled CPM.
Facebook charged us 1.5x more money ($41.27 vs $16.43) for the Engagement campaign than the Traffic campaign. So the Engagement campaign reached fewer people for every dollar spent than the Traffic campaign.
Difference 2: Nobody clicked on the link in the Engagement ads
Next, 3.43% of all the people who Facebook showed the Engagement ad to clicked on the ad. But none went to our website.
On the other hand, even though fewer people clicked on the Traffic ad, half of those people clicked on the link and visited our website, which is the primary goal of this campaign.
Despite the small sample size, the huge difference between the two campaigns was enough for us to pull the plug on the Engagement campaign and focus our efforts on the other.
Before we arrived at this decision, we asked ourselves 2 questions:
- Would we spend $20 (the amount we spent on the Engagement campaign so far) for a visit to our website? The answer was no.
- Are we willing to spend another $20 on the Engagement campaign? To answer our question, we asked ourselves what else we can do with the $20. The answer was to optimize the Traffic campaign and potentially get another 16 visits to the website. So our answer was no. We were not willing to spend another $20 on the Engagement campaign.
Now, you may disagree with our analysis above because of small sample size.
But remember, we are putting ourselves in the shoes of people who have small budgets. We want to balance between experimental accuracy (a luxury we take for granted) and practicality.
So we chose to kill the Engagement campaign.
Update 2: Jan 7 2017 (Day 12)
If you read our update last week, here is the updated version of the same stats we shared:
- Total ad spend: $152.58.
- Reach: 7,722.
- Total ad clicks: 180 (i.e. 2.08% CTR).
- Total traffic: 100 (i.e. 1.16% link CTR).
- CPM: $17.63.
What you see above is the overall results, which means that after killing off the Engagement campaign, we doubled overall link CTR from 0.58% to 1.16%.
Coupled with a 14.5% improvement in CPM (from $20.62 to $17.63), our cost per click improved from $3.56 to $1.53.
Though still far from our original target of $0.20 CPC, we are moving in the right direction.
Between Dec 31 and Jan 7, we did 5 things to optimize the campaign:
- We killed off ads above $6 CPC.
- If the ad happened to have sky-high CPM, we duplicated it again to “give it another chance”.
- If the ad generated at least one link click before it hit $6, we tested a different copy.
- If the ad click CTR was below 1.50%, we tested a different image.
- We published a new article on Jan 3 and promoted it in the same campaign.
Let’s look at the reasons why we did the 5 things above.
1. We turned off ads above $6 CPC.
Our goal was to get CPC close to the $0.20 mark. Thus we wanted to set a CPC ceiling without turning off the ad too quickly. We decided on an arbitrary number of $6 for a start.
Note that if you don’t have any historical data to work with, you should start from an arbitrary cutoff. The exact cutoff point is not important right now. The cutoff will move downwards as you tweak your campaigns more and more.
In the next few weeks, we’ll bring down the ceiling to $5, $4, $3, and so on, until we get to the $0.20 mark.
2. We gave “expensive” ads another chance.
Can I tell you a secret?
I’ve been saying this for years.
But your CPM may be a matter of “luck”.
Sometimes you’ll get CPMs as cheap as $1. Other times you might get it as high as $2,000, just like I did:
Side note: I once spoke with a former MySpace engineer whose ex-colleagues were hired to build Facebook’s advertising engine. They explained that you may get a different CPM each time you run the same ad. I’m not 100% certain if it still works the same way but if you want the full story, comment below.
To overcome this problem, we decided to duplicate the ad and run it again.
3. We tested a different copy if there was a link click below $6.
As you know, we set an arbitrary goal of $6 for each visit.
But there are several ways to get $6 for each visit, such as:
- You get a CPM of $0.60 and a link CTR of 1%.
- You get a CPM of $6,000 and a link CTR of 100%.
- …and millions of other CPM-CTR (link) combinations.
My point is, if we got a visit for less than $6 and the link CTR was low, there is a “chance” we can get the next visit cheaper if we just tweaked our copy a little. We wanted to test a variety of messages and style to see if we could get a higher CTR and improve the overall CPC.
4. We tested the image first if the ad CTR was below 1.50%.
This may sound confusing at first, so try to re-read this section if you don’t get it the first time.
Ad CTR aka CTR (all) tells us the number of people who clicked on some part of the ad.
When you think about it, ad CTR is the group of people whose attention your ad managed to grab.
On the other hand, link CTR refers to the group of people who were interested enough to click on the link.
So the link CTR is always a subset of ad CTR:
From our experience, we know that if our ad CTR was below 1.50%, the likelihood of getting a 1.0% link CTR is low.
Not everyone who reads and clicks on the ad would click on the link.
So instead of testing a copy, we tested the image, which affects ad CTR more than the copy.
5. We published and promoted a new article.
Do you know the number 1 thing that determines the conversion rate on a squeeze page?
It’s not the headline, design, or color of the button.
It’s the offer.
Like what Digital Marketer says often in its podcasts, none of the things you do matter if nobody wants your offer.
Likewise, every piece of content is just like an offer. The offer is the promise of some information that you’re giving away.
For example, the “hidden” offer in reading this article is the knowledge to optimize your very own campaigns. You’re only reading this because you found the offer compelling and the content useful.
So here at JSocial we are also testing different pieces of content and hopefully we’ll find something that many people want.
This will help us find out what our target audience likes and bring down our CPC.
Update 3: Jan 30 2017 (Day 34)
We went quiet for nearly 3 weeks.
Here’s why: our cost per click have improved massively from $1.53 to $0.63 (59% improvement!).
I’ll explain how we did this in just a while. But first, let’s take a look at our stats so far:
- Total ad spend: $477.40.
- Reach: 27,163.
- Total ad clicks: 1,486 (i.e. 4.16% CTR).
- Total traffic: 757 (i.e. 2.12% link CTR).
- CPM: $13.35.
In a nutshell, we are firing on all cylinders.
Ad clickthrough rates improved by 100% (2.08% to 4.16%).
Link clickthrough rates improved by 83% (1.16% to 2.12%).
Cost per 1,000 impressions improved by 24% ($17.63 to $13.35).
We even got 10 new subscribers who signed up via a mix of content upgrades and exit-intent popups.
How did we do it?
In the last 30 days, we did 3 key things and we believe one of them made nearly all the difference.
Key thing #1: We published and promoted 2 new articles on our website (including this one).
Key thing #2: We took and implemented a suggestion from one of our readers’ comments below.
Key thing #3: We monitored the people who reacted to our ad and made changes to our targeting.
Let’s go through each thing briefly.
1. We advertised 2 new articles.
As mentioned in our previous update, we believe that the offer makes the biggest difference to the outcome of Facebook Advertising campaigns.
So this month we wrote 2 new articles, “5 Honest Reasons Why Your Campaigns Suck”, and this mega-post that you’re reading right now.
Then we created ads for each article.
Here’s how the ad for the first article looks like:
…and here’s the ad we ran for this article:
Much to our surprise, one of these 2 ads took off and brought in 702 visitors to-date at just $0.53 per click. This is just $0.03 above our original target of $0.40-0.50!
Since that ad took off, Facebook spent 93% of our budget on it. It was also responsible for all 10 new subscribers we gained.
If you’re not getting the kind of results you want, test a different offer.
By testing different pieces of content, we were able to find a sweet spot and massively lower our advertising costs in just a few days.
If this doesn’t convince you, I’m not sure what will.
2. We implemented one of the suggestions from our readers.
If you look at the comments section below, you’ll see that several readers shared some really cool ideas.
One of them is Reinart:
Reinart saw in our first update that we had killed off the Engagement campaign which brought in zero traffic.
He suggested taking the URL of our Traffic ad to create an Engagement campaign. This way, the Engagement campaign will appear like a Traffic campaign to Facebook users but optimize towards engagement.
So we took his suggestions and ran with it!
First, we created an Engagement campaign and named it “1. Mobile & Desktop newsfeeds – Post engagement”.
Then we created one ad set and targeted Americans between age 29 and 45, just like we did in the Traffic campaign:
We began the test on Jan 16, although the Traffic campaign has been running since Dec 27. Here were the results between Jan 16 and 30:
The first ad set is the Traffic campaign and the second is the Engagement campaign.
As you can see, the Engagement campaign brought in cheaper traffic than the Traffic campaign ($0.52 vs $0.58).
However, 2 leads resulted from the Traffic campaign, while no lead came out of the Engagement campaign.
Initially, we wanted to kill off the Engagement campaign because we value deeper funnel conversions like leads over traffic.
In this case, we decided to keep the Engagement campaign running based on one thing: we managed to reach people that the Traffic campaign alone wouldn’t reach.
Here’s what we mean:
- The Traffic campaign reached 5,656 people.
- The Engagement campaign reached 4,567 people.
- Adding both number gives us a total of 10,223 people.
- But Facebook reported only 9,752 people in total.
- So out of 9,752 people, 471 (10,223 – 9,752) of them saw both the Traffic and Engagement ads.
- The Engagement ads helped us to reach 4,096 (4,567 – 471) more people which we wouldn’t have been able to reach if we just ran the Traffic ads alone.
- This brought us additional 134 visitors to our website at just $0.52 per click.
That’s the (rather lengthy) reason why we decided to keep the Engagement campaign.
3. We monitored likes and comments and edited our targeting.
At the start of this article, we mentioned that we targeted lookalike audiences of our email list. This includes people who have paid us money and people who have never paid us anything, which makes it less than ideal. But it was the best audience we had.
Hence we expected to reach irrelevant people from time to time.
If we saw irrelevant people complaining about our ad, we would look through their profiles and tweak our targeting accordingly.
Using our favorite social media management tool – Agorapulse, we were able to easily stay on top of negative comments like this:
After a handful of replies and comments, we noticed that Facebook was serving our ad to a sizable group of musicians.
Some musicians own other businesses like a small e-commerce store. Others are trying to sell tickets to their shows, albums, or even just trying to get eyeballs for their music videos.
At J Social, we recognize our limited experience in the music industry. As a result, this group of people are not our priority. Hence we decided to exclude “musicians” from our campaigns.
Our next step
Our plan is to retarget the 700+ visitors using a guide that we’ve just created, along with thousands of other visitors that we get organically each month.
The guide is about the 4 Stages that every Facebook Advertiser goes through before they have a highly profitable campaign.
After much deliberation, we thought that this guide would make a good first offer for our readers.
We could be wrong. But we’ll let the data tell us.
In the next update, we will share the lead generation ads, squeeze page, and results of our experiment.
Want to know when we post the next update of our experiment? Enter your email below and let us know where we can notify you.
Next few weeks, we will write about our experiments, ads, landing pages, funnels, and everything behind-the-scenes. If you don't want to miss any of that, enter your email to let us know.