[Follow Along] Paid Facebbook advertising in a small market

[TacoCat]

For those that are following my main thread [here], know that I'm mostly using Facebook ads to drive traffic.
To be honest, most of the time it does feel like I'm throwing shit against the wall and seeing what sticks.

I'm creating this thread with the purpose of writing down my experiments, results and maybe building a PPC advertising approach framework, hoping that some people that know better can pich in to make it better.

The specs of the market:

I'm in the unique situation of beeing in a small market. Thet total Facebook users (for that specific language) is around 1 million active users in the past 30 days.

The benefits of a small market:

- You can literally reach everyone and become a brand with a few thousand $$$.
- The market is relatively cheal 0.05$ - 0.10$ per click, 2.5 - 5$ CPM.

The cons of a small market:

Usual, big market Facebook algorithm rules don't apply.

- The algorithm has very little info on people (if I look at the audience insights, I will mostly get the Top 10 page likes will be the same for any targeting, since there are a few big pages here)
- Lookalike audiences that are made from customers are 10 - 100k big, and will overlap (30% or more) with any other audience that is bigger than 100k users.
- Ads reach high frequency relativelly fast.
- When the learning phase is over for Facebook - The audience has already exhausted itself and I have to switch it up.


The benefit of my product:

- I'm in weight loss, therefore it applies to a LOT of people. 60%+ of population is overweight and has weight related health problems.

The cons of my product:

- I have to tip-toe with facebook ads around the weight loss theme, since facebook will ban any aggresive copy.

What I'm doing currently:

Facebook Ad and Google Search ads -> Landing page:

If Purchase:

No More Adverts​

If no Purchase:

Retargetig Facebook IMG advert, talking about how the weight loss programm looks like
Retargetig Facebook Video advert, showing how the weight loss program looks like​

Retargetig Youtube Video advert, showing ​

Retargetig Google Display ads​

​

How the testing process works for ads:

1. I create 3 Images, 3 texts and 3 Headlines for an ad.
2. The first test is images - Campaign -> 3 Adsets (All same audience with minor age difference with Mobile and Desktop feed placement only) -> Each adset 1 ad (Only the images differ)
3. Run it for 72 hours, see which Image performed the best.
4. Repeat with Text and Headline.
5. I'm looking for CPC and CTR improvements. The campaigns are run for a conversion goal.
6. I will then run the winner to all of the audiences and keep an eye on the ROI.

What I'm still struggling to understand:

• How do you create a solid A/B testing framework, when Facebook algorithm can run one ad for the COM of 15$ and the other one with a CPM of 2.5% and other weird fuckery.
• Why do I get a few days of solid sales and then radio silence.

I sometimes feel like there is no rhyme or reason to this. But then I remember that I'm wrong and there are people out there crushing it.

Results so far:

I'm hitting around 1.2 - 1.3 ROAS on Facebook. But if I take into account Google Search Ads, I think the ROAS is negative.

The interesting part is, that my Google Search ads get a really HIGH CTR - around 30 - 40%, but little to no conversions.

THe landing page is conversting at around 0.4% (Thats less than 1).

I'm testing various sections (not colors or some minor bullshit, whole sections and angles) but nothing seems to be be bringing significant increases so far.

What I'm doing next:

I'm going to test out a different retargeting approach. People are very unfamiliar with the weight loss approach and since the market has been beaten down with different offers, I believe that they don't really see the difference from any other weight loss programm. I'm gonna run video retargeting ads, where I will explain them the differences of benefits of the programme in a Q&A style.

So you visit the landing page, then:

- In the first 2 days, you will see video, where I explain the benefits of this approach
- In the next 2 - 4 days, I explain how it exactly works.
- etc.

Let's us see how this works.

 

[TacoCat]

I have a question that I don't really know an aswer to.

How do you know when to axe a campagin and what are some good signs that it can be optimized?

A lot of campaigns have an ROAS of 1.00 , I'm not sure, if I axe em, or just tweak them.

 

[eliquid]

If you are breaking even, at a slight loss, or at a slight profit.. you keep it running.

If large loss, you axe

 

[TacoCat]

If you are breaking even, at a slight loss, or at a slight profit.. you keep it running.

If large loss, you axe

Thanks for that, the question then becomes, how do you test without stoping the campaign.

Cold campaign A is running at a ROAS of 1.0

Do you test the same audience with different creatives,
Different audience with the same creatives,
Or retarget the vistiors that came from that break even campaign, looking for the best angle?

I can get most of my campaigns to run at a breakeven, or even at a slight profit, (1.2 or 1.3), but where I fail at is scaling it up. Whenever I try to run new audiences, they will NOT convert, basically spending any profit gains that I've had.

 

[Steve]

Retargeting is usually almost free money, so do that.
For testing, use the built-in split testing facility, if you are just breaking even, you can do it to the current campaign.
If it's already performing well and you are worried about borking it, duplicate the campaign first, and run the test on the duplicate.

If you have a lot of reactions on your ads, you are better off duplicating first, else you test creatives will be at a disadvantage

 

[eliquid]

Thanks for that, the question then becomes, how do you test without stoping the campaign.

Cold campaign A is running at a ROAS of 1.0

Do you test the same audience with different creatives,
Different audience with the same creatives,
Or retarget the vistiors that came from that break even campaign, looking for the best angle?

The hard answer is, you try all of that.

However, you don't do it all at once. Meaning, don't try all 3 at one time. Isolate 1 at a time.

This right here is why success seems so difficult for some people. It's not about this magic word or special landing page, it's about the process and system/flow you go through to stack the odds on your side. Like knowing what to change and when and why.

One thought I have for you is, why aren't you capturing the audience's emails and selling them more of the same stuff later on? This would allow you to have a losing campaign that is actually ROI positive on the backend.

That's just 1 idea.

When you can start thinking like this on your own, multiple times in multiple ways.. you will be on the path to stacking success on your side.

 

[TacoCat]

The hard answer is, you try all of that.

However, you don't do it all at once. Meaning, don't try all 3 at one time. Isolate 1 at a time.

This right here is why success seems so difficult for some people. It's not about this magic word or special landing page, it's about the process and system/flow you go through to stack the odds on your side. Like knowing what to change and when and why.

One thought I have for you is, why aren't you capturing the audience's emails and selling them more of the same stuff later on? This would allow you to have a losing campaign that is actually ROI positive on the backend.

That's just 1 idea.

When you can start thinking like this on your own, multiple times in multiple ways.. you will be on the path to stacking success on your side.

Thanks for the advice.

I am collecting emails from that audience. I already got that covered.

As for the system, I 100% agree that there is no magic bullet, I just haven't figured out the process on how to test everything and actually get significant results out.

Since there is the facebook algorithm factor - you can run 1 ad and get great results, then something will happen and that ad won't deliver good results.

My conversion rate will have wild swings withtout an apparent reason.

etc.

I'm having trouble distinguishing signal from the noise.

 

[TacoCat]

Here is a little trick that I picked up:

If you put aggressive Direct Response textin your ad, the ad won't get approved.

If you write a vanilla ad, and then after it gets aproved add a Direct Response text comment with a link to the same place as your AD to your ad, you're good.

I don't think the comments get scanned for ad policy.

 

[TacoCat]

Let's figure out - Why is my CPM sometimes 4x higher than it should be?

I can safely say that there is a local average. Running ads here, when shit isn't going horribly wrong, the CPM for ads will be 3 - 5$.

But then,I will publish new ads and Facebook algorithm will be like - "Nah budy, CPM for these ads will be 20$, even though your CTR is Above average, ad Quality is above average and everything looks good otherwise."

Drives me nuts!

I want to figure out what causes this. Why do some ads get charged 4 - 5x the regular price.

Therefore, I will test every element that I can think of.

- Landing page copy (possibly too agressive)
- Ad Copy (possibly too agressive)
- Ad creative
- Headline

Campaign elements:

- Using campaign with campaign budget vs adset budget
- Optimizing for Conversions vs Traffic
- Optimizing for Video Views

Adset elements:

- Optimizing for different Conversion events
- Including lookalike audience vs saved audience
- Bulding a new audience in the adset
- With or without exclusion audiences
- Automatic placements vs Edited placements
- One vs Multiple creatives in the adset

Random effect:

If it's just an alogorithm beeing a derp, then There should be no rhyme or reason and I should get non-consistent results with any of the tests. I've already tried Copying a campaign or adset 10x and letting them run, they will get the same average, increased CPM.

Let the tests begin. I've underlined the tests that are running now.

Each test is running for a different audience, their overlap is below 10%.

Before, with the control adsets, for each audience, I got that insanely high CPM for some reason. Let's see, where the issue lies.

 

[Amir]

I never heard of landing page copy affecting the CPE. Maybe it's something new. I fucking hate running 'like' ad campaigns. Is that the type of AD campaign you're running? The CPE is way TOO much. I always boost mine, then, auto invite people who liked my post. Every week it'll 2-3 times or more than its previous week, due to facebook's snowball effect.

 

[TacoCat]

I never heard of landing page copy affecting the CPE. Maybe it's something new. I fucking hate running 'like' ad campaigns. Is that the type of AD campaign you're running? The CPE is way TOO much. I always boost mine, then, auto invite people who liked my post. Every week it'll 2-3 times or more than its previous week, due to facebook's snowball effect.

CPM - cost per thousand impressions. not CPE - Cost Per Engagement.

I don't solely focus on CPM, it's just a good number that lets me know if my campaigns are costing what they should.

At the end of the they I'm only interested on how much a conversion costs to me. But when your CPM is 4x higher, the conversion will be more expensive as well.

 

[TacoCat]

It's the ad Copy.

I'm not entierly sure yet, what might be causing it. What wording exactly, but I'm almost 99% conviced that it is the ad text copy.

Because...

I ran tests for different technical setups, but the results were varying, there was no rhyme or reason.

But then I noticed, that one specific ad was getting these bloated CPM numbers, and decided to test that ad against others.

And boom, that ad is the only one constantly getting higher CPM.

I know it's not the creative. Because I already ran the tests for that. It is the text itself. I'm not entierly sure what exactly. But I'm certain, when the facebook algorithm doesn't like your Ad text for some reason, but it's not entierly against the rules, you will get bloated ad costs and I'm pretty sure also a limited reach.

Now, further tests should show exactly what kind of wording it doesn't like.

My money is on the usage of word YOU. Or talking with the user in first person in any kind of sense.

 

[CCarter]

There is also a sentiment filter/algo that will cause problems if you “make the FB user feel bad.” So if your AD is detected to potentially put a FB user in a bad or negative mood then your AD will take a hit.

 

[MrMedia]

There is also a sentiment filter/algo that will cause problems if you “make the FB user feel bad.” So if your AD is detected to potentially put a FB user in a bad or negative mood then your AD will take a hit.

Is this real?

 

[CCarter]

Is this real?

This is a conversation with someone from BuSo about 4 months back - I doubt they'll have a problem with me sharing it, all details have been taken out:

"The trouble for me right now is the changes with FB ads algo, back in 2012 when I'd spend XXX,XXX a day it was all about just how the ad performs. But FB has gotten so much smarter now, they don't want to run ads that make people unhappy. So if you have an ad that is causing negative emotion FB detects that and and charges you more and shows your ad to people that will not convert as well.

So it's a careful balance between using emotion to make the ad work, but also not creating any kind of negative experience - but also it's a black box so you don't know truly what is going on with your ad in FBs optimization engine.

That's the type of marketing I love talking about vs the nature of SEO rankings, but the algos used to optimize paid ads and now it's science behind them and such."​

It's a smart move on Facebook's part since they don't want people to have a negative experience when on their site, so they stop that potential problem with advertisers from the get-go.

Oops - was that one of those things I wasn't suppose to reveal to the world. My bad. (This post might get deleted later on)

 

[MrMedia]

Makes sense and ties into the idea of the true addicting power of social networks as dopamine delivery systems. Negative emotions dilute the power of the addiction.

Fucking dark stuff but genius level too that has to be respected.

 

[TacoCat]

I guess Facebook just really wants this guy to buy. Lol.

***

Anyways, launched new campaigns this morning, edited the ad as per @CCarter suggestions, making it more positive.

Basically just removed the:

"With this widget, you can forget about the problems that other similar widgets cause."

Part. And boom, 4x cheaper ads.

 

[LinkPlate]

I have a question that I don't really know an aswer to.

How do you know when to axe a campagin and what are some good signs that it can be optimized?

A lot of campaigns have an ROAS of 1.00 , I'm not sure, if I axe em, or just tweak them.

Better late than never to answer this :-) I usually look at the organic conversion rate as a base line for testing (your paid traffic should convert at a higher rate than organic, because you control the traffic and assume, you have landing pages). If you don't have enough organic traffic to have a base conversion rate, then I would calculate what conversion rate you need to break even and use that as the base line for testing.

 

[Tiberian]

How are you handling placements?
I noticed that when I started to include more placements such as in search and marketplace rather than just the "optimal" ones , I saw my CPMs go down and interestingly my ROAS actually improved. I will admit that my niche is far different from yours so different rules may apply.