Hey Karen,

Keeping in mind that when it comes to FB's algorithm, most of what I say is just an educated guess... My guess is that the combination of the off hours start time of the campaign, coupled with the much wider placement reach (even if not much was spent there), made the algorithm scatter more as it searched for the ideal segment. And then, before long a quality score of sorts was assigned to the campaign and it just could never really recover.

It could also just be that given more time it will come down, and given more time the old campaign could have gone up.

I would try going back to the original campaign and create a new ad set within that campaign, but with all your new settings, and just see how that does. Start the new campaign early in the morning and see how it does. For this test, stick with just Facebook. If this does well, create a new ad set that just goes after Instagram, etc.

If it doesn't work, try again in a new campaign.

What you have experienced is not a shocker. I have stuff like that happen all the time. I put it down to how things kicked off.

In terms of the campaigns, here is all I really know... There is an old concept of something called a "quality score". This goes back to Google advertising and is a term that marketers through around. But with Google it meant that every aspect of your campaigns had a quality score associated with them. If a url had performed badly, it started off with a few points against it. If a campaign was doing badly, it was hard to create new ads within that campaign because it had a few points against it. And likewise, if something performed well, it started off with a few points in it's favour.

Facebook does not actually have anything they call a "quality score", but in my conversations with Facebook they have told me that their algorithm does something similar and that if an ad was doing poorly it was better to scrap it and start a new campaign, and upload new images even if you had used the exact same images before.

My assumption is that the same goes for positive indicators, and if a campaign is doing well, you would do well to create new ad sets within that campaign. Anecdotal evidence seems to suggest this is the case, but I don't know for certain as the details of Facebook's algorithm are not all public.

In terms of time of day, all I can say is that I think starting an ad early in the morning is ideal, but not at all essential.

Karen Grace said
Thanks for this.

Well, it's two days into my second attempt at a Bjork static ad (within the same campaing as the dynamic creative one that went well), and what's really weird is that it's even worse and the cost per result is nearly twice as much as the overall cost per result for the dynamic creative ad. So weird- I was so hopeful about Bjork's audience as the dynamic creative ad was the best so far

A few possibilities come to mind.

1. You just had an artificially positive campaign the first time around. Just as things can trend down in teh beginning of a campaign, the can trend up as well.

2. 2 days isn't long. I have had campaigns take a week to optimize. 5 days is what Facebook recommends. But often it doesn't take that long.

3. If the audience is small then you may have played out the optimal segment already. 

4. If it doesn't work, start a new campaign. Sometimes things just start off week and a new campaign with no other changes other than a fresh start can make a big difference.

... 

Here are my thoughts:

- Firstly, I didn't realise it was a good idea to upload a new image (I always use the same one.) So perhaps I should try again with a newly uploaded image ? Should I create a new url for the squeeze page too? Or was that comment only relevant to google ads in the past?

I wouldn't stress this yet. Especially if the last campaign did well.

- I also notice with the Bjork DC ad that many of the images and text elements were actually pretty close to each other in cost, so while there were winners, there weren't any that were streets ahead of the rest. Perhaps this explains why the overall dc ad went well on average, but perhaps it's actually more helpful if there are really obvious winners for elements that you can run with? Would you agree with that?

I personally don't think that s a thing. But I could be wrong. I have not seen anything to suggest it works like that though.

Lastly a couple more questions about how to group things. I realise you don't know all the ins and out of fb for sure, so I guess I am asking what do you do?

- I've sometimes been grouping different audience ad sets in the same campaign just because I'm running them at the same time. Is it best to keep campaigns for one audience only, incase an image goes well for one audience but doesn't go well for the other audience, impacting the good one? Or is it just a case of always using unique image files even if they're the same image?

Grouping multiple ad sets with different audiences in the same campaign is fine. If one ad is doing well and one isn't, the one that isn't won't drag the other one down as far as I know. I group like this in almost every campaign I run.

- And at the ad level, I'm wondering about split testing two different static ads after doing DC ads. If I run two ads in the same ad set, do they compete against each other ? Does that mean the prices go up?

No, I would not worry about that.