1st steps with facebook + e-comm (4)
Hey everyone!
I was lurking around for sometime but I see that the best way to gain knowledge is to start acting 
So, I'm currently testing some products from local e-comm shop (childcare products) and I believe that FB is a best place to start with. I have never worked with FB and this is my first step in a long journey.
I've tested few campaigns so far but I have no idea what I'm currently doing and what results I should aim for in terms of CTR, CPC, ATC ratio.. etc., so I hope that you guys can give some thoughts on this.
The funnel I use for this product is simple: ad -> shop/product page. Without advertorial/landing page.
Long story short, my current campaigns stats:
The way I run these campaigns:
I. General campaign setup:
- 3-4 ad sets per campaign
- 3-4 ads per ad set (different ad formats: video, carousel, image)
- conversion based optimization
- ad set budget optimization with $5-10 daily budget per ad set
II. Campaigns launch order:
1. LPV with broad audience (potential reach: 6 millions) - to get some idea how the hell FB works (got 1 random sale) - stopped.
2. ATC with broad audience - to start feeding FB pixel with some data (got less ATC compared to LPV campaign) - stopped.
3. ATC with LAL (3% of those who checked the website) - 38 ATC and 1 sale - stopped.
4. Messed up with testing different audiences and creatives - got 30 ATC and 2 random sales - stopped.
5. IC with broad audience - this one was the "best" one so far 123 ATC and 7 sales, random sales as I can see - stopped.
6. IC with LAL (3% of ATC) - 37 ATC and 1 sales - active.
Let me show you my "best" campaign's stats by ad sets:
As you can see all leads came from a few ad sets on low costs just after campaign was launched but.. I made a HUGE mistake - I changed ads in every ad set where sales came from 
, so all of them went to review and I decided to duplicate them with high budget. Aaand nothing happened. New ad sets got around 20-30 ATC but with 0 sales and this campaign died for me. This was a best day in my journey and I haven't seen anything similar to this one so far.
And a screenshot of my ads in this "best" campaign:
As far as I see the overall performance isn't so awful (or maybe it is.. but I don't know 
) and current ads CTR and quality score looks promising but as I said - I have no idea what I should do next and how can I make it stable and profitable.
* All sales are random and I can't see any patterns here.
* Some ads got few ATC and sales, while there're ads with lot of ATC but 0 sales.
* FB send all traffic to carousel ads like crazy and this format works good. But in order to test other ad formats I have to make ad set per ad format.
So, what do you think guys?
Can you see any potential in this product or maybe I should drop it and start testing something new?
How much should I spend for initial test to see if it's promising when an average order is ~$15-20?
Does this ads CTR and CPC look healthy? What ATC ratio and CR I should aim for?
Is it possible to make it less random and more stable?
Maybe, as a next step I should clone my "best" IC with broad targeting and more ad sets in order to find quality audience pools and spend more money even if there's no sales at first.
Or maybe this product is a shit and with current CR and ATC it's not worth to test it..
So many questions and I'll appreciate if you can answer at least some of them
Very nice beginning to what looks like a really promising follow-along! I can't believe this is your first campaign - good job!
Sorry I missed this thread - had the tab open but forgot to come back to it.
I. General campaign setup:
- 3-4 ad sets per campaign
- 3-4 ads per ad set (different ad formats: video, carousel, image)
- conversion based optimization
- ad set budget optimization with $5-10 daily budget per ad set
II. Campaigns launch order:
1. LPV with broad audience (potential reach: 6 millions) - to get some idea how the hell FB works (got 1 random sale) - stopped.
2. ATC with broad audience - to start feeding FB pixel with some data (got less ATC compared to LPV campaign) - stopped.
3. ATC with LAL (3% of those who checked the website) - 38 ATC and 1 sale - stopped.
4. Messed up with testing different audiences and creatives - got 30 ATC and 2 random sales - stopped.
5. IC with broad audience - this one was the "best" one so far 123 ATC and 7 sales, random sales as I can see - stopped.
6. IC with LAL (3% of ATC) - 37 ATC and 1 sales - active.
Ultimately, setting the conversion event to PUR will bring you the most suitable visitors. The FB algo would know to target users that would be the most likely to purchase, not just click on the ad and then bail, not just add-to-cart and then bail, and not just initiate checkout and then bail.
Having said that: Some advertisers like to set the conversion event to higher-up-the-funnel events like what you're doing. The benefit is you can collect more data for the same/lower budget. The idea is to get enough data so you can build LALs that will (hopefully and ideally) convert better than the original general targeting. So - unless you're really bleeding, you may want to continue running traffic until you have enough data to build the LAL for the next step in the funnel.
But if you haven't tried running with PUR as the conversion event, I would try that first, just to see what the ROAS is. If you're not bleeding out of your ass, I feel that that would be a better way to collect data (again, because the FB algo would then be targeting users that have the tendency to purchase). And with this approach, you'd STILL be able to create LALs for LPV, ATC and IC.
(The fact that your "IC with broad audience" camp was the best-performing one, is an indication that your ads may already be good enough to get people to buy, without involving the other funnel steps and LALs - so definitely go for PUR! And if that works out, no need to mess around with the other camps where you have LPV/ATC/IC as the conversion event.)
Another thing to consider: Most people just skip IC altogether in the initial testing stage, because angles and results tend to be similar to ATC. If you're seeing that in your data, then maybe skipping IC for now could save some test budget.
Regarding "random conversions": It could just be because you haven't spent enough money for any sort of pattern to emerge. It doesn't look like you're bleeding much - in fact there are some ads with nice ROAS that may turn out to be winners. It sounds like you may be stopping all your campaigns before they have a chance to collect enough data.
As for FB preferring carousel ads: When you put different ads in the same adset, FB will very often "favor" one and send all/most of the traffic to it. Personally, when I'm doing initial testing to find out what angles works for each audience, I like to put only 1 ad in each adset, copied 3 times so there are 3 copies of the same ad in the adset. That way I can get a better idea of the "true" performance of each ad if there's even such a thing. When you put different ads in the same adset, there's always the risk of one just being "lucky" and got picked by the algo as the favorite, leaving the other ads untested.
And when you say you're targeting a "broad" audience, are you just targeting country+gender+age? Just wanted to see if there's anything we could look at and figure out for audience targeting.
I would definitely suggest to continue your testing. If the broad audience works out that would be best! Give your ads adequate time to collect data until you can tell whether it's a winner or loser. Pause adsets that are losers, and replace with new ads (either by adding more adsets in the campaign, or by starting a new campaign). You already know not to edit existing ads - not only will this put them back into review mode and possibly "break" its original momentum and performance, but then when you analyze stats you won't remember which ad version resulted in what performance.
I don't even like to edit adsets that are running with good results. Any time you touch something, there's always a chance of breaking it. So I'd normally either pause or duplicate and adset, and try not to tamper with it. But of course YMMV.
Best of luck - again sorry for the late reply! Looking forward to seeing how you do!
Amy
Hey Amy,
Thanks for your reply!
I still try my best to tame that beast
Originally Posted by vortex
Starting from LPV optimization and up to IC/PUR.But after messing around with LAL campaigns and losing some money I can say that it doesn't help much. Probably because the audience just doesn't have the tendency to purchase.
Originally Posted by vortex
So, after losing some money on LAL campaigns I start PUR campaign with just a broad audience and proved creatives and it shows promising results so far
Numbers are low and there are only 1-2 sails/day (probably due to low budget
) but at least I can see that it can work.Current campaign set up is:
1) CBO with $5 daily spend limit + lowest costs as a bid strategy.
2) 2 adsets with optimization for Purchase + broad audience.
3) 3 Carousel ads (same text but different photos) per adset
By the way, this campaign is stopped because I try to manage some kind of time targeting, so I pause it at night and activate it in the morning manually. Does it sound legit? Or maybe FB thinks that I'm crazy and this approach hurts it's algo.. does it better to leave this campaign active 24/7?
So, for the next step I want to make it a little bit stable in terms of sales.
I'm going to clone this PUR campaign with higher budget and 2 adsets with 3 exactly same copies of the best creative in each.
I still don't know how to deal with time targeting. I see that it's possible to use ad scheduling but with lifetime budget only however I don't know how it will affect ad delivery and campaigns performance.
And this is for a broad audience - great scaling potential!
1) CBO with $5 daily spend limit + lowest costs as a bid strategy.
2) 2 adsets with optimization for Purchase + broad audience.
3) 3 Carousel ads (same text but different photos) per adset
By the way, this campaign is stopped because I try to manage some kind of time targeting, so I pause it at night and activate it in the morning manually. Does it sound legit? Or maybe FB thinks that I'm crazy and this approach hurts it's algo.. does it better to leave this campaign active 24/7?
As for the number of adsets: Test various numbers. I've had the best success with 5 adsets when I started playing around with CBO and I've kind of stuck with that since.
Try this: Identify your best ad for your best audience. Duplicate it into a new adset in a new CBO campaign with a $100/day budget, duplicate the ad 2 times so you have 3 same ads in that adset. Then duplicate that adset 4 times so you have 5 same adsets in the CBO.
This approach is from Nick Peroni's posts on FB - I was already using something similar with success before. It doesn't always work, but I find that it seems to work better when targeting a big audience.
As for time / dayparting: I wouldn't suggest to pause it every night, unless you've collected enough data and seeing a definite trend that you're losing money on most nights. What I do is create an automatic rule that would check for ROAS (or any other relevant KPI) at around the middle of the day, and if the performance is bad, the campaign would automatically be paused. And I'd also create a rule to unpause the campaign at midnight again to give it another chance to perform. This is synonymous to buying and selling stocks, where you'd put in a stop in case it drops, and run long for as long as prices are going in the right direction. So if performance is looking good for the day you keep it running, and if it looks bad you pause it out for the day. Can save some money that way.
As for what numbers are considered "good" or "bad" - you'll know after collecting data for a while.
Major KPIs look different for every campaign.Another thing: When testing new ads and/or new audiences, try to either use ABO, or if you must use CBO, set a minimum spend on each adset. Otherwise, some ads/audiences will not get enough traffic to show their true performance. (Same issue as what I was talking about - certain ads getting "lucky" and hogging all the traffic because the algo favors it for some reason.)
I hope you're at least as excited as I am! You obviously have the right products and messaging and the right audience. Targeting broad works so there's scaling potential. Be bold with your budget - get used to spending money! But DO watch your stats like a hawk so you can pause (adset/campaign) when stats aren't going the right way. (Alternatively, create rules like the ones I mentioned above.)
You want to keep collecting data until you can build LALs - THAT's when the fun will escalate!
And in the meantime, don't stop testing ads! Are you using UGC (user generated content)? If not, please try to get some. If so, try to get more!
I look forward to your next update!
Amy
Home > Facebook & Instagram > Follow-along Campaigns