Home > Native >

---

[Followalong] Native Ads Follow Along - Lets go! :) (7)

---

Hey everyone,

I started Native just today on Taboola and am currently promoting some eCom offers.

I'll be trying out offers primarily in the eCom vertical via affiliate networks.

I'll (hopefully) update on a daily basis so stay tuned.

---

Day 1 Stats :

So - potatoes for day 1. The cap was set at 100 USD, the bid was .7 USD.

I hit the cap pretty early on, this is what the hours look like :



So I hit the cap roughly 8 hours in.

Here's the ad data:



And as for the LP, I'm only testing 1 LP right now. So 9 ads x 1 LP x 1 offer for this campaign.

Questions:

1. I hear some people like to clone their campaigns initially, 3 clones, they let it run and allow the system to pick out the best campaign. Then they pause the other 2 campaigns. If I was to adopt this strategy, would I let the 3 clones run for a single day at lets say 100 USD cap, let their AI figure out the best campaign and just go with the one that performed the best? So that'd be 300 USD spent on the first day and then 100 USD per day for the best camp after the first day. Do you guys recommend this strategy?

2. Also with regards to cloning campaigns, do you guys also recommend hitting the clone campaign button or creating a campaign from scratch with identical targeting and inventory?

3. How long do you guys recommend running a campaign with 0 conversions or like lets say -80% ROI before turning it off? From what the Taboola interface tells me, it takes up to 11 days for their AI to become fully adjusted. I'm doing 100 cap per day at .7 USD bid, so do I follow the 10x CPA rule or let it run for roughly 11 days?

4. For site level optimizations, when do you guys recommend messing around at the site level? After their SmartBid AI has collected data on the camp for lets say a week? I hear some people don't mess around with the sites at all and some people blacklist regularly.

---

Hey man-

Great stuff!

Here's my thoughts on your questions... I'd caution you though that there's lots of ways to succeed on Taboola and there are people who do things way different than me who kill it on there. Many ways to make it work (though none exactly easy).


Here's my thoughts on your four questions:


1. That's definitely one option... usually when I start with 2-3 clones at the very start I like to make them slightly different... so different bid levels, or different geo's, or different device targeting, etc (though in your case with physical ecom products I'd stick to desktop at the start for sure). But yeah, it's a balance of wanting to maximize your chances of having a campaign work vs wanting to minimize your spend. In that regard, if you don't know if the offer will work, you might want to just test one campaign at a time to start rather than multiple clones, just to save yourself money.


2. Since the campaign is brand new it won't really make a difference if you clone it or build a copy from scratch, especially if you are using the same Conversion Event. Later on down the road it might make a difference, where you would want to clone it if its doing well vs start from scratch if doing bad and you want a fresh start.

I'm not entirely sure what campaigns "keep" when they get cloned... it does seem as though new cloned campaigns re-enter the "Learning" phase of SmartBid, but at the same time it seems as though they also retain their "juice" in some capacity or another that is beneficial.


3. If you get to 200-300$ without a conversion I'd probably pause the campaign. Taboola always send the best clicks on day one so if you do horrible out of the gate it usually means the offer is not going to work. So yeah, I would let it run for 11 days if its doing okay certainly, but if its obvious after day 2 or 3 that the campaign is not going to work then I would definitely cut my losses. Even if you see it working on Adplexity, that person may have a unique setup you can't replicate, so it really just comes down to testing lots of offer in order to find something that works.

Remember too that its going to take you dozens of campaigns before you "master" Taboola (or any traffic source), so really the faster you can get to that point where you've tried out 50+ camps the sooner you'll start making consistent profits. So don't be afraid to give up on campaigns quickly- you can always restart them in the future if you want.


4. This is the perennial question... as you said there's different philosophies... I don't like to even look at the 'Site' breakdown (unless I'm killing it then just for fun). With that said, if the campaign is sort of on the line where it could succeed or fail, I will block a few really poorly-performing sites that are doing awful.

At the start you'll get a few bad sites that go 0/100$, but one thing I've noticed is that if I look at the site breakdown on a winning campaign where the total spend is like 30-40k+, there's never any such sites in the top 20-50 sites by spend. So that must mean that eventually the SmartBid does stop sending traffic to those sites, which is good. Also its hard to predict which sites are good and bad. Case in point some of my most profitable widgets on Taboola are Pushnami ones, which you would think would be awful (and in some cases are).

Anyway hope that helps!

PS Don't forget to try lots of ads! Super important!

---

Thanks for that breakdown. Your revcontent thread has helped me greatly and I've been wanting to break into some taboola as well.

---

Yeah so I asked my AM about this and he said that the cloned campaign will retain some insights of the creatives and site performance. Very mysterious, their algorithm.



Day 2 and 3.

Day 2 stats :

At the campaign level :




So based on Ralfs presentation at affiliate world, I combined all devices into one camp for one of the camps and just desktop for the other.

I set the cap at 250 on both for day 2 as I was getting capped pretty early on and thought 'lets actually test the afternoon as well'.

So the all devices camp actually performed better. The CPC was set at .43. The desktop one totally bombed again for day 2.

Heres some ad data for the all devices camp :




Day 3 stats:

At the campaign level:




and cumulative ad data on the all devices camp :



I did notice one thumbnail doing relatively very well :




So the all devices campaign I actually decreased the bid to .3 on Day 3 which I'm guessing was a contributing factor into this -85 ROI since I would stop getting HQ pubs. I think a better approach would've been testing out ads rather than decreasing bid from .43 to .3.

The desktop campaign, since it was failing so hard, I decided to make numerous adjustments, I decreased the bid from .7 to .57, blocked the site with highest spend ($116) that had 0 conversions and switched from Smartbid to Fixed bid. However, I did increase the bid of the HQ sites to .85 from the baseline of .57. Now, I did this because at this point, I just wanted to test out how it would work if I switched to Fixed bid and just bumped the bids of the HQ pubs. What ended up happening was that traffic from pubs that Smartbid would normally I think block, flooded the campaign. And since I didn't have the optimizer or any other rule alogrithm in place, the pubs that I actually bid high on didn't really get spend but the pubs that no one wants flooded the campaign (because I turned off smartbid). So in the future if I ever do fixed bid again, I will always need to have auto blocking rules in place.

And as for the desktop + tablet campaign, this was due to the recommendation of the AM who said that the user experiences are similar so advised targeting both at the same time. Performed better than just the desktop one so some merit to that, certainly.

So all in all, I'll have to dump this product and test out some other products.

However, for the 'all devices camp' I may test out ads since it was -40% ROI on day 1. Also, how would you test out that camp further? That one image did 4 conversions. So how would you guys proceed here?


But yep definitely going to be cutting losses sooner on obviously failed camps and going to be testing new products right away.

Questions :

1. For testing ads, how do you guys recommend going about it? 1. Uploading ads to the same campaign or 2. hitting the clone button and uploading new ads there or 3. creating campaign from scratch with same targeting and uploading ads there? Also, I had 2 headlines that performed about the same, one had 3 conversions, one had 2 conversions. I hear that headlines last awhile and the one with 3 conversions was a slight modification of an Adplexity one I saw. So what I'm thinking is testing out 6 images on that one headline with 3 conversions to see if I can improve ad CTR. But, how would you go about testing ads?


2. So for ads and testing them. Lets say a campaign showed some promise. Would guys duplicate that campaign like 3 different times and have 9 headlines and 9 images spread out across those campaigns to test? So for camp 1 : 3 distinct headlines x 3 distinct images, camp 2 : 3 distinct images different from camp 1 as well , 3 headlines distinct from camp 2 as well, camp 3 : 3 headlines and images distinct & different from camp 1 and 2. Also when do you guys start pausing ads? 1x CPA, 2x CPA? So what I'm trying to see is how to test the ads properly. I have absolutely no idea how to proceed here.

3. Do you guys have any tips for headlines and images in the ads? How do you craft your headlines and where do you find your images? I get my headlines from Adplexity, sometimes I craft my own and sometimes I use this tool to validate said headline : https://www.aminstitute.com/headline/ and aim for greater than 25% EMV. For the images, mostly Adplexity and Google. What do you guys do?

4. For the 'all devices camp' I may test out ads since it was -40% ROI on day 1. Also, how would you test out that camp further? That one image in that camp did 4 conversions. So how would you guys proceed here?


Many many thanks in advance. Hope you all have a wonderful day ahead.

---

Great follow along so far. How do you pick a bid when testing? Also, any update on your progress?

---

Great follow along @dokdok!

Judging from your stats the offer is already showing some positive signs, however in order to turn it profitable, you will need to make some thorough optimizations.

Here are a few thoughts:

- Ask your account manager about the conversion rate the offer is having on the network level. Once you have it, you can compare it against your actual conversion rate. Based on your actual stats shared in the above screenshot the situation looks like this:



With the current conversion rate and bid levels, you would ideally need a LP CTR of approximately 50% in order to break even (assuming everything else remains the same).


So instead of shooting for a higher LP CTR, I would suggest to make a similar analysis like the above on the site level (main ones) in order to get a better picture of where most of the landing page clicks and conversions are coming from. At the same time I would look and take actinos the following:

- Block any sites that have spent $25 or more but have not generated at least 1 landing page click, therefore no conversions. With the current conversion rate you would need on average 55+ landing page clicks in order to get a conversion.
- Block any sites that have spent 1x payout ($40) without generating a conversion even tho they may have a nice LP CTR. Quite often it is best to kill a spending site when it is not generating conversions rather than keeping on spending on it with the hopes of a conversion.
- Check out the Avg CPC, ROI and EPC on converting sites and adjust the bids accordingly. If the ROI gives you room for higher bids, try it and see if they convert better.

When it comes to ad level optimization Taboola tends to allocate most of the traffic to a single ad, especially on SmartBid and Optimized traffic allocation. Here you can play around with the campaign cloning to see if another ad will perform better.

In addition to the above, it is VERY IMPORTANT to see if you can introduce any improvements on the landing page that could lead to a higher conversion rate. Maybe add a different landing page style and see the results.
Sometimes just a small change on the landing page can increase the conversion rate by 50% in which case, you may even want to re-test some previously blocked sites.

---

Home > Native >