Home > Paid Traffic Sources > Native

---

How do you cut the crap, without crap all over yourself? (9)

---

Hi all,

Working on Taboola and Outbrain atm.

I know there are hundreds of ways to optimise on this platform. I've lost a fortune exploring a lot of them!

Managers on both platforms seem to have little or no system for best practises on this. They always just pressure me to scale on funnels that are spending a lot but losing me money...

Questions:

1) In what order should I be optimising or cutting the following:

- Banners
- Countries
- Sites
- CPCs
- LPs
- Offers

2) When should I start tinkering and cutting things? e.g. days, certain spends?

3) What are reasonable adjustments to make without risking the algo nuking your camp with shit traffic?

It seems like one wrong move, even a small one, and the whole campaign will either lose all traffic or the CPA will double for no logical reason...

If I get some solid feedback and decent pointers here I'll try and make some time for a follow along!

---

I would do it like this: sites (actual source, needs to be on point when testing your funnel properly, so get rid of the bad ones asap), banners (find good source and THEN test banners) , cpcs (in alignment with banners I think?), lps (CTR is the second important thing in the funnel, and "easy to optimize yourself), offers (once the first two parts are set, start rotating offers), countries (when this all fails switch geo's, assuming you did some decent reseach on the geo you started with)

ofcourse some will overlap, or mix a bit (since I am always looking for better offers)

others might approach this totally different, maybe because I do Facebook mainly, but I think the idea is kinda the same. also your skills, characteristics etc are playing part of which path you choose...

---

In my opinion the two networks are very different, and it also depends a ton on whether you are using SmartBid or not.

I personally always use SmartBid on Taboola, but others have other strategies too.

On Outbrain lately I have been using their version of SmartBid "optimize for conversions", but it doesn't seem quite as "Smart" as the Taboola SmartBid.

On Taboola, I would recommend this:


Create multiple duplicates of each campaign with each duplicate having slightly different targeting, slightly different angle of ads, whatever... treat each campaign almost like an "ad group".
Then spend 90% of your energy on finding better ads and editing the ads (I let ads run on 'Optimize' but there you can also do 'Balanced')
Spend 10% of your energy looking at the results by publisher, but only block the absolutely horrible ones (like 0$ over 100$)
If a campaign isn't working within 300-500$ spend, just pause it and duplicate it so the new version retains the list of blocked sites.
After enough iterations of tweaking ads + angles + very tiny publisher edits + duplicating, you should eventually find success if it is possible to do so.
Always be deleting worst performing ads and always be uploading new ones.
Continually split test new landing pages (but don't put as much time into this as you do ad creation as its more important)
All of the above presumes you have a great offer, as obviously everything else follows from this.



On Outbrain I would do this:

Create one single campaign per device type and don't do a lot of duplicating like on Taboola (as it takes longer to 'warm up' new campaigns on Outbrain).
Batch geo's if possible rather than doing individually (algo seems to like this), although its not always possible of course.
Spend 50% of the time creating better ads.
Spend other 50% optimizing publisher and blocking bad ones very quickly by lp ctr (if average is 40% and something starts out at 5% I would block it) and less quickly but still pretty quick by CPA (if you need a 20$ cpa and it has one conversion over 60$ I'd probably block it...).
Continually delete poor performing ads and upload new ones.
Continually split test new landing pages (but don't put as much time into this as you do ad creation as its more important)
All of the above presumes you have a great offer, as obviously everything else follows from this.



That's just what has worked for me, other folks could have exact opposite take!

And also disclaimer that I haven't had a ton of success on Outbrain lately... more so on Taboola... although in the past when running different types of offers I've definitely had ones where Outbrain did much better!

Hope that helps and yes, follow-alongs are great you should consider doing one! Would love to see your progress!

---

Rough order:

Offers (most important)
Ads
Bids
Countries
LPs
Sites

I have maybe 5 sites blocked on a Taboola sub-account for an offer with 6 figures in adspend. The traffic quality there is awesome, so if you're blocking sites your offer most likely sucks or your ads are low enough CTR to prevent you getting any real volume at decent bids. Instead of blocking sites that convert but have high CPA, just drop the smartbid down 30-50% and see if it still spends.

First priority of a new campaign is simply bidding high to establish the ads within the algorithim. I like to spend 250/day to start, see how it does and what the ad CPM is, how close to break-even it is, before usually doubling the cap and dropping bids by no more than 10%, so like from $0.65 to $0.60 at higher cap.

If you get ANY conversions and your ad CPM is decent enough to still drive volume at lower CPCs and CPMs, because you're starting by overbidding, then there's a very good chance you have a winner on your hands. It then just becomes a matter of how much volume you can drive given your goal for ROI and how limited you are by cashflow.

Generally I wait a few days before making cuts to big things like ads, LPs, and sites, assuming I'm going from 250 to 500 to 1000/day.

Don't make drastic changes to bids in the beginning, eg more than 10%. Save that for after campaign has spent a few thousand. There can still be swings on big sites, I've had MSN sites convert like crazy at volumes then something happens with ad performance, ad gets blocked, etc and spend drops and CPA doubles or triples. It happens. I feel like it's largely still volume related though as a 1k/day campaign on MSN is still small and fluctations are going to occur in the beginning. See if there's any dayparting you can do for starters, eg weekend vs weekday or just up bids during certain times.

---

Well if the weekends are consistently poor then I'd manually drop the bids then. Usually weekends do better, but I also experienced a drop off this Sunday morning which was unusual.

I'm just now getting back to running Outbrain but inconsistent performance and quality issues like that are exactly why I jumped ship to Taboola and got going there first.

I'd work on testing some better ad creative first and foremost. What's your ad CPM at right now?

---

"Outbrain, inconsistent performance and quality issues" - never a truer word...

CPM is $0.73 on that camp.

---

Sorry to interupt, but thedudeabides

What do you mean by a Taboola sub account?

---

You have a Taboola 'network' account, and then for each campaign or vertical you will have a sub-account made.

---

Thank you

---

Home > Paid Traffic Sources > Native