Home > Paid Traffic Sources > Native

---

Native Test #2 - $154 spend - $30 revenue (proven product) Taboola (13)

---

I am testing new product since my $2000 failure.

So far spent $150, and made $30 revenue.

I am blocking every placement that isn't a news website.

Running only desktop.

I only got 1 sale, and that was a fluke on day one. What to do?

---

What is the logic behind blocking non-news websites?

What does their performance look like? Are you getting landing page clicks from these websites? What about the other ones? What's the average LP CTR you are getting until now?

---

The logic is when I spy on Adplexity, 98% of the winning campaigns, top 10 placements are always news sites.

This tells me that they have the most traffic. And the offers get the most sales from there.

---

The sites you see on spy tools are not necessarily the best ones the campaign you’re spying is running on.

On the other hand, if you’re trying to target sites based on what you see on spy tools, you might get to higher competition (higher CPCs) if other affiliates are using the same strategy. Yet, other than that the ad was served on those site, you have no other metric showing you that those sites work.

I’d suggest optimizing towards the actual performance your campaign sites are showing. Landing page click through rate is a pretty reliable metric to start optimizing.

---

@Atnakov - I think you're overthinking things on blocking the news sites. First of all, what you see in Adplexity can be deceiving... once you're on there for awhile you'll notice that the "Top Ten Sites" on your own campaigns don't always match your actual top ten sites. Furthermore there's lots of things that get magnified or hidden on Adplexity based on the complexities of scraping ads all over the internet, so it's hard to use Adplexity as a guide towards actual site optimization... it's better for just seeing general trends, seeing landing pages, finding good images, headlines, etc.

Especially on Taboola, blocking sites is the last thing you want to do. On Revcontent or Outbrain it can be okay, even encouraged (if they aren't performing well), but the problem with Taboola is that A) if you are using SmartBid then blocking lots of sites will just confuse the SmartBid and you'll do poorly, or B) if you are using Fixed Bid, there's 20,000+ sites so blocking sites manually is futile anyway...

This is what I would recommend you do. Others might do things differently but this is what works for me.

Find an offer that works - upload 12 headline/image combinations including those you see working and a few of your own - split-test a lander roughly similar (but not identical) to what you see working and a more custom one you make trying to "beat" it - set to SmartBid - set images to "Optimized" - set to Balanced budget at 3000$/month (1500$ if you're worried about losing too much money) - bid just slightly over where you think you'll need to bid. Let sit for a week without doing anything - don't block any sites - only block ads if they spend 1.25X CPA without a conversion of it they never start spending after a week of being active.

That will give Taboola's SmartBid time to actually make things work for you. If after a week the campaign failed, start from scratch with new targeting or with a new offer.

I lost tens of thousands of dollars on Taboola when I started trying to "make" things work like I would on Revcontent (aggressive blocking/whitelisting/bid-changing/etc) and it wasn't until I just gave up and trusted the SmartBid that I began having success. Ever since then its been by far our biggest and most profitable traffic source.

---

I like the idea of trusting the algorithm and avoiding excessive campaign management. But in an extreme case, if the algorithm has spent $100 on a placement which has not converted and has a zero Landing Page click thru rate, then surely that is a junk site & it makes sense to manually block it? Especially since I have visibility of poor LP CTR rates which the Taboola algorithm is not aware of?

I don't understand the point about about manually blocking sites 'confusing the Smart Bid', could you explain more about this?

Similar issue as the previous question i.e. trusting the system vs manual intervention.

Using Optimized delivery the algorithm just seems to pick one ad based on a small amount of data then just sends 90% of the traffic to that, when it's very clear that there's not enough data to conclude which is the best ad.

Even worse, the Taboola help guide suggests that optimized delivery is only based on CTR and is NOT picking the ad with the best conversion rate (Although the Smart Bid is bidding based on conversions so sites which convert best should get more traffic)

Surely it's better to use Even Ad rotation and maunally pick the best one?

---

As a matter of fact, Taboola is totally unaware of the landing page clickthrough rate or anything behind the ad click for that matter. And I think this is why there is a mixed feedback and experience on discussions about Smart vs. Manual bidding.

When having your campaigns on Smart Bidding, Taboola algorithm tends to "supposedly" run a test on the site level in order to find the sites that deliver better ctr and hopefully conversions based of the different bid ranges their algo delivers your ads. However, it's not very clear how Taboola's algo behaves on the site level besides the fact that it plays around with the bid level from -50% to +99% or maybe on a wider range. With that being said, there are also many cases where Taboola tends to bring up considerable volumes of junk push sites from one day to another that can quite easily drain more than 50% of your daily budget without any promising stats nor any conversions at all.

So basically, better kill fast whatever site that does not deliver, rather than keep on spending on such sites without any promising stats. Obviously it's best to make a thorough testing so that one does not block any site prematurely.


This happens almost all times. Whenever you have images on Optimized mode, only one of the ads will get like 90% of the traffic, leaving everything else behind with very small to no traffic at all. Practically if one wants to do a proper ad-level split-testing needs to set ad delivery options to Even, then make sure their ads have the right amount of bait in order to satisfy both their results as well as those that Taboola's algo is looking for and influences on the average cpc.

---

I agree with platinum. What i would add is spending 150$ that will most likely be distributed on dozens of websites, getting 1 sale from the first day it's just luck, not consistency. Secondly, when you start running traffic on taboola you need to give it a few days though, this way you'll get some data.You also need to look at the quality of your traffic, is it tier1 tier2? It often happens that 90% of your budget might be spent on a shitty traffic source because of the ctr.
From my experience, smart bid is a good way to start, but it's more used for scaling rather than initial testing. @platinum i agree with even distribution settings, this way you give your ads even chances to test, then you can change to optimized mode once you have a winner. I personally prefer fixed bid strategy, but you need to put in a lot of hours (but it will be worth it) and make sure you use a third party optimization tool. It's a life saver once you know your kpis, you can put in automated rules in place and run on auto pilot.:

PS: you're missing out a lot by not testing mobile also.

---

This conversation is interesting.
My personal approach is to have 2 different campaign, with smart and fixed bid(on creative balance even or optimize I'm already studying which is better)

On fixed i apply rule for block placement that not perform (i have low budget i block at 1xCPA)
On smart i leave running for 5/7 days before do something, first days usually go bad

I try to test different kind of copy and creative and improve after see the ctr, after that add new copy and image based on the best one.

For the moment i'm a newbie but i try to build my workflow.

Jacopo

---

Hey Jack_l,

I have a few questions regarding the Taboola:

1) Did you select the relevant audiences on Taboola campaign while you start testing?
2)Did you even keep the site running if the LP CTR is very low after 100 clicks, like LP CTR<5%?

Thank you.

---

Hey @fuyuzhelianmeng - So, the way I do it might be totally different than how other folks do it, and I know there's lot of people killing it on Taboola who do things way different than me, so I could be an outlier, but in my case I basically just let the smartbid do its thing and the campaign either works or it doesn't... so for question #1 , I don't target audiences at all I just do 'run of network', and for question #2 , I don't typically look at the LP ctr on Taboola camps, I just let the SmartBid optimize via the CPA amount (since the SmartBid can't see the LP ctr obviously).

I try to run lower payout offers though on Taboola... so if I was running something with a $60 or $100 payout I might optimize by LP ctr and block the really low ones. Or create a conversion event to equal the LP click-through and let the SmartBid optimize based on that.

But yes, if I'm not using a SmartBid-equivalent on a native network and am optimizing manually I definitely block things with super low LP ctr's, so if the average was 40% LP ctr I would definitely block one that got to 100 clicks at <5% with no conversions.

Hope that helps

---

Hey Jack_l, Thanks for sharing this method. As a Native Newbie, I tried many methods people suggested. This one gave me different outcome these days after testing.

A few weeks ago, AM in Taboola suggests set the strict daily budget to at least 300 per day with smartbid. I followed it. Guess what? I spent 1432 USD but only get 4 conversions, -87.04% ROI. Taboola just suck all the budget but give very few conversions.

Your method is different but give the more promising results.

Another question, if you run the the campaign one week and found out this campaign is promising, would you keep the monthly balanced budget or change to daily budget? Continue using the smart bid or change to fixed bid?

Thank you.

---

So again, I know folks doing well on Taboola using totally different methods, but the way we do it is to always leave it on SmartBid, always leave it Optimized, and always leave it on Balanced monthly budget.

I've tried the other budget categories and I just like balanced better, seems to give us better results.

We do change the balanced monthly budget and bid quite a bit. We just divide by 30 and treat it like a daily budget. Although it will range up and down 50% either way typically on a daily basis.

It seems to me the secret is to find the sweet spot where you are bidding enough for the good impressions and where the budget is getting you the max amount of good traffic possible without having to 'overstretch' and send you bad impressions.

If your bid is 6 cents and your monthly budget is 15k (300$ per day) but the campaign is only spending 100$ a day and the CPC is 9 cents, that would suggest your bid is too low and/or your budget is too high, and the smartbid is going to find any click it can to feed you whether the people clicking are likely to convert or not.

On the other hand if your bid is 6 cents and the budget is 15k (300$ per day) and your getting clicks at 5.5 cents each and it's spending 350$ a day, you're in the sweet spot and you can consider raising the budget (if you're profitable) or lowering the bid (if you're not profitable and need to get your cpa lower).

---

Home > Paid Traffic Sources > Native