Home > Native >

---

4 Figure Profits Running Less than $1 Avg. Payouts on Native [Search Arbitrage] (23)

---

Disclaimer 1: The following campaign was launched in order to better understand TheOptimizer.io user needs with the aim to provide them with added value support in terms of best practices.

Disclaimer 2: The actual ad spends, revenue and accounts used in this case study belong to a friend of mine who was kind enough to let me take over one of his campaigns without limitations and share them here.


Following my previous guide on Running Search Feed Arbitrage Campaigns on Native, in this thread I’m going through the process of running and optimizing native to search feed campaigns.

In order to make this case study/guide more comprehensive, I will break this down to 4 main topics. This way, even those who are not familiar with this kind of campaigns can understand and replicate the logic.

• Finding potential good performing feeds/verticals
• Building a solid tracking setup
• Launching and optimizing the campaign
• Results and Key Takeaways

Let’s start!


Finding Potential Good Performing Feeds/Verticals

As always, competitive intelligence tools are extremely useful to find good offers/feeds, as well as getting the right ideas about what images and headlines to use. So before launching this campaign I went to Adplexity Native and ran a query with the following criteria:

Search Query: In Ad Info (enter keyword) | Results of at least 5 to 30/45 days | Country: US | Sort by: “Running longest” & “Received Most traffic”.



The logic of searching for results based on the ad keywords stands behind the fact that in most of the cases the keyword or vertical of the feed is present right in the ad headline. So, to get a clear idea of the potential feeds you can promote, check first with your search feed provider account manager for a list of the top performing feeds, the see what looks more interesting for you.

Knowing that I was going to promote a 2-Click flow Google search feed, I went over an extra step in order to make sure that the search volumes and competition were high enough to scale the campaign. Signed into my Google Ads account, headed to Tools > Keyword Planner and after selecting the country (US), I added a few commonly used search terms related to SUVs.



What I was most interested to see in this report was the Avg. Monthly Searches, Top of Page lower and higher Bid.

Having decided with what search feed to go, it was time to consolidate the tracking. Part which I consider to be pretty important, because with this type of campaigns we have a delay on the confirmation of conversion revenue ranging from 24-36 hours. Check this thread for details.


Building a Solid Tracking Setup

Considering that most search feed providers nowadays are capable of reporting an estimated revenue for the conversions in near real-time (this varying from every few minutes up to every hour depending on the provider), I wanted to use this estimated reporting to my advantage. Instead of just placing the destination feed link directly on the ad destination (like most people do), I passed all the traffic through a tracker (in this case Voluum). And here’s why this was pretty important step for me:

• Both the ad network (Outbrain) and the Search feed provider had over an hour of delay in reporting the performance on their dashboards.
• Using a click tracker, I could see what was converting or not in almost real time and at the same time use this information for optimization proposes.
• In case I can find a similar feed on another provider, I can easily split-test two or more providers, or even replace the main offer without having to touch the ad destination link risking to cause any delays from the re-approval process.

Outbrain Config:

On Outbrain I created an event-based conversion called “search” assigned to the search category found under Outbrain's predefined categories.



Note: I think assigning the conversion event to the right category is pretty important, as it will better engage the algorithm to optimize towards the correct conversion event you are looking for.

Voluum Config:

On Voluum I created a separate Outbrain configuration specific for this type of campaigns, this way this would not mess up with the other conversion events or send inaccurate info to Outbrain. As you may already know, their tracking pixel tends to be quite unreliable and under-report conversions.



Also, as you can see from the above screenshot, instead of using a standard postback URL for this traffic source config, I used an event-based postback. This way having an event-based postback setup on the search feed provider side, these events can fire smoothly to Voluum, then from this last one to Outbrain. This way when I get the confirmed revenue for the conversions, I can effortlessly upload them as main conversions without risking to inflate the conversions count on Outbrain. Inflated conversions should definitely influence how the traffic source algo behaves - never had the courage to test it on something that was working or showed promising results.

Launching and Optimizing the Campaign(s)

At first, I started this campaign with manual optimization in mind, somehow as a challenge for myself, but soon turned out not to be the best choice. This because the campaign was targeting US and I am based in EU, so monitoring and optimizing it manually would have been "mission impossible" for me. Especially with my current daily schedule where 36 hours a day are not enough.

Anyways, I went ahead and created the campaign on Outbrain as follows.

Targeting: United States; Desktop Only; Day parting enabled
Bid: Starting at $0.06 CPC
Budget: $20/day (1st day)*, then slowly increase | Pacing: Accelerated
Conversion strategy: Semi-Automatic [60% Optimized] / Optimize for Conversion – Search



A few notes on my current campaign setup here:

As you may see from the above screenshot, the campaign was set to use Semi-automatic conversion bid strategy for the "search" event . Also it's worth noting two additional settings here.
The first one is the "Max CPC limit 100%" - This means that until I was ready to bid individually on specific sections, Outbrain's algorithm would be playing with my section bids from -99% to +100% in relation to the campaign level bid in order to get the desired results.

The second, enabling an A/B test experiment. I decided to go for 60%, which means means that if the conversion mode would result in a lower CPA compared to the manual mode, 60% of the traffic of this campaign will be put under conversion mode. In other words, Outbrain's algorithm will do it's best to keep 60% of the campaign's traffic under control and deliver the results I was looking for.

Creatives:




Important Note: As you may see from above, I allocated only $20/day for the first day on propose. Knowing that Outbrain wouldn’t give me a chance to block the usual high volume / low competition publishers or sections for the first couple of hours, it was better to just burn that $20 and call it a start after that.

Once the first amount of allocated budget was consumed and I was able to start blocking any of the pubs I didn’t want to get traffic from, I let the campaign consume its daily budget for the first three days while blocking any poor performer daily.

Right at this point the campaign was doing around -60% and was showing promising results. But because I was targeting US, I wasn’t able to follow through, so I connected both Outbrain and Voluum to TheOptimizer Native in order to give the campaign a spin. As the saying goes, work smarter not harder.

The initial optimization rules were pretty basic. Practically I was excluding any section that had generated 10 clicks and no conversions. This may sound a bit too aggressive, but based on the feedback I received from the account manager, I had to reach at least 15% conversion rate to make this work out.

After about a week or so, running on this simple optimization approach the results were not there yet. Indeed, I was close to breaking even, but as you may know Outbrain has a default block limit of 30 Publishers and 100 Section per campaign – I had already reached that, so I was forced to pause the camp for a couple of days.

It’s worth noting that the account was recently created, had no ad spend history of any kind, no account manager, or anything that might have helped me in regards to Outbrain’s decision on lifting the default block limits to keep the current campaign going.

Surprisingly it turned out that the limitations were lifted by just submitting a support ticket. Obviously, I explained what I was running and briefed on my optimization strategy. In their reply, they confirmed I had the limits pushed to the max. If I’m not mistaken 1000 Publishers and 1500 sections.

Now back to turning the campaign ON.

During this time, I had also collected enough data to start creating some more “intelligent” auto-optimization rules.

Instead of creating generic auto-optimization rules, I went ahead and spent some time looking at my campaign stats, or to be more precise, looking for specific patterns. This way I would be more confident with my optimization approach since this way my optimization would be driven by the data I had collected instead of just throwing rules that could negatively impact my results.



From this moment on, the campaign was virtually running on auto-pilot.

All I had to do was to manually download revenue generating click ids form the provider in a CSV format, then upload them on Voluum. Also, the stats of the confirmed revenue were scheduled to be automatically updated on TheOptimizer side using Automatic Updates.




Campaign Results and Key Takeaways

As you can see from the below graph, the results of this campaign ware pretty satisfying.




Key Takeaways:

• One doesn't necessarily need to have a high payout offer in order to generate profits on Native - you can generate profits even with a payout ranging from $0.3 to $1.5 a conversion.
• Always invest time and efforts on building a good relationship with your reps. I must say that the account manager played an important role in the success of this campaign. Don't skip this.
• Focus on understanding a traffic source behavior as much as you can, especially when it comes to how it reacts when playing with bids and budgets.
• The best and proven way to properly optimize a campaign is to rely on campaign's data. Data-driven decisions tend to be more reliable in the long run - Don't learn this the hard way!
• Also it's best to adopt your optimization strategy based on the country / vertical and device targeting. Although this applies to all campaign types, for this ones, I think this is a must.
• Use a click tracker whenever possible. Even for experienced media buyers running arbitrage (especially native to search) a tracker will give additional insights you cannot get elsewhere.
• Native to Search Arb Specific - It is really important to be able to exclude at least the current and previous day results when working with ROI based rules. Things can change by a lot in 24 hours, so it's worth the wait.
• Unlike Facebook, with native you can double or triple your campaign daily budget without risking to screw things up.

That's all for now folks! I hope you enjoyed it!

In the reserved section below I will share some additional details on the automatic rules used in this case study, how I came up with them, and how I scaled the initial campaign.

As always don't be shy to ask any questions you might have in the comments down below.

Cheers, Losid

---

Now back to the optimization and scaling details.

Optimization Logic and Auto-optimization Rules

My best bet to come up with the right optimization strategy and rules was to look at my campaign stats. Having the cost, pending + confirmed conversion and revenue (excluding the current and previous day) in TheOptimizer, I just had to select a date range with enough data available to go over the stats.

Even though I had a rough idea on how I was going to break down performance levels and optimization actions (see graph below), I still wanted to make sure I was in the right direction.



Using column filters on TheOptimizer I was able to quickly verify my optimization approach and adopt it to the above logic.




Publisher level rules:

As you can see from the rules list my main optimization focus was on the Section level. For Publishers I had a pretty simple rule that was taking care of the budget wasting publishers, just in case they were going to result in less than -60% ROI.
Important note: Because Sections are part of a main Publisher, I didn’t want to risk blocking any potential publisher that could’ve had over 50% of the sections in positive ROI.

Important note: Because Sections are part of a main Publisher, I didn’t want to risk blocking any potential publisher that could’ve had over 50% of the sections in positive ROI.

Section level rules:

Here is where I applied the logic visualized in the above graph. I was practically cutting anything that had a low CVR and ROI worse than -50%, while tweaking bids for when the ROI was from -30% to 0.

You may say that there’s a gap between the two ranges, but because I could never know what the conversion payout might be for an additional conversion, I thought it was best to risk it and not over-complicate my approach.

Here's the full view of my actual rules for blocking sections:



These two rules are identical in the condition settings, however one of them is selectively excluding sections from the block actions. These are sections that have showed good results and I wanted to give them a bit more clicks to express their potential.



These two other block section rules, were taking care of sections that were excluded from the aggressive section blocking rule providing full coverage on the thresholds I had determined from analyzing the stats.

Here's the full view of my actual rules for changing section bids:



For sections that had a ROI between -20% to 0% I was updating bids everyday at 10PM CET (an hour I was sure the previous day revenue was updated). Then for sections that had a ROI between -30% to -5% and a CVR grater than 8% above $0.9 I was changing bids every 2 days based of the last 2 weeks performance. This way I was able to adopt bids to my Avg. EPC in a wider date range.



Along with the above rules, I had another simple rule to Unblock Profitable sections. This way I could easily turn on any potentially profitable sections that would have been previously blocked for missing conversions and revenue.


The Scaling Process

For this type of campaigns the scaling process isn't necessarily vertical. Meaning that you shift the daily budget from $50 to $200. Even though the risks of having the campaign screwed up by this change isn't present, it is still best to distribute the spend into multiple campaigns - scale horizontally.

Doing such a horizontal scale, gives you more chances of finding a better match between your "campaigns + ads" and the traffic source optimization algorithm. It rarely can happen that two or more identical campaigns end up competing with each other - you would need to have 50+ identical camps to risk that.

Also, the original campaign that started targeting all devices in one place, was then split to specific device targeting per cloned campaign Desktop, Tablet and Smartphone.

Hope the above details help on getting a better idea on the optimization logic.

---

I'm going to sticky this thread as well - it's a "must-read" for anyone planning on giving native search arbitrage a try!

Automation can save so much time (spent on checking stats constantly) and money (by not "overrunning" - continuing to run something past the point of cutting).

Automation can also save a person's mental health - in not having to worry about campaign performance taking a dive (e.g. while we're sleeping) and zapping the budget when we're not checking stats.


All the "Key Takeaways" are gold! This one though really surprised me.

Native traffic isn't cheap, and typically you'll see higher-payout offers and products being promoted. When I was running lead gen offers for that newbie Outbrain tutorial, I was surprised to find that offers with payouts of under $5 could work for native.

And here you are showing us that payouts as low as $0.30 can work - mind officially blown!

With this method, we no longer need to spend 4-figures a day to optimize a native campaign (although we could!), because it takes so much less money to cut a section (or tweak its bid) than if we were promoting, say, a nutra offer with a $xx payout.

Many, many thanks for your posts @platinum!



Amy

---

Love it @platinum! Really fascinating seeing how you do your Optimizer rules too...

And how you use Google Ads for research too... that's one thing we haven't done at all I'm ashamed to say...

But yeah, we ran SUV Search Arb too recently - on Taboola... were killing it in November and early December but epc's went way down in January so not we're only spending like 20$ a day breakeven after having been over 500$ a day at a profit for awhile in Q4... guessing it has to do with what the car companies are bidding on search clicks right now, hoping it will go up as we get further into the quarter...

Search arb seems very hit or miss (I'd say 90% of the verticals we've attempted have failed) but when it works it really works, and I love that its so much easiest and cheaper to test vs ecom and lead-gen campaigns...

---

Very nice case study platinum! Thanks for such a detailed post with all the settings!

---

Thanks a lot for your case study!

I'm still waiting for Theoptimizer to implement such a "24 hours" rule (or make it variable, say 1/3/12/24/72 hours), the current setup (today or today+yesterday) is not accurate.

---

This is incredible. The creatvity combined with the data and optimization techniques are fascinating!

Some dudes are just on another level. @platinum is one of those dudes.

---

Thank you for the love guys!

I've also added some more details in the first comment as promised. Especially for the rules because a good part of them had settings not visible in the rule list.

Furthermore, the approach I've used in this case study can be easily replicated for other campaign types as well. It just needs to be adopted to the payouts, conversion flow and conversion rate of the offers.

---

Actually January tends to have a slight drop in the performance of such campaigns, however, usually it can be fixed swapping the keywords on the backend of the search feed. That is why I emphasized how important the role of the account manger is.


Indeed some providers stronger than others for specific verticals, I think it's worth testing more than one provider. Also, if you're running mostly US traffic then its a bit harder to get the camps in shape, because the competition is pretty high. But other Tier 1 and 2 geos can work pretty well. Again I can't stress enough how important are the keywords behind the feed.

---

@platinum How do you deal with the search feed networks not passing Revenue in their postback for the real time conversions? (They are stating that Google doesn't fire it to them). That's the case for System1 & Domain Active at least.

Have a great weekend!

EDIT: some bug caused me to post this a bajilion times, would love to if some moderators could assist

---

@roiter123 Please stop spamming, this is my final warning

Thanks for bringing the thread up again.

---

@roiter123 they don't pass real time conversion reporting, however they can pass intraday conversions using postbacks. Just explain them what you are looking to achieve (ideally send them a ready made postback URL) and they will be able to assist. Never had any issues getting this done, so can't say much on the topic.

---

Thanks for the reply @platinum!

They are stating that they can't fire intraday postbacks with revenue in them... Also system1 are stating that sending clickid via the sub_id token would eventually become a problem when running on high volume (data pipeline congestion issues).
So either I'm doing something reallyyyyyy stupid or seems like I would have to work with no revenue conversions (intraday) and then upload rev to the tracker after around a day (without firing to the TS).

Will the TS algorithm be less effectively profitable when no revenue is being passed?

---

Just ignore same day revenue, it's an estimation anyways. If you can get conversions reported you're all fine. The traffic source won't be affected at all for not sending revenue to their conversions. Put a big 0 in the postback payout parameter and you're done

---

Thank you man, this helped me a lot.

---

Have you figured out this question? because they tell me the same thing. They tell me that I cannot use in sub_id token unique clickIDs. And this, in turn, means that I will not be able to use the tracker in work.

---

I'm simply running with System1 without revenue... @jack_l told me he doesn't use a tracker at all and just passes postback to traffic sources and optimizing from their platform (not sure if he's passing revenue to the TS though, probably optimizing based on average CPA and average payout, CR).

---

wow losid, you always blow me away with your advanced content, love it.

will have to try this looks like a lot of nerdy fun goodness

---

Yes... we never used a tracker with System1... we just passed conversions straight to Taboola/Outbrain/etc and optimized inside the traffic source dashboard... we did not pass revenue (don't think its possible)...

We did build a custom reporting thing in Tableau for like 2000$ of developer time though... that tracked all the data on payout/epc/average epc per campaign/epc per hour of day/day of week/etc... to be honest I don't think it was worth it though... maybe if you're spending xx,xxx per campaign somehow, but even then, I think those variables are constantly changing anyway based on what the people buying the search clicks from google are doing... so its hard to optimize based on that data anyway..

I would just pass conversions to the traffic source for optimizing sites/ads/etc, and then also track your campaigns in Excel at a 24 hour lag, by subid, using the daily report they send you each day...

So...

SeniorHousingAU_Taboola_AllDevices
Date - Conversions - Revenue - Spend - Profit/Loss
05.01.2021: - - - -
05.02.2021: - - - -
05.03.2021: - - - -
And so on...

Then you can see which campaigns are making money and which aren't... and you can always go back and check on them in 6 months or a year if you want to see if there were any patters (x vertical does better in summer, y does better on weekends, etc).

---

Hey @platinum, in regards to your optimization rules - can you please explain this?

---

And... what's the point of having those restrictive conditions from the bottom up like EPC being above 0.07 or CVR being above 8%? What's the point of those restrictions?

---

Hi @platinum,

I was triggered by search arbitrage after seeing a video of you on Youtube, didn't realize you were also on STM.


Search arbitrage looks like a really interesting niche of AF. Years ago I managed a portal site that was a Google Search Partner, managed to direct link to hot expensive keywords in newsletters and earned a lot of cash that way for that company... kind of search arbitrage :-)
And been doing Search Ads for like +15 years so I favor Search over Social.

Have a few questions:

- any thoughts on the budget needed to get this started? Not to get to XXXX a day, but like to get profitable and get XX of maybe 1XX a day.

- When I do some (manual) spying I can only see like keywords on cars and kitchens ... I guess there is a whole range of keywords. Care to list some ideas?

- I suppose you're free to choose your keywords (not the search partner network)?

- Do you have to play with exact/phrase/broad? Or is this just the keyword wathout matching options? Makes a big difference on Google Ads.

- the landingpages of the networks are 100% fixed? Or can you choose from a few landingpages? Or is this not really important for the conversion?

- When I look at publishers with native ads I'm only seeing like a small percentage of them being search ads. Most of them are 'regular' native ads. So I was wondering about the volume of these ads? But I guess you can do these in many many geo's...


At the moment I'm also seeing a lot of search ads with really crappy translation to my 'native language', so I guess there is an opportunity to at least win the bettere copy contest (again not sure at this moment if this will affect the conv. rate).


Hope you don't mind all the 'noob' questions


Thx for any insight

---

Hey @s14b23, let me answer your questions below

In the last couple of years, search arb has become way more competitive than it was before. Not that it's impossible to dominate (automation helps a lot with that), but it's not as easy as it used to be.
On top of the competition issue, keep present that search feed providers still require some decent amount of ad spend, so I would say around $300 - $500 a day should be fine to start with.

I don't think manual spying is the way to go for such ads. You can catch a few of them, but keeping in mind that a native ads campaign gets traffic from 2000+ publisher sites, it's impossible to see everything manually. It's better to go for spy tools, they will show more reliable data, especially the "days running".

Actually there is no control over the keyword match types. It's all managed from the search feed provider as well as overwritten by big G most of the time.

Landing pages are also managed by the search feed provider. They are constantly testing new landing pages and design variations that may yield better results. As long as the traffic is legit and compliant, they are the first ones interested in getting the best landing page variation in the eyes of your visitors.


What you see as "Sponsored search results", "Search Ads" or whatever other naming on them, that is automatically added by the traffic source (ad network). Some of them add this text, some don't.


My advice would be to start with low competition GEOs first. It's okay get started with your own country and other ones that speak the same language as yours, but always keep present the RPMs and overall performance of a geo. Your search feed account manager should be able to assist you with better up-to-date details on this.

---

Home > Native >