IFTTT – Creating a Data-Driven Campaign Optimization Strategy for Native Ads

Data-Driven Campaign Optimization Requirements

First things first, no matter how many steps your campaign funnels have, you will absolutely need to track your visitors’ journey from A to Z. This helps both measure the results of your campaigns as well as collect meaningful data crucial to creating your campaign’s optimization strategy.

So, to track our campaigns and collect the data, you will definitely need a tracking system in place. Here are some options you might want to utilize.

• Traffic source pixel implementation (a must-have)
• A tracking platform able to track the full path of our funnels
• Google Analytics (able to provide demographic info about the campaign audience)

For starters, the traffic source pixel is an absolute must-have! This is because, if you are running native traffic through Taboola, you are going to get better results by having Taboola’s tracking pixel installed on your promoted pages vs. not having it. The presence of the pixel will enable Taboola to capture visitors’ journey and use that information in their built-in optimization algorithm (aka. SmartBid) to improve your results. 

In addition to the traffic source pixel present in your funnel pages, it is highly recommended to use an external click tracker to accurately track the performance of your campaigns outside Taboola or any other native ad network for that matter. Besides the benefit of tracking your funnel visitors journey, you will be able to get additional valuable insights for your campaigns and benefits like:

• Landing page click-through rate and/or funnel step success rate
• Visitors geographical insights (country, region, city).
• Visitors device model, connection type, display language, etc.
• Server-to-Server conversion tracking.

If you can’t add a click tracker in your set of tools, then Google Analytics might be just fine, just make sure your event-based goals are properly set up, this will avoid inaccurate tracking that can lead to erratic decision making when optimizing your campaigns.

Once you have a well-tested tracking implementation in place, you can move forward with the next steps.

Campaign Planning and Data Analysis

When it comes to planning most of the time the focus goes to choosing the right offer, images, headlines, and ad copy; but part of the planning process is also the process of analyzing and calculating a few preliminary KPI estimates in regards to your campaign.

So, whether you are an affiliate marketer, agency or product owner, you must understand or better calculate the average success rate of your product (aka. offer) using previously collected data from various traffic sources. If you don’t have access to such information you can rely on industry-specific benchmarks you can find on Taboola’s blog or other data research platforms.

Campaign Planning

Using previously collected or industry-specific data, you can easily estimate the minimum thresholds of some vital KPIs of your campaigns right before even sending traffic to them. And for native advertising campaigns, the LP CTR is one of the most important metrics you should care about. This metric will tell you right from the beginning whether a publisher site has any good chances of becoming profitable or not. 

To estimate the average LP CTR you will need to get close to break even you will need to use the following formula.

Example:

Let’s assume you are promoting a campaign with the performance stats.

Payout: $ 25        |     Conversion Rate: 9%

Ad Spend: $500    |     Campaign Bid: $ 0.65

Using the above formula to estimate the optimal LP CTR, we should end up with an Est. LP CTR of 28.89% for a conversion rate of 9% from the landing page clicks.

As such you can create the following optimization rule:

Obviously with an average amount of 3000+ publisher sites per campaign it is hard to follow through and check each publisher site manually, so a better way would be to create an auto-optimization rule in TheOptimizer Native like the one below:

The above automatic rule will follow the same “if this then that” logic and can run every 10 minutes checking each campaign publisher individually while using both traffic source and tracker information.

The above formula can also be used to start blocking poor-performing publishers even before getting any conversions to avoid huge losses, however, let’s move to the next step for a more feasible approach.

Data collection and Analysis

When starting to send traffic to your campaigns you will slowly start to collect actual data about them. This way you will be able to easily compare your preliminary estimations to the actual results generated from your campaigns. 

One important thing to keep in mind is the fact that you need to let your campaigns run for a while in order to collect meaningful data. Jumping into decisions right after the first or second day of a campaign going live, won’t get you anywhere. As a rule of thumb, you should let a campaign run for at least 5 days to a week, then dive into your data to identify money bleeding patterns.

During the analysis process, it is very important to look for specific patterns amongst publisher sites. You will notice various patterns among different publishers, and that’s pretty normal. Some publisher sites will show a lower than average LP CTR but with higher LP CR, while some others will show higher LP CTR with lower LP CR. There’s nothing to worry about here, it is pretty common to have such results. For this very reason, it is important for you to dive into your campaign statistics and analyze KPIs like ad-level click-through rate (CTR), pre-sell page engagement rate (LP CTR), ad-click to conversion rate (CVR), pre-sell conversion rate (LP CR) and Avg. CPC.

Once you have determined a set of specific patterns you can move to the next step.

Data-Driven Campaign Optimization – Building Your Optimization Strategy

Now to the most important part. Let’s dive into the three main optimization interventions that will help you meet your goals.

Blocking Bad Publishers

The first thing you want to do is to start blocking poor-performing publisher sites. After all, there’s a good reason why Taboola has an ample limit of 1500 publisher site blocks per campaign (and an additional 1500 on the account level). 

Using the same formula used earlier to calculate your estimated LP CTR, now you can use your actual campaign data for determining your LP CTR publisher block thresholds. The process of recalculating the average threshold is important because it can turn out that your offer shows a higher conversion rate from the estimated one, which leads to a lower LP CTR range requirement. Alternatively, it may show a lower conversion rate from the estimated one, therefore increasing your minimum LP CTR threshold.

Using the isolated patterns mentioned earlier you will also be able to address different publisher groups based on their results.

Changing Publisher Bids

Another important factor in the success of your campaigns is your bids. As a good rule of thumb, it is always best to start with an average higher bid then slowly drop it down to the optimal levels. Although this strategy might look easy, sometimes you have to be careful not to throttle it too early.

Here’s what you need to keep in mind. Although you have started with an average high bid, Taboola’s algorithm will make sure you’re not overbidding on all sites. This means that it will place a lower bid to those sites with low competition and stay within your limit range on those with higher competition. Sometimes it can also go above your current bid, but that’s part of the process. High reputation and competition publisher sites tend to require higher bids because they convert better. 

On the other hand, if you want to throttle your bids on specific sites you can do it by lowering or increasing the bid on them. To do that you’ll have to rely mostly on your results, particularly your EPC and ROI. As a good rule of thumb, you should aim to have a lower average CPC compared to your EPC.

Using the same “if this, then that” condition-based logic, you can start tweaking your bids to match your actual EPC or take control over the bids that are placed on a specific site. Here again, you have to keep in mind that you will need a reliable amount of conversions or traffic before starting to change bids on a publisher site. Doing it too early may turn out to ruin the performance of that specific publisher site, so better wait for some significant amount of conversions before doing that. Here’s an example:

It’s important to emphasize that when changing bids, you need to be aware of the bidding strategies that Taboola provides.

Option 1: Leave publisher site bids on SmartBid mode and lower or increase the bids to match the performance of that specific site.

Option 2: Switch the publisher site bidding strategy to Fixed and take over the bid control of that publisher. When doing so, be careful to do it once you have enough data and conversions to back up the bid levels you are trying to control.

When running with Taboola keep present the following bid strategies: SmartBid and Fixed Bid

The difference between the two is that when running on SmartBid, Taboola’s algorithm will play around with your bids in the range of -50% up to 200% of your current bid. That means that if you are bidding $0.60 CPC your cost will vary from $0.30 up to $1.20 CPC.

Split-testing Ads

Ads, the combination of headlines and images are your first impression to your targeted audience, therefore it is important to thoroughly test ads and make sure your campaigns run with the best ones. They also influence a lot when it comes to your average CPC, so make sure to use appealing images and headlines. This step is vital for your campaigns.

Native ad networks like Taboola, offer you the ability to either optimize ads on your own or let their system decide which ads to focus on. But for split-testing proposes, you’d better let the ad traffic allocation to even, this way you will make sure you are properly testing them and then focus on the right ones.

When utilizing “Optimized” creative traffic allocation, Taboola will split-test all ads in the first few days of a campaign, then based on their results, it will pick one and drive most of the traffic to the winning ad. Please note that factors like CTR and Conversions are the most influential ones.

When utilizing “Even” creative traffic allocation, Taboola will make its best to serve all ads equally. Using this traffic allocation mode you can easily split-test ads and decide for yourself which one to block and to keep based on the results of your ads.

Here as well, you can use “If this, then that” conditions like the one we used to block publisher sites, but this time for pausing ads.

Improving your pre-sales page 

Last but not least, this is the most overlooked part of campaign optimization. Most advertisers take for granted that their landing pages are the best possible they can have for that product and often fall victim to a poor-performing landing page. 

As a matter of fact, optimizing your landing pages for better results is part of the process too. Sometimes introducing small changes like a different headline, hero image, interactive or personalized callouts, can easily lead to better results. Just make sure to test each change one step at a time. Mixing things up can easily tear the campaign apart and give the opposite effect.