Home > Paid Traffic Sources > Native

---

Get your highest ROI by running on autopilot: Presenting Auto-Optimization at Voluum (6)

---

Tired of searching for profitable traffic sources? Pausing placements and adjusting bids taking too long? Thanks to years of development (and bunch of smartass data scientists) we’ve solved all your problems at once, by creating our Auto-Optimization feature. It’s a free to use, built-in feature for Native ads optimization in the Voluum DSP platform.

Voluum DSP is a high-performance platform oriented towards getting your native campaigns the best payouts. Thanks to exclusive traffic from Revcontent, MGID, Outbrain, Polymorph, and Liveintent among many others you can easily run campaigns on multiple Ad Networks at the same time. Moreover, we offer a huge variety of optimization features, giving you – the user – complete control over your spending strategy with full transparency in where your traffic is coming from. Additionally, we focus on satisfying performance-oriented advertisers, where we know, each conversion counts. Therefore not only can you fully adjust your campaign targeting to your specific needs, but you can also support your campaigns with our Auto-Optimization machine learning algorithm which optimizes campaigns for you.


So what exactly is Auto-Optimization and why is it so awesome?

The usual journey of any media buyer is a cycle of running a campaign, making mistakes, learning from said mistakes, adjusting targeting options and repeating. This cycle can equate to thousands of dollars spent just to “probe” the traffic and collect whitelists.



The ultimate weakness of this cycle is how much time we waste on the testing process. Each action is supported by hours of checking reports and analyzing which traffic segments bring the best results.

This step by step optimization revolves around analyzing traffic segments which are responsible for campaign performance. An example of traffic segments for a campaign is for example “sites” as displayed below:



In the above diagram, we see a simplified form of the optimization process where the advertiser will pause Site X and Site Z for not delivering optimal results.
In the below diagram, however, we can see extended granularity by adding “Browser” type to each “site”. Instead of pausing all traffic from Site X, the advertiser can pause only the underperforming “Browsers” from Site X and keep “Browser A” which is profitable.

As per example above, we can see that in fact, even though Site X didn’t look profitable at first, it was profitable under a specific browser (combination Site X – Browser A). So, is Site Z in fact not profitable? To check that let’s add once another dimension – device. As you can see in the below diagram, with the additional “Device” dimension, certain traffic from Site Z can be profitable.



In conclusion, pausing Site X and Site Z would have been harmful for the campaign’s performance because there are in fact profitable segments when we go deeper.

What does that mean for you? Our Auto-Optimization feature will identify these granular segments responsible for delivering results.



That means our machine-learning model will decide to bid more on specific traffic segments for you (like Site Y – Browser E – Device A), and will drop bids once it’s certain that specific combinations don’t deliver good results (for example - Site X, Browser B – Device B). As per this example you see only 3 levels of granularity, but our Auto-Optimization feature works on almost 20 dimensions!


Examples of traffic dimensions our machine learning algorithm uses:

Bundle/Site
Widget ID
AdExchange ID
ISP
Connection type
Region
City
Device Model
OS version
Browser version
Browser language
Hour
Country
Creative ID

Also, our algorithm is able to differentiate between traffic responsible for “visits”, “clicks”, or “conversions” so you can set the Auto Optimization feature to favor certain types of user action. If you still want to manually optimize campaigns, our Auto-Optimization feature can work in the background as you pause/adjust your bids.


So, how do we start using Auto Optimization?

Auto-optimization is enabled for all Voluum DSP users without any additional fees. You can select it in the targeting section of the campaign’s settings.



As you can see, this feature can optimize your campaign towards specific performance metrics – iCTR, CPV, CPC and CPA (for CPA optimization – please contact you Account Manager at Voluum DSP).
The only thing you need to specify is the optimal value for your selected metric goal. You can check out a more detailed guide on how to set this up.


It’s always learning, and the more data it will eat, the smarter it gets.

Unfortunately, we are not fortune tellers (not yet). That means to make auto-optimization work, your campaign will need to gather minimum events needed to identify profitable segments of traffic. The first version of the optimization algorithm for your campaign is created after 100 successful events (visits for CPV and iCTR, clicks for CPC and conversions for CPA optimization). Once your campaign reaches the minimum events, your campaign’s Auto-Optimization algorithm will be initialized with your specified goal. Your campaign will take its first optimization steps, but with each added impression – our optimization algorithm will get smarter and make better and more granular campaign adjustments.
Below, you can see a timeline of this process:



Day 2 – 100 events reached, first version of the algorithm appears

Day 2-8 – campaign collects more and more events with different parameters, effectively increasing algorithm accuracy

Day 8 onwards – Auto-Optimization is on full-steam, buying 80% of its inventory from the best-performing dimensions according to the algorithm

Why will only 80% of the traffic be bought accordingly to the optimization algorithm? This is strictly because we know that traffic streams can surprise any buyer, disappearing placements, spikes of traffic can affect your ROI. Therefore the campaign will always buy 20% of its traffic from not verified traffic streams to possibly enrich it’s dataset and learn even more. It’s always learning, and the more data it will eat, the smarter it gets.



So, in the end, what we did was take the classic optimization process of any old media buyer and set it on hyperdrive by adding a few dozen layers of granularity and dimensions to the traffic we optimize.

If you’ve got DSP campaigns running now without Auto-Optimization, stop what you’re doing, log in to your account and turn it on for your campaigns ASAP! Then, take a seat, grab yourself a beer and let our algorithms do your optimization work for you.

Not a Voluum DSP user? Please find more details here.

Finally, we will do our best to share more insights about setting up an optimal goal and real life case studies in the near future.
Stay tuned!

---

Seems to be working really well for iCTR in recent tests.

Does making changes to a campaign, eg setting a new auto-optimization goal value or type, require it to start over with fresh new data?

And does it ever pause things for you entirely, eg a placement, creative, browser, etc or merely lower the bid to a really low point for that combination?

---

Great to hear that!

As for changing the goal value - this is actually what you should be doing. It’s better to start from some lower goal for iCTR or a higher for CPV and then adjust it over time once you identify promising traffic segments with conversions.
Our AMs are always happy to help you with the most optimal goal.


And when it comes to changing the AO model, let me use some example here:

Let’s say you are running CPV AO, but you want to change it to a new CPC or CPA goal.

You got 100 visits (learning period = min. 100 visits for this type of model) so the system already started optimizing and the AO works in the following way:
80% of the data analyzed is based on the optimized traffic and 20% is based on non-optimized traffic, so the algorithm is constantly learning with this 20% of data so to say.

Once you change the model from CPV to CPC or CPA the system will no longer look at the 80% of optimized traffic, but at the 20% that was non-optimized (clicks in case of CPC and conversions in case of CPA) + any clicks and conversions before the CPV was running. Providing that you already have 100 clicks for CPC or 100 conversions for CPA, it starts optimizing immediately.

If there’s not enough events for the new type of model the system will look at:
1. 20% of non-optimised traffic (when the CPV was running)
2. any events before the CPV was running.

It will just limit buying from segments with crazy-high CPVs, CPC etc.

---

It is nice that AO doesn't get reset once you got from iCTR -> CPV -> CPC and eventually CPA BUT there is a major flaw in the algo.

As per the documentation of Voluum auto-optimization gets reset if new creatives are added to a campaign or get paused/deleted by the campaign edit tab (but not if you go to Report -> Creatives and pause them from there).

This means that savvy affiliate/marketers/advertisers get penalised for being creative and coming up with new creatives and angles because every time they do that to a campaign the AO data gets reset and back to square one. This is quite a flaw and until it gets fixed so that adding creatives doesn't reset the AO data gathered it is quite a pointless feature for anybody serious enough about their campaigns.

---

Hey,
that's a fair point and a feedback we already took care of. Adding, pausing or deleting creatives does not re-start the algorithm anymore. We've changed that recently - we just did not update our documentation at that time. Now it's all good. You can check, we've updated this part and all should be clear now.

My apologies for the confusion!
Thanks,
Justyna

---

Hi Guys,


After a short break we are coming back with some real-life examples of campaign results here at Voluum DSP.

First of all, let’s split our optimization based on three types – visits, clicks, and conversions.



Below you can find an explanation for each of them:

1. CPV – focuses on delivering visits to your Native campaign at specific price – for example, if your goal for CPV optimization is $0.20 - our algorithm will do its best to deliver such visits – it will focus on buying from dimensions responsible for visits at that average price.
2. iCTR – This works really well when you want to test out which placements have best visit to impression ratio. The higher you go with the goal, the more demanding the auto-pilot be when filtering for the best placements.
3. CPC – same as for CPV, this optimization type focuses on delivering clicks for a specific price. However, as usual you will have much less clicks than visits – the algorithm will need more time to learn (to gather 100-400 clicks) but once you will have enough clicks, it can be much more accurate than CPV or iCTR.
4. CPA – same as for visits and clicks – focuses on delivering conversions at a specific price. You need to remember that the goal itself is not enough – you need to input a goal below your payout to increase your ROI (so for example, if your payout is the same as your CPA goal - auto-pilot will drive the campaign to be breakeven). This type of optimization usually takes the most time to learn, as having 100-400 conversions on a single campaign level can take awhile. That is why this option is enabled only for clients with assigned Account Managers.

So, now that we’re done with theory – let’s jump into practice – real campaigns at Voluum DSP.

Below you can find examples of campaigns running on auto-optimization:



The optimized segment - where our “auto-pilot” was applied remains profitable and at the same time the not-optimized segment is unprofitable.

Type of Auto optimization: iCTR
Goal value: 0.35



The optimized segment brings higher ROI than non-optimized

Type of auto-optimization: iCTR
Goal value: 0.2


So let’s dive deeper and check out how Auto-Optimization works on a daily basis for the following Voluum DSP Campaign:

Below, you can see a goal set at iCTR – 0.32



Campaign last 7 days performance:



However, to better understand what actually happened, it’s better go on each day level:



As you can see, the campaign has been created on 4th February 2019. However, after campaign creation, our auto-optimization model could not effectively optimize the campaign as there were not enough events to “learn” for the model. A sufficient number of events is usually between 100 and 400 (for iCTR such events are visits) so the campaign started to optimize itself on 4th day – 7th Feb. The goal value given by the advertiser was 0.32 iCTR and therefore, you can see a significant difference between the optimized and non-optimized traffic segment – 0.36% (optimized) to 0.17% (non-optimized) just within the first day.

Our model detected the most promising traffic sources on the combination below:

Application Bundle ID or Site (most promising), Creative ID, Connection Type, OS version and Device Model

Then, it started to apply higher bids for the most promising application bundle ids or sites (or decreasing bids if iCTRs were not met). You can see how adoption of the campaign looked like – in blue – data for optimized segment – in red – non-optimized segment:


Impressions & cost:



Bid change:



So in this scenario – optimizing placements with highest iCTR helped campaign get more conversions. Of course it can not be that obvious (high iCTR doesn’t always mean more conversions) but at least your campaign should limit spend on the lowest iCTR-placements.

Given that all adjustments above were performed without any manual bid adjustments, auto-optimization “pilots” your campaign towards your desired goal without you lifting a finger as soon as it has enough data to start.

To sum up, this example shows how we can reduce hours of campaign optimization via machine learning algorithms. We’ll be adding more examples of other types of optimization in coming posts to give you more insights about how it’s working and why you should try it out. Stay tuned


Waiting for your questions, feedback and of course some comments from those who use it!
Thanks,
Justyna

---

Home > Paid Traffic Sources > Native