# Voluum Documentation

### Auto-Optimization

Voluum DSP DSP Campaign Optimization

In Voluum DSP, you can apply a machine learning algorithm that automatically optimizes your campaigns towards your specified performance goals:

• iCTR (Impression click through rate) which optimizes the number of clicks in your banners

• CPV (Cost per visit) which optimizes the cost of visits

• CPC (Cost per click) which optimizes the cost of clicks on the landing page; this option is only applicable when a lander is added to the campaign’s flow

Once your Auto-Optimization enabled campaign is activated and gets traffic, the machine learning algorithm will start. The algorithm begins with a learning period where it acquires data about what type of traffic performs for the campaign. The learning period continues without optimization and ends once enough data deemed statistically significant is gathered. The minimum data for optimization to begin is 100 visits (iCTR, CPV) or 100 clicks (CPC).

Once the learning period ends and the optimization algorithm starts, your campaign dedicates 80% of the daily budget to buy traffic deemed optimal by the algorithm and the other 20% to continue testing for new traffic.

In order to set an appropriate optimization goal, you should adjust it gradually. First of all, take a look at the value of iCTR, CPV, or eCPC without optimization. You can easily find it in the campaign's report. This value gives you an insight into your current campaign's performance and how you can improve it with Auto-Optimization enabled. Then, edit the campaign you want to auto-optimize and set the optimization goal.

While running a campaign with Auto-Optimization enabled, you can perform the following manual adjustments:

• Pausing ⇄ resuming sites, applications, creatives, operating systems, etc.

• Changing a micro bid value

• Changing campaign's targeting options such as: bid, unlimited / total / daily budgets, spend strategy, max daily spend per site/app, max daily spend per ad exchange, frequency capping, start and end dates, day parting, ad exchange, device type, operating system and its version, connection type, source type, mobile carrier

All above listed changes will not reset the optimization findings for a given campaign.

If you want to turn on Auto-Optimization for newly created campaigns, we recommend that you verify the campaign's performance for a half of day without Auto-Optimization enabled to determine that your goal is reasonable.

#### Turning on Auto-Optimization for Your Campaign

2. Go to Campaigns. The Campaigns view will show up.

3. Edit the campaign that you want to auto-optimize.

4. In the BASICS tab of the campaign's form scroll down to the Auto-Optimization section and click the Add optimization link:

1. From the drop-down menu select one of the optimization goals: iCTR, CPV, or CPC.

2. In the text field next to the selected goal, provide an amount of money that you want to set as a threshold for your campaign.

5. If required, change other settings of the campaign.

6. After providing all changes, click the Save button. The campaign form closes. Your campaign is going to be auto-optimized within the set goal.

### Caution

Voluum Note: Keep in mind that it is not allowed to modify the below listed targeting options during campaign life-time:

• Creatives

• Countries

• States / regions

• Cities

• Retargeting lists of audience segments

• App / site retargeting lists

Adding any of these options will result in resetting the Auto-Optimization model findings and restarting the campaign learning period.

#### Analyzing the Campaign Performance

Once Auto-Optimization is turned on, you can track the campaign's performance in reporting:

1. Go to Campaigns. The Campaigns view will show up.

2. Select the campaign and click the Report button in the menu above the report. The campaign's report opens in a new tab in Voluum.

3. Go to the Optimization tab: In the newly displayed table you will see the performance of the traffic bought organically, Non-optimized, and according to the model recommendation, Optimized.

4. Now, you can compare the results for both types of sources.