Introduction

Following the launch of Smart Bid, our new pricing model, and our previous blog post that explains how this new feature works and what are its benefits, we would like to showcase how an advertising campaign evolved over time, using this feature. This case study is based on the experience of one of EXADS ad network clients, with one of their advertisers. To be specific, we will see how by using Smart Bid with the Native ad format for a range of Dating offers, the advertiser was able to reach their Target ROAS (Return on Ad Spend) of 120%.

Campaign Overview

• Offers: Dating
• Ad format: Native
• Advertising Payout Model: PPL
• GEO: US

Native: An attractive Ad Format, complex to master

The advertiser's objective was to identify new traffic sources that would generate additional profit using the Native ad format. For this, the advertiser launched a Native campaign in the US. The advertiser was rotating several Dating offers with different PPL values, therefore the conversions had dynamic values. This was an important factor in choosing to use a Target ROAS bid strategy instead of Target CPA strategy for the Smart Bid campaigns.

• Objective: 120% ROAS
• Main constraints: Offers capped at 200 leads per day.

Initial Campaign Setup

Country: US
OS: All
Device Type: Mobile
Traffic Type: All
Zones Targeting: All
Categories: All
Frequency Capping: 1/24
Smart CPM: $0.20
Daily Budget: $800

First-week results 

One week after the initial campaign was launched, the advertiser informed the ad network of its poor performance. Moreover, they were considering abandoning this format altogether as it was proving to be "unmanageable due to a large number of ad zones". At that very moment, the campaign had a ROAS of 48% (ROI -52%).

The first step taken by the ad network was to analyze the campaign's statistics. This analysis revealed several points that contributed to these poor results:

• A high number of zones: A format such as Native represents a real challenge for any media buyer. The manual configuration of an adequate bid for each zone can be time-consuming and tedious, especially given the large number of native zones offered on the ad network.
• Poor distribution of spending between zones: When launching a test campaign targeting thousands of zones, the main objective is to identify high performing ad zones that could generate profit. However, 90% of the campaign’s spending was concentrated on four zones and  the campaign reached its daily budget after 10 hours. These high-volume zones were holding back the campaign and the campaign needed to allow for more spending on other low-volume zones that were potential sources of profit. 
• Underperforming dimensions: Additionally, some browsers and devices were impacting the overall ROAS of the campaign.

In the table below you can see the advertiser’s results from their initial targeting and bidding strategy:

The second step taken by the ad network was to tackle each of the points mentioned above. This called for a change of strategy.

Smart Bid : The way to go

The Native ad format is indeed a format that offers thousands of zones which inevitably requires a large amount of time in its management: setting up a correct bid per zone, blocking unprofitable zones,etc. The ad network recommended switching to a Smart Bid payment model. This would fully automate the campaign according to the advertisers target ROAS. For this strategy, Target ROAS was suggested rather than Target CPA, due to dynamic value for each conversion.

The ad network also advised the advertiser to create two campaigns, to better distribute the budget and the daily cap of the promoted offer. Campaign 1 targeted the four zones that reported 90% of the spending from the initial campaign, and Campaign 2 targeted all zones except those from campaign 1. The objective of this strategy was to ensure that the offer cap remained allocated to the second campaign targeting lower volume zones to ensure proper testing. 

Using the data acquired during the initial campaign, the advertiser set more relevant targeting for Campaign 1, by removing underperforming devices and browsers. In addition, the advertiser removed the worst-performing variations from the campaign, which maximized the chances of reaching the target ROAS that the advertiser wanted to reach.

For Campaign 2, the recommendation was to leave the targeting settings as they were on the initial campaign. The reason for this being that the spending on the initial campaign needed to be more significant in the targeted zones to be able to draw relevant conclusions.

For both campaigns, the Target ROAS was set at 100%, which means that the initial objective was to reach the break-even point

Please note: It is crucial to have a Target ROAS that is not too ambitious at the launch of a Smart Bid campaign to avoid over-optimizing the campaign to the detriment of volumes. Of course, the Target ROAS can be adjusted as the campaign progresses and according to its initial results.

Setup Campaign 1:

Country: US
OS: All
Device Type: All except Huawei, Vivo, and Oppo
Traffic Type: All
Zones Targeting: 4713984,470346, 4945834,4461824
Categories: All
Frequency Capping: 1/24
Bid Strategy: Target ROAS: 100%
Bid Optimization: Maximize Volume
Daily Budget: $400

Setup Campaign 2:

Country: US
OS: All
Device Type: All Mobile
Traffic Type: All
Zones Targeting: All except 4713984,470346, 4945834,4461824
Categories: All
Frequency Capping: 1/24
Initial CPM: $0.20
Bid Strategy: Target ROAS: 100%
Bid Optimization: Maximize Volume (week 1-2) Maximize Target ROAS (week 3)
Daily Budget: $400

First week results using Smart Bid

After one week, campaign 1 reached a ROAS well above the target (138%). There was a significant increase of the CPM from $0.20 to $0.32; this means that the Smart Bid identified performing segments on which it positioned itself with an aggressive bid to maximize volumes; by contrast, we notice segments on which the campaign did not deliver any more traffic after the first day, due to a ROAS well below the target. 

Campaign 2 recorded a ROAS of 71%, below the advertiser's target. However, the results were more than encouraging, as the zones that spent the most and collected significant data, were on target. The overall performance was impacted by the large number of zones that did not have enough data for the Smart Bid to take over, so they were bidding based on the initial CPM configured, equivalent to a regular Smart CPM campaign. This campaign simply needed time to allow the Smart Bid to identify more optimizable segments.

Second-week results using Smart Bid

At the end of the second week of testing, Campaign 1 recorded a target ROAS that was still above the target and significantly higher than the previous week (149%). The CPM was almost identical, $0.33 vs. $0.32.

Campaign 2 also recorded a target ROAS with a clear progression passing from 71% to 83%. Of course, it would be necessary to let the campaign run for several weeks for it to start being profitable due to the large number of zones. Given the significant volumes, it was suggested to the advertiser to switch the bid optimization from “Maximize Volume” to “Maximize target ROAS '' in order to focus on the ROAS and not on delivering conversion volume.

Third-week results using Smart Bid

After the third week, there was no significant change in the ROAS for Campaign 1 and the CPM had stabilized, so the conclusion was that most segments had been fully optimized at this stage.

Concerning Campaign 2, once the Bid Optimization changed to maximize Target ROAS, the ROAS improved  from 83% to 120%, which was the advertisers initial objective. We noticed that several segments were no longer receiving traffic due to the change in optimization strategy prioritizing ROAS instead of volume. All zones with an ROAS below target had mostly recorded a significant drop in CPM, leading to a decline in traffic. As can be seen in the graph below, overall, the change in optimization strategy clearly allowed the campaign to focus the budget on segments with an ROAS above the target set at the campaign level.

Conclusion

Smart Bid is a powerful feature that allows your advertisers to optimize bids on specific segments, which is impossible with other bidding models. Smart Bid is the recommended model for campaigns targeting a huge volume of zones, as it will guarantee their optimization and relieve advertisers of the tedious tasks that managing a complex campaign requires. The success of Smart Bid campaigns will depend on the data collected, so it is essential for your advertiser clients to leave enough time for their campaigns to better calibrate the bids according to the Target ROAS.