Native AdOnline ad case study

A common way to acquire traffic is going through display campaigns to redirect users to the product one wants to promote. Campaign’s efficiency will rely on many factors as traffic quality, ad location, cost and so on. But one of the most important thing is to tailor the ad to perfectly match the targeted audience which is not done most of the time.

A company I worked for was partnering with one of the top 20 classifieds websites in the world, with international audience. This partner wanted to try to boost his display revenue by creating a new native ad spot. The ad location was already existing on the website pages, but it was used for classical display with an ad server displaying random ads which a user could have already seen on other platforms.

My team got in charge of designing and producing this native ad that it matches technical and traffic nature constraints from the partner. We would host it on our servers.

The Native Ad Design

Foliage Illustration

The easiest part in this project was designing the ad. It must perfectly fit the website layout, using the same visual code than the displayed offers on the page. For this, we went for an HTML ad. We reproduced to pixel-perfect quality the offer slot, adding a big CTA and we added a sponsored offer mention that users could distinguish the ad from a normal offer.

Normal offer layout Sponsored offer layout

Offer and native ad layouts - The native ad uses the exact same layout as the normal offer. An image on the left, a title on top and a description below. A CTA at the bottom and a sponsored offer mention have been added to differentiate the ad from the offers.

The ad must be responsive the same way the offer slots were, to receive traffic from any device. After few browsers testing, we got it right, the ad was correctly displayed, no matter the browser or the device, mimicking the website offer slot behavior.

Traffic Source Constraints

Traffic Illustration

In classical online display campaigns, ads are generated based on a product and not based on the traffic nature and constraints. Usually, a company wants to promote its product, goes through several ad networks which will display the ad on their network of partners websites, produces the different ad formats and then hopes this ad will be profitable.

The ad content is mostly not optimized for the user, displaying the ad next to content the client doesn’t want to be associate with (latest YouTube ads scandal here, Procter&Gamble cutting its ad budget here). Based on user’s cookies, retargeting campaigns serve again and again the same ads to the very same user for days.

As opposed to regular online display campaigns, our partner had several requests to be sure the native ad was optimized based on many factors:

  • The ad content should be relevant to the webpage content the user was accessing.
  • The ad content should be relevant to user’s location.
  • The native ad would be displayed twice per page on two different slots and each of these slots should have a different content.
  • The number of viewed pages during an average user’s session on the partner’s website was 60 pages. The average length of a user’s session on the website was 10 minutes. The partner wanted to be sure that the same ad content would not be displayed again and again to the same user while browsing the website and the offers.

Based on these requests, we started to dynamize the native ad content to fulfill them.

Adapting Ad Content

The native ad should consider two parameters to display the proper content to users: where the user was physically located and what page category he was accessing. The ad like the offers displayed on the page should be local with a relevant content.

About the local dimension of the native ad, we used a mixed solution. First, based on parameters sent to the ad by the website which classifies offers per area, we knew from which area the offers that the user was consulting were coming from. Then, we were using geoIP location to pick the right ad content from our database, overlapping it with the area data sent by the website. The goal was to provide a very local ad content to a specific user that the native ad “blends” well within the webpage offers. For mobile traffic, the user’s location was not used as geoIP is not reliable.

About the relevance of the content, we used again parameters sent by the website. Based on the page’s category the user was browsing, we were selecting the proper ad content from our database.

Based on all of this, we were adapting native content on the fly, displaying the right product picture and information, matching the category and the location from where the user was from and in what area he was looking for something.

Adapting To User’s Behavior

Users' Behavior Illustration

Websites have different audiences and these audiences have different behaviors based on who they are what they are browsing. Our partner had a very good picture of what his users were consulting, how they consumed the content of the website, how many pages they would browse per category and how long they would stay on the website.

Based on the number the partner gave us, we set up an ad content strategy to ensure that all along a session on the website, a user would get as varied as possible ad content each time he was browsing a new page, within the area and category constraints.

We used cookies and PHP user session to ensure that:

  • On each page he was browsing, he would see the two native ad instances with a different content.
  • For a 10 minutes duration, which was the average user’s session time on the website, none of the previously displayed content in the native ad would be served again, unless we didn’t have enough content available based on the category and area constraints. In this case, we would display again some content the user has already seen.

In Production

After another round of tests on partner’s test pages, the native ad was rolled out on a single category to see how it performs.

The partner was concerned about two things:

  • Is the raw performance of the native ad worth it?
  • Will it cannibalize revenue from other classical display ads?

The very first day, it generated around € 900 revenue for roughly 8,000 clicks. For the next following month, revenue per day kept growing up to € 3,000 and a click average range between 9,000 and 10,000 clicks per day. EPC was around 0.3. The partner was happy with the revenue growth trend from the native ad as we fine-tuned its behavior during the first weeks.

The other good news was that the native ad was not impacting the revenue from the other display ads. The native ad was likely targeting a different population segment than the one generating revenue through classic display ads. It meant that this native format and the way we handled it was generating an extra revenue for the partner.

Since then, the partner deployed the native ad on other webpages categories with the same success.