The context

As part of a branding campaign during autumn 2019, a few months after a name change, the gas transporter, and storage company Teréga, advised by the agency Makheia, called upon Qwarry to carry out part of a display campaign and a video display, carried out in parallel on YouTube and programmatically with third-party cookies. This programmatic campaign – a first for the advertiser – began in September 2019 and is still ongoing.

Qwarry, a company co-founded by Julie Walther and Geoffrey Berthon in February 2019, uses natural language processing (NLP) to analyze the page content of many websites, determine which contexts are most relevant to the advertiser’s campaign, and then bid on the ad space via its DSP. “Our agency presented Qwarry’s solution to us, and their ethical positioning without exploiting user data was close to our company values”, explains Dominique Boquillon, the communication director of Teréga.

The process

Continuously, the deep learning algorithm based on the NLP developed by Qwarry analyzes millions of URLs to determine a semantic and sentiment scoring on each content and page. “We do the scoring ahead of time because our computing powers do not allow us to perform all the steps in real-time and respond in less than 200 milliseconds to a bid request”, explains Julie Walther.

Before launching the campaign, Qwarry analyzed Teréga’s semantic field and deduced close semantic segments (on targets BtoB: industry, urbanism, agriculture, for example) to link them to each target defined by Teréga (consumers, companies and institutional journalists). The start-up then operates via its own DSP to position the formats on contexts (site pages) that it has “whitelisted” in these segments. Without using geo-localization data, Qwarry operated a national targeting campaign, except on the local target, which was targeted primarily via regional news sites.

On the investment side, no amount for this campaign has been disclosed. However, Qwarry specifies that it pays itself with a percentage of the CPT (not communicated) and that it operates on campaigns with a minimum of 5,000 euros and an average of 10,000 euros, to justify the time linked to the analysis phase of the advertiser’s semantic field and the creation of segments in the case of Teréga. Julie Walther specifies that the added value of her solution is visible in campaigns with branding and “top of funnel” goals like this one, more than on performance-oriented operations: “On this type of campaign we believe that we are complementary to other classic solutions using user data”, she adds.

Read the entire article on Mind News.