Evolving Display Advertising Through PDCA. Continuously Optimizing to Achieve Results.

Following Part 1, we asked Mr. Yasuhiro Matsuno and Mr. Hiroyuki Tamura of Nextedge Dentsu Inc. about device considerations when delivering display ads and post-delivery optimization through PDCA cycles. What key points should be considered in actual operations to drive results?

*Nexedge Dentsu Inc. became "Dentsu Digital Inc." on July 1, 2016.

■Smartphone? PC? Effective Display Ad Usage Tailored to Each Device

──Are there devices that are more or less compatible for delivering display ads?

Tamura: Yes, there are. Take job recruitment campaigns, for example. The KPIs are member registration and job applications. For member registration, smartphones have a significantly higher conversion rate and lower CPA. Conversely, for job applications, PCs have a higher conversion rate and lower CPA. Member registration involves fewer fields to fill out, making it convenient on smartphones. Job applications, however, require filling out more detailed information like career sheets, making PCs more suitable.

Understanding these device-specific characteristics allows us to focus promotions on smartphones for registrations while reliably capturing applications via PCs.

Matsuno: I feel smartphones are becoming a crucial channel. Especially for e-commerce sites focused on low-involvement products like inexpensive daily goods, and clients requiring member registration, the majority of conversions now come from smartphone ads.

The dominant trend in smartphone display ads is clearly the "in-feed" format. "In-feed" is a type of "native advertising" that appears between articles on social media or curated media sites. Yahoo! also uses this format, and in our actual case studies, we've seen significantly higher efficiency compared to PCs in many instances.

Furthermore, even for products perceived as "inefficient on smartphones, not leading to conversions," there are actually many cases where "users see the ad on their smartphone and then complete the final conversion on a PC." This highlights the importance of a cross-device approach.

In accounts we managed, when smartphone conversions were set at 100, cross-device conversions—where users viewed on smartphone and purchased on PC—added approximately 10 to 20 conversions. This prompted a reassessment of smartphone ad effectiveness. Google is now able to capture this data as well.

■Continuous improvement for better ads

──Please explain the process after display ads are delivered.

Tamura: In PDCA terms, the process involves Plan and Do before delivery, followed by Check (evaluation) and Action (improvement) afterward. For display ads, we create multiple variations by altering elements like the catchphrase, color, or banner size. We then select the most effective ones, ultimately retaining those with the highest conversion rates to refine them into optimal creatives. We cycle through PDCA as quickly as every week or even daily (!). That said, we never fixate on the last surviving display ad. We constantly generate new ideas, always aiming for something better. So, it never ends (laughs).

Matsuno: Regarding Check (evaluation), while we sometimes receive requests from clients, we actually have more opportunities to proactively propose changes. We constantly think, "What if we tried this? It might boost effectiveness," and we do it with excitement. I think of effectiveness verification as being like conducting experiments.

──Regarding the Check (evaluation) phase, what aspects do you focus on when verifying effectiveness?

Matsuno: There are three main points. First is formulating hypotheses. We consider what might happen if we target a different segment in addition to the ones we've been using. For example, with smartphones, we might hypothesize that conversion rates could increase during commuting hours or lunch breaks. Once we establish such hypotheses, we verify them by comparing them with actual data.

Second, clearly defining "which metrics to verify and along which axes." We precisely decide which metrics to track—clicks, CTR, conversions, CPA, etc.—and determine whether to compare via period verification (before/after the initiative) or conduct A/B testing with simultaneous delivery. Once decided, consistency is crucial, so we strive to reach full agreement with the client before proceeding.

Finally, it's crucial to alternate between macro and micro perspectives. If a single-day test shows only marginal results, try a week-long test. If overall effectiveness improves, break it down to segments or creatives to pinpoint the most impactful elements. The final review is vital; steadily accumulating this know-how is how we improve results.

Tamura: What I focus on is not just committing to client requests, but clearly establishing whether the requested cost per acquisition (CPA) is derived from the client's sales figures and is reasonable. For example, if a client requests a CPA of ¥10,000, and I feel it's unreasonable, I calculate and present what I consider a reasonable CPA. I believe thoroughly aligning on this point solidifies the core KPI.