Can AI be the respite for the weary programmatic trader?
Programmatic traders are being underutilized as most of their time is spent managing mundane complexities and disparate processes. There is also a lack of workflow and efficiency, as traders must juggle up to eight platforms a day, says Dominic Powers, the chief executive officer of CrtlShift.
The findings are based on a survey that CrtlShift released last month, which found that 48% of respondents used five or more platforms daily, most teams (40%) used two to four platforms a day and almost 20% of respondents were using eight platforms or more daily.
The survey, whose respondents makeup of 35% of trading leads and 65% of traders, noted Google (Search 360 and DV 360) and Facebook accounted for at least three of these platforms, while the rest were adtech from other trading desks.
“When you speak to an agency or brand they say, 'We are just using a couple of platforms.' They don't think it's a tough task to have their traders trade. But obviously, what we saw from the survey is that a large majority of the traders are using four or more platforms,” Powers explains to The Drum.
“20% of the traders are using eight platforms, which they must go into day-in-day-out. It takes them two minutes to log in and do something. That is a lot of platforms to go in and out of like Facebook, Google and a couple of demand-side platforms (DSP). Name another industry that actually has to use that many platforms to get their job done.”
Powers points to the finance and the human resource industry, as examples of where professionals are mostly using single platforms. He says it is an irony that in the marketing industry where it is all about automation, there are still a lot of manual processes.
This means there is a possibility if the complexity of the platforms increases, or if the number of platforms goes up, the service company must look into hiring more traders. Powers questions how much time and resource they allocate to retraining their team when platforms evolve.
“We are asking people to do manual work, which leads to significant amount of money that is being spent in the people managing them to see how much spend can a trader manage,” says Powers.
“Traders are also being evaluated on how they manage branding, how much can they manage in performance and how can they grow. That is why we did the survey, because for us as an overall industry, it is good to understand what kind of efficiencies remove from the ecosystem.”
The survey also found 42% of respondents saying that 50% of their reports were done manually as they must stitch together data from multiple adtech platforms such as search, social, programmatic and ad servers.
Over half of respondents (57%) in South East Asia said that over 50% of their reports were done manually, while 28.5% said over 80% of reports were done manually.
Traders who worked on intensive performance campaigns suffered the most as more than 80% of reports were done manually. They also had to make more than 10 optimizations per campaign per week.
According to respondents in the survey, campaigns with budgets above a million and below US$50,000 required the most manual reporting. While smaller campaigns were driven by performance considerations, larger campaigns, most likely branding campaigns, saw the need for customized insight.
Powers says the prospect of having to manual reporting means that agencies are struggling to find talent in programmatic trading, as the average tenure of the trade from the report was about one and a half years.
“We have seen historically because of these inefficiencies, traders are moving on every one and a half years. That means there are things that the industry needs to be doing for us to push forward,” explains Powers.
“For us, this is complicated because it is about automation and aggregation. We have got to find ways to make life easier for the people who are the knowledge workers in this industry. If you speak to agencies, what they want to do now is get more from the same resources because they have more and more budget going towards programmatic.”
He adds: “Programmatic is going to be more of a $70 billion industry this year and maybe 90bn a year after, but at the moment, you cannot find the people to execute. That is why we need to be more efficient.”
Will AI help automation?
To help weary traders with their workflow and efficiency, implementing artificial intelligence does the dual task of delivering great performance through highly granular campaign optimization as well as removing the bid optimization workload of the trader, says Rahul Vasudev, the managing director for APAC at Scibids.
He explains there are some things that humans do well, like understanding the client goals, translating those goals into something that a DSP can understand, and generating insights. However, there are also limitations like how many variables they can look at, how adept at mathematics they are and how they are being able used to predict performance, for example.
“The latter is where AI comes in to act as a tool for traders, optimizing up to 22 million variables, making changes to the campaign every two hours and picking the best impression for every client,” he says.
In addition, using AI to automate the ingestion of log level data from the major DSPs means that the trader has more time to focus on setting up the campaign for success, ensuring the right brand safety filters and supply, rather than trying to connect the systems, according to Vasudev.
“Apart from this, just the sheer volume of data that the AI can wade through to make intelligent predictions, means that the trader essentially delivers 40-70% improvement in campaign performance,” he explains.
When it comes to reporting, Vasudev points out many agencies and trading desks are setting up reporting suites that are reducing workload for traders. He says this is a straightforward job of creating API connections from DSPs and ad servers to reporting platforms.
What AI can do, he says, is transform the campaign away from bland clicks and conversions, basically from a spray and pray approach to cherry-picking but at scale.
“The role of the trader changes to the creation of insights, interpreting what the AI is discovering that can be understood by their clients. This is an exciting new skill set that we see traders developing,” he explains.