MiQ Sigma is the product of 15 years’ hard work. We’ve always strived to lead the programmatic industry and make it better, and we believe that MiQ Sigma allows us to continue to do that. But we also know the world is a different place to what it was 15 years ago.
Embracing AI is inevitable and necessary to stay competitive in the market. And we know that championing this game-changing technology comes with a new set of challenges - especially around ESG (environmental, social and governance). We take our vision of being a responsible advertiser in the ESG space very seriously, so we have made sure to keep this in mind throughout our Sigma journey - from development to launch.
John Goulding, our Global chief strategy officer, sits down with Sara Axelbaum, our Global EVP of inclusion and ESG, to answer the most pressing questions following the launch of MiQ Sigma.
What guiding principles did you consider for how MiQ Sigma would be created?
The vision for MiQ Sigma was to create a unified experience for world-class programmatic campaigns across any platform, channel, or KPI. That means being both sustainable and high-performing.
We followed five key principles as we developed Sigma:
- Experimentation: validate our findings with small, quick experiments.
- Objectivity: make sure that we’re always championing the best ideas (because great ideas can come from anyone).
- Focus: keep the most common use cases at front of mind.
- Marketability: build something that will be genuinely useful to our customers.
- Pragmatism: balance the long with the short term.
How do you balance innovation and maximizing profit with responsible AI deployment - especially in a fast-paced industry like ad tech?
We’ve been developing AI solutions for around 12 years now, so striking this balance isn’t new for us. Developing Sigma has been an iterative process, so there has been a lot of testing and feedback from end users along the way. With responsible advertising as one of our ESG pillars, we made sure to keep best practices in mind throughout. We’ve followed privacy centric principles, such as those set out by GDPR, which include thoughtful consideration around the use of AI. And, of course, we’ve also been constantly consulting with our legal, privacy, and compliance teams, running each use case past them to ensure their safety.
Knowing that embracing AI is the key to futureproofing, but also that it generates significant carbon emissions, what was MiQ’s philosophy to Sigma as it applies to the environment and sustainability?
We include our IT costs - which factor in data processing and model training - in our scope 3 emissions calculations featured in our annual carbon footprint reports. These are a fraction of our overall scope 3 emissions and, even as our AI usage grows, they will never be as high as our business travel or the emissions of our media buying, for example.
So if MiQ Sigma can allow us to buy media more efficiently, and run campaigns more effectively, it should be a net positive for the environment, as we continue to strive to run our business in the most sustainable ways possible.
With this in mind, we will be monitoring these emissions and keeping an eye on how they grow over the coming years.
How did you train the model to mitigate bias, especially in the development of the audience personas on Sigma?
This was challenging, as LLMs (large language models) are trained on the entire Internet. As this can be very generalizing, it naturally posed some risk to our audience persona tool. So we committed to working on two key things when generating our personas:
1. Safety
We mitigate as much risk as possible, so we don’t generate inappropriate audiences that contain offensive content, with a large keyword blocklist to prevent end users from abusing the tool.
2. Representation
We provide the LLM with demographic data at the national level and ask it to match its outputs to a representative gender and ethnicity distribution. Not only is this what marketers want and expect, it’s also the right thing to do according to our company values.
There probably will be instances that we can’t predict, but we’re committed to continually refine and teach Sigma’s AI. To continue improving, MiQ Sigma has a handy help tool where users can submit their feedback.
A big part of social responsibility to employees is job security. How are you looking at the impact of Sigma on your workforce?
MiQ Sigma is intended to enhance the employee experience, supporting them with the best tools to make their lives easier.
How many MiQers eat lunch at their desks, work weekends or long hours? MiQ Sigma will give employees a better work-life balance, while still scaling and delivering programmatic that’s far from average. The hope is that, over time, we can all manage more revenue and campaigns, while also avoiding burnout and excessive hours spent at our desks. In fact, we believe that Sigma will empower us to fulfil our ambitions and continue to grow as a company. And, while our headcount may not grow as fast as our revenue (due, in part, to the efficiencies provided by Sigma), it will still grow over the coming years.
Becoming specialists in our new AI-powered tools will help us retain and expand upon our ‘people-powered’ advantage, while also becoming more creative. Being an MiQer means you have to know and learn a lot of information to do your job: Sigma allows everyone to have up-to-date, easy-to-access information at their fingertips.
We already employ some of the smartest people in the industry. Imagine what they can do with Sigma.
What other usability considerations went into the development of Sigma?
From the beginning, we interviewed a broad cross-section of employees from across the business to hear how our tech could make their lives easier. This has been a driving force for the launch of Sigma.
In terms of visual design, inclusion is always at the front of our mind. We always strive to design for everyone, inclusive of low vision, dyslexia and the autism spectrum, among other things, making sure that all our data and information is presented clearly.
The development of ‘dark mode’ is just the beginning of how we’re making Sigma a platform for all. And watch this space for the rollout of our multi-language support.
How are you addressing the privacy and security of client data?
We have a dedicated team that works on data governance, applying technical controls to follow the principles of the major global privacy regulations. This includes minimal use of data and limits the potential for human error. Within this governance, client data is permissioned only for use by the relevant client team.
How are you looking at AI holistically for MiQ’s future?
In simple terms, we have the opportunity to improve results for clients, and make working at MiQ more fun. Tools like Trading Agent do both of these things - as well as turbocharging our work.
A future working alongside these powerful agents means instant access to knowledge, powerful optimization tools at our fingertips, and a more connected, seamless workflow. Operating it as responsibly as we can is table stakes. We’re only just getting started - and we can’t wait for the next few years.
