Should CMOs see Algorithms as Friend or Foe?
The conversation on AI and the future effect of algorithms on marketers continues post DMEXCO and today played host to a panel session hosted by Scibids on this hotly debated topic.
Using algorithms to buy media us nothing new with companies such as Criteo having built their business on non-human traders for years. There are a number of standout reasons for this, primarily that machine learning is a low resource and overheads with high outputs. Machine learning can analyse every touch point and optimise to each part of the consumer journey.
So the big question is how can CMOs make algorithmic buying work for them.
The first clear point to make is that this should not signal the end for expert 3rd parties as it is hard for brands to take it in house and at this early stage, there is minimal appetite from brands to make that leap. However, advertisers are wanting to have more ownership of a more advanced measurement framework to manage a cohesive, omnichannel approach.
First party data is the starting point as its knowledge about customer and clever modelling should be welcomed by CMO’s to bring it together. Marketers need to understand off the bat that outcomes short and long term are a constant push and pull, which are affected by variables on the data.
It goes without saying that DR campaign and branding campaigns require different kepis. This is where algorithms can teach and guide us, and we can put up different line items in response to what the algorithms say. Marketers then need to extract the relevant data to learn more about that first party data. The use of this data needs to be as responsive and fluid as much as the users are.
One thing that has become clear is that CMOs are becoming more demanding within DR - needing to satisfy multiple KPIs simultaneous. For many, gone are the days of just driving online purchases; now the focus exists on attracting the customers that go on to be the most profitable over time and this is driving technology to be more customisable around that advertisers’ specific data points. It is apparent that the need for driving more specific transaction types exists in all channels, from partner marketing to programmatic.
CMOS need to own their data and tech and looking at right operations models and this is where it gets messy and complicated. The answer? Simplify it and work with specialists to still deliver campaigns but layer it with the intelligence on what your KPIs are and what you want to get out of it in terms of business objectives.
So does this ring the death bell of agencies and platforms? No. They have one job to do and that is to prove that the advertising worked against the brands’ business objectives. They also need to validate that algorithms helped - even if they are not the silver bullet that they claim to be.
For example, bid multipliers of custom variables really helped towards a step to delivering algorithmic buying and allowed us to move away from creating line items which creates efficiencies and value.
One point to note to the agencies is that often the marketers don’t care how it works - they just want to know that it has worked, and this is why measurement is so critical. Does the CMO want control? Yes. So they want to look underneath the bonnet? Often, no. Let’s use the analogy of going to a restaurant to eat a meal; most people don’t want to know the ingredients that were used or where it was sourced and how it was made. Some just want to eat the food and enjoy the fruits of the expert chef.
In the same way many CMOs just aren’t interested in the tech behind marketing. Too many are also caught in the headlights, but a lot are seeing it as an opportunity to take back control. What is clear is that CMOs need to look at costing models and how to get the insight out of the black box. There is no doubt that there is a new dimension to the CMO role and they need to be able to start talking about data science, be keen to learn more about how algorithms are allowing us to move beyond ID level and have a better view of the overall outcomes.
Drive efficiencies with what the data scientists do and make improvements with your own knowledge of your brand and customers. CMOS need to step up to the table and be involved in the objectives and KPI settings and then use algorithms to meet business objectives.
Ultimately, this will help keep the role of the CMO alive!