Can AI answer that age old marketing question?
SourseAI's CEO, Tanya Hyams-Young, weighs in on the age-old debate between differentiation or distinction, arguing it doesn't really matter because AI makes it possible for organisations of all shapes and sizes to accurately measure what works best for them.
The debate between differentiation or distinction continues to rage on amongst marketers. While there’s certainly merit in these discussions, every time it comes up, I always think the whole thing is a bit pointless. Couldn’t we just focus on trying to get accurate attribution data so we could make decisions based on the real-world performance?
If we all used marketing mix modelling (MMM), we would actually understand the impact of spend on acquisition and retention, and we wouldn’t need to debate reach vs targeting. But the obvious challenge with mix modelling has been very similar to the reach or mass marketing approach. Not many marketers in the real world could afford it. Until now.
Open AI?
2023 was the year AI went from pipedream to something that almost all organisations could use. Adobe research suggests that 68% of brands are already using generative AI on marketing campaigns. That’s great. But the real value of AI isn’t A/B testing EDMs, drafting copy or pre-filling briefing documents.
AI can be the strategic tool that makes mix modelling a reality for almost any organisation. It’s a tool that can deliver clarity and remove ‘gut-feel’ from decision making. And that’s the ‘bad’ news from Adobe’s study – only 31% of Aussie brands are using AI at an organisational level.
This is almost certainly a legacy issue. It used to be the case that if you wanted to use AI and data science to be more strategic and do something like mix modelling, you had to hire a team of experts, allocate a significant budget and a year (or more) for them to build the function within the business. With the rapid rise of tools like generative AI, that’s no longer the case. While ChatGPT isn’t quite the answer, there’s now a range of AI tools available off the shelf that allow marketers to get really strategic.
Theory vs real world
This brings me back to the differentiation vs distinction debate. Because it doesn’t really matter which is right. It’s more important that marketers understand which is the right approach for them. With AI widely available, it’s much easier to build the models and crunch the numbers, so you can actually measure the performance of your marketing activities and attribute the impact of certain tactics or channels. And it can be achieved in months, not years.
But that’s only the start. AI and mix modelling bring accuracy and predictability to decision making for marketers. Rather than running a scenario plan based on a previous year’s sales figures, a little science and a lot of gut feel, marketers can use AI to build and run models that take into account broad marketing datasets to forecast an outcome. AI and mix modelling will help marketers scenario plan per channel, per campaign. It would take into account spend, seasonal information and so on, so you can really optimise costs and predict outcomes. It becomes as accurate as a CFO forecasting sales or profitability based on a range of scenarios.
Going further
With AI doing the grunt work, it doesn’t mean the job gets easier, or that AI can replace a good marketer. Marketing will always need strategic thinkers and problem solvers. AI merely frees us up to use data to think more and solve more complex problems.
For example, you can dive deeper into the cause and effect of loyalty. If churn is high, it’s very easy to say you have a loyalty problem, instead of examining the quality of customers you acquire. Often the cause of churn is a discrepancy between the marketing promise and the product. Focusing on conversion without time and effort on education will help in the short term by increasing base growth, but rarely influences lifetime value or average margin per user.
Combining AI and MMM, you can start to spend more time on more sustainable metrics, understanding why and predicting when customers leave. Or you can delve into the customer journey and education to help improve the quality of acquisition. AI and MMM gives you the attribution insights so you can explore both sides of the loyalty argument and make decisions based on science, not gut feel.
Tanya Hyams-Young
Pursuit of data perfection?
Data then becomes the most important piece of the puzzle. While the adage ‘a stopped clock is right twice a day’ is a good warning when sourcing data for modelling, it’s also important to remember that a perfect dataset is almost always out of reach.
This is where AI can be such a valuable tool – because it allows you to test and iterate. It makes it easier to identify the ‘must have’ data rather than the ‘nice to have’, so that you can get up and running quickly and avoid the dreaded pursuit of perfection. It can also help you to quickly identify whether issues with your models are real, or just a result of imperfect data.
Final thoughts
Ultimately, it’s most important to remember that every marketer faces their own challenges and budget constraints. We can debate theories and rules of thumb until the cows come home, and some will enjoy doing it. But there will never be the perfect formula for every organisation.
With tools like AI becoming more freely available and easier to use, organisations of all shapes and sizes can now realistically aspire to a mix modelling approach to attribution and scenario planning. Then marketers themselves can decide if differentiation or distinction works best for them. And really that’s all that matters.
Tanya Hyams-Young is the CEO of SourseAI.
Subscribe to the daily newsletter
