Article Tech

How to use AI to support strategic marketing planning

Marketing planning is a fundamental requirement within any successful business for a number of reasons:

  • It helps to define clear, measurable objectives which align with overall business goals; whether that is to increase sales, expand the market share or launch a new product.
  • It can help with aligning resources (financial and human) to achieve the overall goal by prioritising initiatives that will attain that goal.
  • Refines targeting within the different channels to reach overall objectives
  • A well-defined marketing plan can help businesses differentiate and therefore be more competitive in a challenging market
  • By understanding and evaluating market trends, businesses can proactively look to address them and adapt more quickly
  • By measuring the key KPI’s and understanding what can influence trends within those, and therefore identifying areas to improve for the best return on investment (ROI)

Through developing and executing a well-crafted marketing plan, businesses can effectively navigate the complexities of the marketplace and drive sustainable growth and success.

With the launch of AI in November 2022, and now being more than a year on, the fears are that if you do not adopt it in some way within your business, those that do will have a distinct competitive advantage.

How to use AI to create a high-level marketing plan

We have been exploring how to utilise AI in one area of Marketing planning, which is the creation of the overall high level Marketing plan with budget and conversion allocation across channels (the strategic marketing plan)

In a typical planning process, a company would be taking their current yearly plan with budget and conversion allocations and assessing, on a regular occurrence, where they are against that plan. They might perform their own analysis to understand trends and the market, to then influence the figures that would be entered into the latest version of that plan.

Here at Fusion Analytics, we have been looking at how AI can enhance or compliment that process, through developing prompts to feed into an overall end to end planning process.

Using AI to provide a view of channel split

For those of you that have delved into using open-source AI platforms, you might have discovered the art is in the prompt you put in, in relation to the quality of what you get out. We will demonstrate this here:

As an example, I have a budget of £3m to spend and need to achieve 1-million-unit sales, so could put in the prompt of:

This provides an output of:

Whilst this looks on the surface to be quite useful information as it is providing a split of channels and allocated budget to them, with some context, on closer inspection it has a number of flaws.

The budget and percentage when added up is more than the budget given, the channels are provided at a high level so for taking action on what to do would be quite unclear, it isn’t providing how the sales might be attributed and this of course isn’t taking into account what channels you actually might use at present.

Taking this a step further you could look to refine it a bit more:

This generates:

This is a more easily digestible output, with some reasoning behind the budget allocation.

However, as a business you would want to know how to spend it across a year, as we know it does change with seasonality and known peaks and troughs

With that in mind you might then do a further question:

This gives an output of:

This looks to be useful, splitting it out by month and applying some level of difference across the year, be it simplistic.

However, still some flaws in that the totals add up to more than the budget you had, for example Digital advertising was £1m allocation, but the by month split adds up to £2.5m

It also typically uses the same proportion across the year, whereas usually it is optimal to change your mix as the market peaks and troughs.

Therefore, as demonstrated it is an art of understanding what exactly you want to get out of AI, and applying some logic and common sense in terms of the flow of the prompt to get to the desired output. Then once you have that to experiment with combining the prompts to get to the output with one prompt.

So, after some iterative refining we eventually arrived at a prompt and output we were happy with that would feed into our AI generated solution.

Getting to the right prompt to make it more actionable

Through our investigations we have come up with a list of tips in getting the right prompt for an ideal output. These are:

  • Define a role that you want the ChatGPT to play
  • Be specific in terms of the requirement/what you are trying to achieve
  • Define the task that you want it to do
  • Provide details that are known eg. Groupings/segments
  • Note any restrictions to avoid
  • Specify the desired output and structure of that
  • Provide examples
  • The use of results and data in the solution

The next stage in the development of using AI within planning, is to enhance the results by adding known data and business results. Until you do so, what AI provides is essentially a more intelligent summarisation of theoretical research.

Integration into the Webfusion toolset

Here at Fusion analytics, we have developed an AI module within our Webfusion marketing planning tool, which includes a range of pre-tested AI questions including what channel mix you should use and how to adapt to different market conditions.

WebFusion facilitates the planning process cycle described above, helps to automate the process, eliminate user error with excel spreadsheets and consolidates into one version of the truth.

Therefore, AI is a natural progression enabling enhanced decisions into this tool and process. The tool also holds all of the present and past strategic marketing plans, which can be one of the data sources to feed into the AI generated solution.

Our AI module is designed to help provide independent advice on how you plan your marketing budget.

Should you wish to discuss the planning module or how we are utilising AI within this process please get in touch