Here we are with the second instalment of our Essential Econometrics series. If you haven’t already read Part 1, then I suggest you start with that. Otherwise, read on for Part 2.
6. Using Econometrics for what it can do best
We’d be lying if we said that Econometrics was a unicorn and was the answer to everything. It isn’t. We often tell people that they don’t need a model. After all, predicting the outcome of a radical change from the norm is not an econometric strength. For example, a publisher wanted to know what would happen should they switch their paid publication to free, but honestly, no model could solve that mystery.
Other examples:
- What about that £3 price point? Sure, an Econometric model can draw some fancy curves, but that won’t tell us if customers are ready to actually pay £3.
- What impact will my new product have? Who knows. Is it any good?
- We used 10 pieces of digital copy, and spent £2k on each – which one worked best? Again, you’d be drawing a dead end with Econometrics
While Econometrics is useful, it’s important to remember that it’s not a foolproof solution to all issues.
7. Data – the unsung hero!
The data collection process is often overlooked or underestimated but is a crucial stage.
When it comes to data analysis, it’s easy to assume that the modelling stage is the most difficult and time-consuming. However, the truth is that data collection can be the most laborious and detailed task. To build a model, you need to collate data, usually weekly for at least 2 years, for all variables you want to include in the model. This means that consistency is key, and any gaps or changes in supplier can cause problems. Always begin by considering which drivers are relevant to your project and source the data accordingly, rather than relying on whatever data is readily available.
On the bright side, once you’ve completed this stage, you can plot all the drivers against sales, which can be an illuminating experience for management. Unifying all your data in one place often provides valuable insights in itself, even if the process is sometimes arduous.
How data collection feel to every econometrician
8. Setting yourself up for success – Buy in
You’ve done the data gathering, you’ve got your model, you’ve got the results, what is the next step? The truth is, it may be too late if you didn’t have buy-in from colleagues from the beginning. Some may criticise the input data, others may argue that the analysis is flawed, and others may suggest that there are missing components in the model. (Unless, of course, the model tells them that everything they did was great, in which case, they may remain silent!) However, this is unlikely. Additionally, there may be winners and losers, brands and activities that require more or less investment. While it’s noble to bring transparency to your business, some may not like what they see.
A plan is necessary to prepare for this. Involve key personnel from each department, including finance, right from the start. Talk with senior stakeholders about the need for a single version of the truth. Setting the groundwork for establishing a learning culture is equally important. It appears that the modelling aspect is the easy part, but our experience suggests that this is often where people tend to overlook the importance of bringing colleagues along with them. It’s better to influence them upfront than to delay it until the end, where it’ll be more difficult and painful.
Buy-in critical. Staged high five optional
9. Getting from ‘did’ to ‘do’ – Optimisation
Now that we have a model in place, we can use it to determine the best marketing strategy for our business. However, optimisation is a complex and nuanced process that requires a careful combination of inputs, costs, and business constraints. Questions such as how much we should allocate and where, whether we are prioritising volume, revenue, or profit, and if we’re willing to invest more in our key brand at the expense of others, must all be considered.
Despite the difficulty of the process, optimisation provides two benefits. Firstly, it brings the model to life by running iterations that help to clarify the results:
“Oh, that didn’t have much impact.”
“Well, that was expected based on the presentation.”
“Ah, I understand now.”
Secondly, it forces the organisation to think critically about its goals and limitations. In our experience, the value of “doing an optimisation” comes from the process of asking and answering these difficult questions, rather than just the answers themselves. It’s crucial to take the time to ask these questions and avoid the temptation to delegate the task to your econometric consultancy, as this approach is unlikely to yield satisfactory results.
10. Testing, Testing, Testing…..
Hypotheses testing is crucial in creating successful models. When refreshing the model, it’s important to test different activities at an appropriate scale to see an effect. Conduct these tests in an as scientifically manner as possible to ensure accuracy.
When building a model, the initial data is often a fait accompli, which may not be ideal for the model. Activities may have been run simultaneously or always on, making it difficult to isolate their effects. However, the next iteration of the models – the update – offers an opportunity to test different scenarios, such as turning specific activities off, and observing the results through the lens of your models. This way, you can restructure your plan and misalign activities to tailor the model to your needs.
Consider which decisions your company can practically make and how to test them. This approach will turn your model into a live tool that helps with future planning, rather than just measuring the past.
So, there you have it. Our 10-point Econometrics basics guide. Hopefully this has given you a better understanding and if you’d like to go a bit deeper, download our free ebook ‘Guide to Econometric Modelling for Modern Marketers’.
Or if you have any questions, please feel free to get in touch with us.
Tom Lloyd