Whilst we can’t guarantee that it will (you might be doing everything perfectly), it’s highly likely that it will. It’s the industry standard approach to measuring and optimising media and we find that it typically improves ROI by at least 20%.

Econometric models aren’t the answer to everything; whilst they do some things very well, they cannot answer some questions. Typically, where spends are very low or answering very detailed questions around specific media formats (e.g. is one format of digital display ad better than another), or where there’s been no variation in deployment (for example, if a long-term sponsorship has always been in place). Also, it cannot predict the performance of activity that has not been run.

The models are generally very reliable as long as there are no changes in the fundamental dynamics of the market or the brand. We’ll often recommend, after the model is run, calculating predictive forecasts to see how well the model fits the actual ongoing sales data. When these begin to diverge, it’s usually time to update the model.

• If your media strategy needs changing.
• Which channels are providing the best ROI’s.
• How to increase ROI by optimising channel mix.
• How to optimise media laydown.
• What the optimal level of total media spend is, whether or not business growth targets are achievable and how much of that can be driven by media.

• You can use the model to run different media laydown scenarios and predict outcomes. This helps in testing various strategies before implementation.
• Factoring the econometric model findings into your planning process. Combine these insights with other data sources to create a robust strategy.
• Determine what aspects of your current strategy are working and which aren’t. This forms a key foundation for making informed decisions.
• Treat the econometric model as a living document that evolves with your strategy. Regular updates and reviews are essential.
• Make econometrics an indispensable part of your planning process. Many organisations wouldn’t dream of planning without it.

It’s not uncommon for senior managers to have had either no experience of econometrics, or in some cases, a negative experience. We’re very used to taking senior stakeholders through our process to reassure them and to explain what exactly they’ll get at the end of it. We can provide example outputs and case studies of where we’ve addressed similar situations.

The best time to build an econometric model is before the annual planning phase, which often starts over the summer depending on an organisation’s financial year (yours may differ).

If you’ve recently incorporated a new creative, a great time to conduct rapid ROI assessments is one month after the new creative has been running.

Plan to update the model within the year. April is usually a good time for this (depending on your organisation’s financial year), ensuring that you’ve updated insights before the usual planning phase in August / September.

It’s a different sort of project from others that you may have run in that it involves assimilating data from all areas of the business that have probably never been aligned before. It takes time; there will be glitches. We’d estimate that at least half of every econometrics project is spent commissioning, collecting and validating the data. We’d typically allow 3-4 weeks for data collection and validation, 2-3 weeks for modelling, then 1-2 weeks for presentation & follow-ups. A typical project from start to finish takes around 6-8 weeks. If you need urgent analysis and have data ready to go, contact us. Sometimes, we’re able to turn around a project in less than a week.

Usually, we’ll use the most recent three years unless there’s a reason to go back further or not as far (e.g. data collection only started 2 years ago).

Planning is critical, focusing on the key questions you need to answer and ensuring you have the data to answer them. We also believe very strongly in stakeholder engagement up front to tackle any issues, set expectations and engage with any potentially difficult stakeholders well in advance.

In the initial stages, where we’re still discussing and scoping the projects, we usually suggest (after signing an NDA) looking at your data and see if we feel it’s suitable to model. There is no cost or obligation at this stage, and we’ll be candid about whether we feel the data isn’t of the right quality to model.

We need weekly deployment and spend data. For TV, this is typically weekly TV Ratings; for digital channels, we typically model impressions and /or clicks. For some channels we even model spend if no other deployment data is available.

The starting point for most econometricians is 3 years of weekly data for the KPI you are trying to model (sales, web visits, new customers, quotes and so on) and the activity you’ve done to support it. Models can be built with less data (we’ve successfully delivered projects with as little as 26 weeks of data), but less data means less information, fewer insights, and fewer of your questions answered.

It’s important that we take account of non-media factors so that we avoid mis-attributing effects to media. Typical factors we’d routinely test include: seasonality, weather, base, economic factors, consumer confidence, distribution, price and promotions and ad hoc events.

We’e a secure cloud-based FTP server. We provide you with a secure login for the duration of the project, where you can upload your data.

We’ve secure file servers and are fully GDPR compliant, we also use secure data transfer via FTP.

No, we don’t need PII. We only need aggregated weekly data (total number of sales, new customers etc), which is fully anonymous.

We’ll retain the original files and create formatted versions of them to use in our models. We’ll also chart it and present it back to you so that you can check that you’re happy with the data going into the model.

Yes, we’ll help you identify the right data to put into the model and arrange calls with the various people in charge of the data sources. Some further things to consider include:

• Do you anticipate issues with getting any data? Does it require special analysis internally?
• Are there any drivers for which you don’t have data?
• How long will it take to get both internal data and media data from agency partners?
• Will you need us or another supplier to source data for you?

We use pooled multivariate linear regression to build our models, which we fit using OLS and Bayesian analysis. If you’d like to talk in technical terms about the modelling techniques, we’re very happy to do that.

Bayesian refers to approaches that blend prior information with current data to make better predictions about the future. Moreover, in every model there’s a myriad of decisions that the model-builder makes. There are commonly 4 areas where we’ve found that Bayesian has made a significant improvement to the models we’ve built: 1. Poor data quality – situations where the data wouldn’t normally permit a model to be built 2. Collinearity – in other words, when things happen at the same time 3. Model stability – when results change significantly from study to study 4. Complex effects – when we’re trying to tease out subtle effects.

Adstock measures the memory of your advertising. We measure Adstock as an exponential decay, in effect like radioactive decay, where the effect gradually dies out over time. We find that different media channels, and even different creatives, tend to have different rates of Adstock decay. Adstock rates are measured from your data.

Twice as much is not usually twice as good. That second helping of ice cream isn’t quite as good as the first scoop, as taste buds are overwhelmed and appetites sated. So, it is with media. Twice as much is not twice as good. Doubling the deployment of media within a particular period will not typically result in a doubling of the sales driven by media.

Diminishing returns are an unavoidable fact of life. Knowing how they affect your brand helps you deploy the right amount of media in the right way – and boosts your ROI.

We help you with that. We typically start with a round of stakeholder interviews to elicit hypotheses about what may have driven your business. After collecting the data, we test these hypotheses as we build the model, to see whether they are true.

Sales is the most common, but any response metric that you measure can be measured, for example: new customers, repeat orders, leads, enquiries, website traffic, brand awareness (if it’s tracked regularly and consistently).

All and any for which deployment data exists. Such as:

• Offline – TV, Press, Magazine, Cinema, Out of Home, Radio
• Online
  • Search – paid (PPC) and free (organic)
  • Display – standard or programmatic
  • Video on Demand (e.g. YouTube), Broadcast VOD (catch-up, e.g. ITV, C4)
  • Social – promoted (paid) and unpromoted (community managers) – e.g. Facebook, Instagram, Twitter, Etc
• CRM, eCRM

It’s important to involve stakeholders from key functions early in the process; this may include finance, sales, BI, your media agency and more besides. They need to be comfortable with the inputs, what we’re doing, and why. If you don’t have buy-in at the start, then, come the results, people will pick holes in it if they don’t like what it’s saying. Not every activity will have been a success, so you need to be prepared for that and have those conversations up front, not at the debrief.

We believe in collaboration. As well as working with our clients, we work closely with their media agencies. Our background is in media, and we understand the challenges that planning and executing media campaigns can bring.

Over the years, we’ve developed effective working relationships with many media agencies, large and small, which has enabled us to produce powerful results for our clients.

Again, we can help you with this. What are you hoping to learn? In an organisation of any size there is usually a long list of potential questions. From your long list of questions, which ones are critical? Solicit questions from across the organisation. Then get in touch for an informal discussion.

As a rule of thumb, we would say we’d need at least £1-2 million in annual media spend. This will ensure you get sufficient value from your investment in econometrics.

You’ll have a clear understanding of your business drivers and the role of media within that. Specifically for media, you will gain an understanding of what is or isn’t working. You will know which media campaigns and channels performed better and, therefore, what is likely to perform better in the future.

Econometrics quantifies the ROI of media channels and provides a high-level overview of your marketing mix. Once the modelling process is complete, scenarios can be run through our optimisation software to explore how adjustments to the channel mix and annual budget allocation could maximise impact.

An econometric model will inform your media strategies by analysing past performance, identifying high-impact channels, optimising budgets, and guiding data-driven decisions for future campaigns. By using the model to reveal how media works for you, we can help shape your media strategy into one that will best deliver your business and marketing goals.