10 ways to get more from Marketing Mix Analysis

I'm currently working on a white paper about Marketing Mix Analysis and how the industry needs to adapt to meet new client needs. As part of this paper, I've created some "best practice" advice points that I wanted to share and get some feedback on. Please feed back any comments.

10 Ways to get more from Marketing Mix Analysis:

1. Be realistic about what the analysis can help you achieve – drop words like optimisation and replace with words such as “improvement”. Bravado is a great quality when used correctly but its place in analytics is questionable. Realistically, systematic measurement and improvement programmes in marketing can improve ROI by an average of 10%-15%. However it is not going to happen every time and when it does, it is unlikely to be effort free.


2. Focus on quality not quantity of results– many mix projects suffer from over engineering – both in terms of data used for analysis and detail in results. The old adage of Keep It Simple Stupid (KISS) applies as much for mix modelling as it does in any other area. Start with analysing the key metrics that determine business success before moving further down the value chain. Think of “money at risk” – that should clearly determine the key issues to be analysed. Simply adding dimensions of complexity may bring a new depth to the problems that can be solved but is that really where the real value is going to be derived from?


3. Beware of averages – average measures are a big danger in marketing for two reasons. Firstly, given that there is no such thing as an “average” customer, it is going to be very difficult to create a campaign that will satisfy them. Secondly, average performance may be misleading – it is the marginal returns that are meaningful – i.e. the return from the last unit invested. In an industry that rightfully eulogises the term “diminishing returns”, there is a woeful level of understanding about “marginal” performance.


4. Create genuinely meaningful reports – if an organisation is to make changes in marketing strategy then it needs to be clear to all those who are touched by this as to why changes are taking place. Clear and concise reports can help organisations share knowledge efficiently – long winded one hundred page presentations are not the best method to communicate change.


5. Use simple language where possible – enough said. The people who need to respond to marketing analytics tend not to be the most mathematical proportion of the population. Comparing results in statistical terms is easy for analysts but not for laymen. However the wider marketing community have a valuable part to play in the interpretation itself and needs to be included in discussions.


6. Standardise results– Not all models are built equal. Ask whether you can compare the range of potential models being presented to you on the same basis. If you can’t, how can you benchmark and improve performance. Standardised reporting may be inflexible but it is repeatable and enables quick comparisons between results. Standardised reporting enables one of the favourite techniques of modern managers - benchmarking. Marketing teams often hate it but it is valuable.


7. Focus on the actionable areas of the business – Report writers love to provide elaborate explanations for esoteric findings from their models but often they end up reporting on non marketing issues to marketing teams. Whilst some if this information may be useful, teams are often time pressed and it is marketing issues they need to focus on.


8. Systemise the analysis process– one-off reports may be useful to organisations seeking an occasional review of their activities but only continuous monitoring will lead to real improvements in performance. By systemising the reporting of marketing analytics, real processes can be adopted for test and learn marketing. The CMO council have shown that organisations that have managed to adopt these processes perform better and it can be no coincidence.


9. Develop test & learn processes – many successful marketing organisations have figured out ways to experiment with marketing in order to prove to themselves what is working. In almost all these cases, the organisations in question are able to action findings quickly once they have been identified. If your marketing organisation makes changes to their plans once a year as part of an annual planning cycle then you are missing an enormous opportunity to improve marketing performance.


10. Combine it with other analysis - Marketing Mix Analysis can be a blunt tool - it tends to give high level results. It addresses some important business questions but isn't a panacea. Never overlook other research techniques.

Any other thoughts or comments?

modelQED Launched - finally!

Today we finally launched modelQED™, a marketing mix analysis tool (econometrics if you prefer) which will hopefully help to democrotise the usage of this powerful analytical technique.

For many years, i've figured that marketing mix analysis was actually a fairly well understood tool by many advertisers. There are many articles saying that it needs to be performed by professionals with years of experience. However my personal experience was that for 90% of the models i've ever created, the simple straightforward model was often the best. My econometrics knowledge was only used on the most specialist assignments. For the rest of the time, i was following a well defined process which in most cases gave the correct answers. modelQED™ is designed to deal with these challenges in record quick time and leave the hard core analysts to tackle the remaining challenges.

As with all things marketingQED are trying to do, we've focused our efforts on making modelQED™ both simple to use but also powerful in context. By this I mean that the users interaction with the package is made as easy as possible whilst the analysis delivered is exactly what would be required for the majority of mix modelling. The feedback to date is that we've largely achieved this. I think we will continue to develop the product focusing on even more simplification and user friendliness.

modelQED™ solves lots of issues for modellers but this is only the beginning. Give me a call or drop me an e-mail if you want a demo of this truly unique tool.

We have many other products in the pipeline. Hopefully the market will find these tools useful.

Any comments on the screenshots etc - please post or e-mail. Feedback always appreciated.

A new term emerges for online analytics

Pete Affeld at Numeric Analytics has coined a new phrase to describe the analysis of online advertising using econometrics - "3rd generation" web analytics (see http://www.numericanalytics.com/pdfs/AdServ.pdf).

What I like about this is that it acknowledges that there is now a realisation that you cannot analyse online advertising in isolation from other actives. I've noticed this trend as i talk to online media companies about their analytical capabilities and it certainly seems to be a good time for these organisations to reassess how they define ROI for their customers. If Pete is right then all those TV and Radio analysts will be moving into online in the near future. However as others have noticed, there is plenty of money in the bigger media channels to justify top researchers fees but less is going to be available - hence the need for better tools, training and processes in marketing research. It's this last reason which gives me faith that our new venture marketingQED will be a success.