Viewing entries tagged


The Silver Economy in Holland - an example of a data driven Horizon scan.

The ‘silver economy’ is a term used to describe how the increasingly healthy and demographically significant +65 population could be of greater economic significance in the future.  Thinking about the silver economy could highlight considerable economic benefits and many governments and businesses are thinking about how to better engage with this increasingly significant demographic.  Working with our associates at Future Consult we did some analysis to help the Dutch Rijkswaterstaat better understand what the implications of an increasingly significant silver economy could be for Holland.

To do this we applied a form of topic modelling and expert mapping, that is sometimes referred to as ‘Context brokering’ today.  This post covers how this analysis was conducted, in a separate blog we’ve detailed the key findings from the analysis and the subsequent discussions it was used to facilitate.

Using context brokering to understand strategic trends.

To understand the silver economy and the benefits it could bring there is a considerable wealth of knowledge available around ageing generally.  Ageing and the silver economy relate to research in the fields of demographics, society, health, the economy and even as far as resources and infrastructure, so they are very complex, multi-disciplinary areas of study.  To conduct any analysis on this subject, we thought it best to reflect such complexity and design a basic data gathering method that bought in data from a wide variety of open sources reflecting the different sources of data.  So we based the way of gathering the data on the following process:

Doing this, we defined an initial search that returned 18 open reports detailing the silver economy, society, ageing and limited the geographical range to Holland, or Europe more generally.  After gathering this data as reports we then applied machine reading techniques to extract the most significant keywords for the combined string set of all the documents, doing so allowed us to sample the most frequently occurring key terms:

This data was then analysed further to look at the interconnections between the key terms and, a further level of analysis was conducted to ‘tag’ further terms and specific trends and ideas with the intention of labelling and discovering any interesting ‘outliers’ or signals for new and novel ideas for trends.

Doing this analysis allowed us to start to resolve the complex issue that the silver economy represents into a series of different topics, from the most discussed topics to the least.  This kind of information analysis (albeit from a small dataset) enabled us to generate a simple, ‘topic map’ to inform and guide further facilitated discussions with representatives from across the Dutch Government, Academia and Industry.  This approach, provided a clear context to start discussions around initial assumptions in real data and provides the earliest start point for evidence-based decision making.  The full dataset for the analysis is available here and the ‘rich picture’ produced using Gephi is available here.  For those wishing to engage with a dynamic data visualisation that illustrates trends and interconnections in the master data set, this is all provided in the gephi, rich-picture visualisation to access this data, please contact the team at  For those interested in the ‘top level’ strategic narrative around the data, please see the image below and the discussion of the specific trends and themes (and overall feedback on the technique) is available at the following blog post - The Silver Economy in Holland.



Context brokering - how do you apply it?

To better understand what context brokering is and how it can be applied in decision making, it’s worth considering the following hypothetical example:

A CEO of a large UK multinational organisation specialising in mobile phones has asked the business development manager what the international strategy for engagement in Africa is.  This happens in a board room and, as often happens, the BD manager knows nothing about Africa because he’s worrying about Brexit and Donald Trump, like everyone else.  The CEO isn’t happy about this, so she asks the BD manager to prepare a full briefing on the strategic options for improving their role and relationship in African Markets.  After this, the BD manager goes away and runs through a few options.

Option 1: Expert Literature Review

There is the ‘tried and tested’ option; commission an expert on Africa to produce a paper that tells them a range of strategic issues.  Once the paper has been delivered (probably at considerable expense that directly relates to the urgency) someone in the BD Managers team will condense them into a powerpoint presentation, perhaps with a detailed report of research that they can reference if challenged.  Is this good, is this bad?  Well, it’s good as it does give you answers that can be put back to the board (arguable in a linear, bulleted powerpoint format).  This traditional approach also suffers from limitations as it depends on the scale and the process through which the data has been assessed (often the biggest value has been given to the analyst who compiled the report and learned the associated knowledge in its production).  Such reports can easily be biased and often based on a small range of reports that are limited to the number that the analyst can comfortably process in the time available to them.  Also, if its based on a small number of people and papers, the assessment is at greater risk of being biased toward particular issues or outcomes.  

Option 2: Produce a context map  

An alternative option to commissioning a single expert is to produce a context map.  At present,  this does represent a significant cultural change to how many organisations currently conduct their strategic planning.  Context brokering works on the principle that the best thing to do early in your planning, is to define and gather as much data as possible and then summarise what you believe the insights and themes around an issue could be with some kind of qualifier for how valid you think the data could be (relating back to the source data to illustrate where the insights came from).

So going back to the considering the future of Africa, using data visualisation and mapping tools a context map (or topic model) can be produced.  Such a map summarises the data behind an issue and provides a start point for strategy making.  This produces a map that is a lot more engaging and derived from a wider range of sources (there is theoretically no upper limit to the number of reports that can be analysed and mapped, although at present our own experiments at Simplexity Analysis are around 1000 documents). Such outputs are less static than bulleted lists and can be used in facilitated sessions with experts who can then interact with the context and add their own insights as required to further enrich our understanding of the context. Have a look at the one below produced to provide a context for future strategic issues surrounding Africa.


Mapping the data around an issue in this way can be daunting.  What was the domain of traditional research and literature reviews is now increasingly contested with data scientists and analysts talking in numbers and code and arguing in shades of technical purity around who’s process for mapping is more accurate (is it a complete reflection what’s in the data) or quantifiable accuracy (if you take qualitative data, is it worse than hard number predictions)? Perhaps this is why its challenging for decision makers to interact with such new techniques as context brokering does represents a cultural change - the best way of addressing this, is probably to be open and honest in the data used to make the assessment, the assumptions behind it and the limitations in the development of the context.  In the past, that’s what the weight of a large volume of research would convey.  Now, it’s probably the scale of the data that has been analysed.

Which option works best?

Concluding again with the Africa example, what’s better - a bulleted powerpoint presentation of ideas, referenced with a weighty research tome (that, lets face it, few people are going to read).  Or a map, summarising a range of options that can be discussed and assessed by the board, or through associated activities that equate to action for the board to sanction and the associated data available for analysts to reference further as required?

For more information detailing differences of approach for mapping and analysis, please see the following presentation that outlines the differences between conventional analysis and data driven approaches.


1 Comment

Context brokering. What is it and what does it mean for strategy?

Context brokering, is a relatively new term that broadly relates to using data to provide a context around a particular issue (other techniques like topic modelling or ‘concept testing’ are sometimes also used to describe a similar analysis process).  Context is particularly valuable where people, in business or government, need to derive insights and understanding around a particular issue, that can be highly complex and involve a large range of data.  This is where context brokering has a strong link to strategy, which is where someone, generally a leader has to take the data and do something with it.

Forming a strategy, or even a plan, generally requires an understanding of what’s going on.  Having a good understanding of the context allows you to both understand what decisions you might need to make and any assumptions that you could be making.  This really isn’t new.  As an analyst, the first thing everyone tells you to do is start by understanding a particular problem or issue.  To do this we generally start by gathering data.  There are many ways we can do this and the choice of method usually depends on the time and resources at our disposal.  But however we do it, be it from simple google searches through to a detailed literature search, the aim is the same - to gather as much data as we can to inform our understanding of a particular issue or topic.   

So, to form a context we need to first gather data and then decide on our approach on delivering a particular outcome (our strategy).  In the modern, data rich world, this is often quite a challenging thing to do - we now live in a time where it’s not about too little data, but too much.  We continually face questions about how reputable our data is, so understanding how to refine and understand the data is becoming increasingly important.  Traditionally, this used to be limited to how much information the human gathering the data could process.  So, in a way, we’re roughly limited to say around 10 reports of maybe 20 pages a report, perhaps a 100, if you’ve got an inhouse team of people and some analysis processes to help you triage and summarise the increasingly complex research data.  

Today though we are a lot less limited by human processing.  There are many options and dashboard solutions that enable people to gather a lot more data and make sense of what is being said.  Making sense of data is now increasingly important and the range and the scale of the data is increasing.  So in some ways, the data available to be understood is far greater, potentially less biased and not limited to the human processing bottleneck.  But, this creates a new range of issues - how accurate are the processing algorithms applied to them and how and where should the human intervene to select out the most important aspects of the data for context?

And this is the challenge we now face - attempting to balance tools and techniques that allow us to gather and structure more data, whilst providing a useful, accurate and informed context that enables us to make better decisions and form policies and actions.  And this is where context brokering can help, but it can be a complicated process that yields deceptively simple outcomes, so to understand how it’s applied and how it can differ to traditionally applied techniques it’s worth considering an example.  Have a look at this post that explains things a little more.   

1 Comment


A data-driven forecast for the future of healthcare.

Back in November 2014, we produced a forecast that was entirely data derived. Unlike a lot of futures reports, this analysis is based entirely on openly available data and analysed and visualised in a manner that illustrates all of the available data used to derive judgements. We believe this is important as it means we can reduce bias in our assessments but also when we make predictions for the future we can produce quantified assessments to reflect our belief in whether they will happen or not.  We've now developed this analysis significantly and have used it to test ideas and trends in a wide variety of areas, but, if you're interested in the future of health, we've now made the analysis available here.

But, for those of you, too busy to read the actual report, the main 'Top 5' findings are detailed below [caveated appropriately!]

Top 5 trends in Healthcare 2015 - 2050.

1. Fee paying healthcare is likely to increase out to 2050.

Insight - Because of greater demand on health systems (ageing, obesity and disease), the rise of new healthcare markets and strategies (from emerging markets) and increasing technologies and medications to promote and prolong life, fully funded state-based healthcare is unlikely to be sustainable out to 2050.

Judgement - There is a probability of 0.8 that by 2050, countries like the UK will deliver a far greater proportion of their healthcare through private agencies. State-based provision is likely to become increasingly difficult because of the continued evolution of diverse healthcare demands and increasingly complex technical requirements of future treatments. By such a point, states are likely to focus on facilitating access to affordable healthcare and promoting healthier lifestyles.

2. Global obesity rates are likely to increase over the next 30 years, prompting significant initiatives to address them.

Insight - Without coordinated intervention global obesity rates are likely to increase out to 2050. Basic projections suggest that if global obesity continues at its current level, an estimated 2 billion people in a global population of 7 billion in 2013 (contrasting with 857 million from a global population of 4.5 billion in 1980), then by 2050, around 30-60% of the global population will be obese. In total numbers, if the global population reaches 9.5 billion by 2050, this will represent a range of 2.7-5.7 billion obese people.

Judgement - There is a 0.95 probability that the levels of obesity in the global population will increase from 2014-2050. This trend will be driven by higher calorie diets as lower activity levels become the global norm. However, the problem may become so significant, so quickly, that policy reforms, new technologies and medicines may provide the necessary interventions to mitigate this trend.

3. Out to 2050, states are more likely to occupy the role of facilitating healthcare access as opposed to direct provision.

Insight - Over the next 30 years, the rising cost of healthcare and the increasing diversity of technologies and medicines to promote health and prolong life will mean state-based care strategies will be increasingly costly to maintain. This is likely to lead to many countries developing less costly models to promote and facilitate access to healthcare, guaranteeing a level of access to the least well off citizens alone, whilst enabling access (through part funded and tax incentive schemes) to the majority of their citizens. However, due to the variability of national strategies and priorities, there will be considerable variation in the political attitude to toward state-based healthcare.

Judgment - There is a 0.65 Probability that governments will move to roles based on facilitating access to healthcare as opposed to being the direct provider.


4. The use of healthcare data will be increasingly important for healthcare treatments.

Insight - Out to 2050, improvements in sensor technology, data collection and increasingly available open data will drive metric collection and increasingly sophisticated trials and health strategies. Such developments will change many perceptions on the use/protection of health information and patient confidentiality.

Judgment - The use of healthcare data will increase out to 2050. It is a certainty that data (once it has been approved for confidentiality and legal consideration) will be collected and used to improve the quality of human healthcare.

5. Policies to encourage healthy behaviours and lifestyles are likely to become increasingly important.

Insight - To reduce long term health issues government and company policies are increasingly likely to promote healthy behaviors and lifestyles to reduce long term costs on industry and the state. Such strategies will be more cost effective to implement in the long term and reduce the treatment of symptoms rather than the causes. However, certain specific requirements such as the guarantee of basic security and emergency responses to save lives will remain key ‘duties of care’ that will need to be maintained.

Judgment - There is a 0.7 Probability that policies to encourage healthy behaviors will increase over the next 30 years.

Judgment - There is a 0.95 Probability that the duty of care of governments to maintain and protect the health and safety of their citizens will endure out to 2050.


As well as weighting our top five findings we've also collected some 'outliers'. These are the rare, and very low reported trends. When you get all these together they can make for interesting reading, just think, if we'd done this exercise in 2009 - where would the term 'healthcare metrics' appeared?

9 'Outlier' trends for the future of Health

1. The next pandemic may not be flu.

2. Both Japan and the EU may suffer from a shortage of trained healthcare providers in the future.

3. Long term chronic illness (such as diabetes or forms of cancer) could represent significant healthcare issues in the future.

4. Hypertension could be an increasingly significant healthcare issue.

5. The rise of counterfeit medicines and synthetic narcotics could be of potential significance to the future of human health and the pharmaceutical industry.

6. The increased use and sophistication of biomarkers could be significant for addressing future health challenges.

7. Cognitive systems that sense, act, think, feel, communicate and evolve, could be increasingly important in how we understand and improve the healthcare solutions at our disposal.

8. ‘Localisation’ and the local environment could be increasingly significant for how healthcare options are delivered to the surrounding populace.

9. A revolution in farming and agriculture could improve or alter health dynamics anywhere around the world.

Any questions?  Get in touch, at