All Models Are Wrong (but some are useful)

12th May 2023

Statistician George E.P. Box coined this aphorism in relation to his own discipline, but it’s since been appropriated by almost every field of human endeavour. International Advisor to Global Association Hubs Martin Sirk explores how it applies to association business models, with a little help from his new AI assistant.

The world is a complicated place: we can only make sense of it through heuristics (mental short-cuts) and simplifications, and our perceptions are designed to ignore the vast majority of sensory data in order to build workable hypotheses about what’s going on around us. Model-making is what we humans do, both consciously and automatically, all the time. It takes a shock to the system – a dramatic optical illusion, for example – to remind us that our models are inherently untrustworthy, and need to be carefully interrogated to identify whether, and in what way, they are genuinely useful!

When it comes to describing organizations and the way they function, we need to employ even greater scepticism about models’ accuracy and usefulness. Models can fail by being over-simplistic or overly-complicated, by underestimating decision-making complexity, by lacking diversity. They may focus on the wrong variables, ignore critical external factors outside the model, or be overly static in nature. 

Every time models are used, it’s vital to remember what can go wrong and to put in safeguards to prevent them being used to generate over-confident conclusions. It’s also essential to be really clear about what questions a model is designed to answer: questions about efficiency or competitiveness are likely to require very different frameworks than those about engagement, creativity or inclusion. And at all times, we need to remind ourselves that “useful” doesn’t necessarily mean “correct”!

Association Business Models & ChatGPT

The traditional starting point for discussions about our sector’s business models appears to be “primary sources of revenue”. That’s certainly the case if you ask ChatGPT to generate a list of “association business models”! Alongside “membership dues and fees” as the default option, my AI assistant came up with nine different models that were actually categories or sources of revenue, from “events and conferences” to “consulting and professional services” to “certification and training”. Many association CEOs will be reminded of Board strategy sessions on business models that evolved (dissolved?) into transactional arguments about pricing and share of revenue by category of activity.

We need to do better than this to find genuinely useful models that will help us reinvent the future of our associations, and ChatGPT is an incredible source of inspiration if asked the right questions! A short exercise with some lateral thinking generated 57 additional models*, with many providing unique starting points for analysis, fresh perspectives, previously unimagined opportunities, potential restructuring or novel relationships between stakeholders, and ideas for experimentation. Here are some of my favourite suggestions:

A query about international company business models included the options “strategic alliances”, “franchising” and “licensing”. Innovative business models and those for high uncertainty environments came up with “subscription-based models”, “freemium”, “contingency”, “employee well-being”, and – something every association could consider – “outcome-based models”. University business models included “continuing education” and “consulting-based”, whilst taking slightly more eccentric perspectives, why not consider nature-based business models such as “ecosystem services”, “biomimicry”, and “co-creation”, or even sci-fi business models such as “reputation-based” or “AI-driven”. 

All of these responses are focused on activities and methodologies, and don’t really address how the association is actually structured and the inter-relationships within the model. Asking ChatGPT for “business structures” rather than “business models” overcomes this very neatly: choose from hierarchical, flat, matrix, network, team-based, functional or even holocratic (defined as where authority is distributed throughout an organisation, with decisions being made collaboratively). Another query about “business philosophies” elicited “six-sigma”, “total quality management” and even “servant leadership”, adding interesting new flavours to the smorgasbord of appetizing choices.

Expanding the universe

Using an AI tool in this way doesn’t help an association identify a single “correct” model, instead it dramatically expands the universe of potential models that can be creatively applied to analyze an association’s current situation, to imagine its desired future reality, and to chart a range of options for moving from the former to the latter. Viewing an association and its ecosystem through multiple, widely-divergent prisms is an ideal way to gain new insights, identify new pathways, and encourage a dynamic mindset. 

This avoids a major problem highlighted by Magda Mook, CEO of International Coaching Federation: “The concept of “static model” is just wrong on arrival. We had to change many dimensions of how we financially sustain the organization during the pandemic, and we did this very successfully.” 

Martin Sirk is International Advisor to Global Association Hubs, a partnership between Brussels, Dubai, Singapore and Washington DC, which is committed to promoting the societal value of international associations, and to stimulating the discussion of important issues through events and articles such as this, which appears as part of a collaborative partnership with Boardroom.

An even greater challenge is to encourage all the association’s stakeholders, and especially the board, to embrace a model-based way of thinking about their organization and its place in the world. Tom Reiser, CEO of International Society of Thrombosis and Haemostatis expresses this well: “You may have the best and most sophisticated models but if you cannot bring the board, committees, members along it may not be sustainable and may actually risk that board and staff are not working hand and glove. You need to know your audience and to bring it along on the journey.”  

This is especially true when thinking about global development issues, where traditional business models frequently fail to identify critical variables, especially with regard to cultural factors and unexpected sources of competition or collaboration. That is when it’s important to work closely with partners who have the expertise to help you build customised models that incorporate the most useful local, regional and global dimensions, such as the partner destinations that make up Global Association Hubs – Brussels, Dubai, Singapore and Washington DC!

*The full listing of ChatGPT’s responses to 10 ten prompts on “business models” is available here. 

Hit enter to search or ESC to close