Only Dead Fish

Automation and Wicked Problems

Glyn Britton made a good point in his piece about how AI and automation may actually free us up to be more human by enabling people to stop doing work that can actually be better done by machines. 

'...what if we turn the narrative around? Look at AI and automation as a opportunity to repair, strengthen and deepen the exclusively human disciplines of empathy, intuition, judgement and creativity...'

It reminded me that I'd read last year (in one of the few empirically supported articles on the subject of whether automation actually does replace human jobs, which uses a study by Harvard economist James Bessen) that there is only one of the 270 occupations that are listed in the 1950 US Census that has since been completely eliminated by automation: the lift operator.

Instead (as a McKinsey analysis also quoted shows), whilst few jobs are likely to disappear completely it's likely that automation will replace at least some work in almost all job areas. So as with most things the answer to how automation might progress lies in the subtleties. One thought that might inform this is considering the different types of problem that we may typically have to solve in a  job. Some areas of work are repeatable, relatively simple and unchanging - these are obvious candidates for automation since they are the domain of efficiency and best practice. Other areas may be characterised by more complicated problems that involve more variables - there is likely to ultimately be a significant role for automation here too, although since this is the domain of expertise it's likely to require more sophisticated automation involving machine learning perhaps, and it may be a time before these areas are fully automated. Complex problems however, have many variables but these many variables are also changing all the time, which presents learning challenges that potentially require the most sophisticated AI to solve.

There is, I realise, a fourth type of problem - so-called 'wicked problems'. These are not only complex, with multiple changing variables, but some of these variables may be characterised by contradictory, incomplete or hard to define information and the problem itself may be continuously evolving, thereby potentially enabling a better outcome but denying a complete resolution. It's likely that automation can bring scaled assistance of some kind to solving every kind of problem area, but the nuance of this is in how human and machine are blended, and I do wonder if the real value for those exclusively human disciplines that Glyn talks about will increasingly be in those complex and wicked challenges that characterise so much of our human existence.

The Role for Hyper-personalisation

I don't usually like pre-fixes like 'hyper' as they rarely add much meaning (personalisation is just personalisation, right?) but I think the extent to which Netflix uses sophisticated content personalisation is sufficiently distinctive to justify the hyperbole.

In a recent post on the Netflix technology team blog they describe, for example, how they use data on individualised viewing history to not only personalise the rows of films and shows that are presented to you, the titles of the rows and the content that populates them (using tens of thousands of sub-genres), but how they also use static imagery from films to show different film cover images to different users in order to appeal to the myriad different tastes that they serve:

‘…we don’t have one product but over a 100 million different products with one for each of our members with personalized recommendations and personalized visuals.’

So the artwork chosen to represent a film to you may be influenced by anything from the kind of genres that you’ve shown an interest in previously to the number of times that you’ve streamed films with particular actors or actresses in them. 

It's important that data like this is used responsibly of-course, but what's interesting about the way they talk about this is the challenges that they say this brings - not least that of making the algorithm that controls artwork personalisation work in tandem with the existing one that powers content recommendation. The former has to sit on top of the latter without disrupting its work. To work well, artwork personalisation requires a huge amount of data and a wide range of source material which can make attributing tangible value to algorithm changes really hard. An overly complex artwork personalisation system might impact the load performance of the home screen (Amazon use product combinations in their recommendation algorithms for a similar reason). And then the algorithm has to adapt as more content is viewed, making it an ever changing formula.

It’s a true engineering challenge but one that has brought measurable improvements to how users discover content. And I think there's an interesting thought here about the role for such sophisticated use of individualised content. The debate about highly targeted personalisation vs mass-market messaging in the media and advertising industry is often more polarised than it needs to be - we need both of-course. But we often suffer from the results of poor execution (like most retargeting), poorly aligned incentives and measures (it works well enough for advertisers or publishers not to be so concerned about trade-offs in user experience) and poor thinking (forgetting that if we don't like something, other people probably won't either).

Right now there are a surfeit of examples of poorly delivered personalisation and targeting. But that doesn't mean that personalisation itself is inherently bad. Alongside Netflix for example, I think Spotify's Discover Weekly has consistently given me considerable value over a number of years. So there are already some great examples out there of how sophisticated individualisation of content can enhance customer and user experience by no small degree, but given the availability of the data and the technology these examples are surprisingly few. Put simply, we need to better.

The Reality of How Marketing is Changing

Each year Scott Brinker's amazing work tracking and mapping the changing marketing technology landscape reminds us of just how much the discipline and function of marketing is changing. Last year's survey mapped more than 5,000 separate martech solutions and you would have been forgiven for believing then that the market must surely have reached its peak. Yet this year's survey is just out and Scott has charted no less than 6,829 marketing technology solutions from 6,242 unique martech vendors.

This explosion in marketing technology demonstrates a quite fundamental change in the discipline, and just how far marketing has broadened in scope and scale. As recently as 2011, when the survey started, there were only 150 vendors.

But in amongst the startling headline numbers, there were a few interesting points that Scott noted in his write up. As you might expect, ranking the vendors by valuation reveals a classic long-tail of thousands of smaller solutions providers but its notable that this long tail is also evident in the technology stacks that marketers are building (shown by an analysis of martech stacks from 'The Stackies', the awards that Scott runs).

So it's clear that marketers are becoming adept at augmenting larger CRM, marketing automation, social and CMS systems with more niche services that can integrate well. And as Scott notes, it demonstrates an interesting dynamic in the market - the very low barriers to entry on the supply side (cheap infrastructure, readily available open source resources, on-demand global talent) combined with this expansion of marketing scope and scale into customer experience, content, and an explosion of channels and touchpoints.

It was also interesting that predictive analytics which was represented as a separate classification in last years survey has now been integrated into multiple areas, reflecting how machine learning capability has now been integrated into multiple different services. The burgeoning category of Salestech is also represented on the graphic showing how deeply entwined sales now is with marketing - this widening scope to encompass what might be called customer experience or go-to-market technology where there is little differentiation between the old classifications of marketing and sales is what Scott calls the 'new reality of marketing'. 

IT is no longer the singular channel for technology acquisition and orchestration. Whilst marketing may have one of the largest tech stacks in a business it's clear that multiple functions are now spending big on technology with the average enterprise using over 1,000 cloud services. 

Combined with a renewed focus on data quality (not least due to GDPR), marketing functions are becoming more sophisticated every year in the adept combination and application of a broad range of technology solutions to 'evolve from fragmented silos to a more holistic digital architecture'.

This is no small change.

Marketing is changing forever.

And it's not going back.

The Rise of Zero-Based Budgeting

Zero-based budgeting feels like one of those trends that is potentially pretty significant but which few people are talking about. I seem to be coming across it more and more. It's no doubt sprung from the increasing pressure that marketers are under to demonstrate the value of their investments, and the increasing accountability that comes from better data and analytics capability.

Rather than using more traditional budgeting methods in which previous performance, sales, revenues or costs are used as a base to extrapolate future budgets, ZBB starts from scratch with a zero base and no assumptions, and effectively involves re-evaluating and therefore re-justifying every line on a budget. Programmes or activity which have been justified then form the basis for the budget, rather than an incremental increase from last year’s numbers, and it assumes no expenses to be pre-committed or balances to be carried over.

ZBB is seemingly a response to pressure for growth and efficiency, and the need for organisations to examine every financial investment with scrutiny, but it carries with it some notable advantages and disadvantages.

On the plus side, ZBB might be regarded as a more accurate and efficient way to allocate resources, particularly in the context of rapidly shifting environments. Using last year’s budget as a base can often lead to hidden assumptions and inaccuracies, and a failure to properly take account of changed circumstances. Since ZBB effectively questions the need for everything, it forces marketers to re-evaluate all expenditure and assess the value of every piece of activity, ensuring that surplus, low value or unnecessary spend is identified and cut or re-allocated. It might also lead to the discovery of more efficient ways to achieve goals, prevent unnecessary budget inflation and it involves staff in the budget process.

One of the main disadvantages of ZBB however, is often cited as the amount of time and resource that can be dragged into the process. Since it ensures that every budget line is considered, ZBB can be a highly time and resource intensive process, and potentially lead to staff feeling as though they need to repeatedly justify every element of expenditure. This can be somewhat frustrating if not done in the right way.

This feels like a significant shift, and the fact that ZZB is a growing trend in many organisations and marketing teams means that the impact is also potentially widespread. It could mean that suppliers find it harder to secure longer-term relationships when the value that they bring is questioned so regularly. It may also mean that it's more difficult for marketers to protect long-term investments and in some circumstances, it could potentially lead to projects or innovations being killed off too early because they have yet to demonstrate sufficient value. But then the counterpoint to this is that when things are moving so quickly questioning everything that you do is not necessarily a bad thing.

Either way, I'm surprised that there's not more debate about this potentially significant shift.

N.B. I'll be talking about ZBB in my latest Digital Shift trends webinars for Econsultancy, scheduled for Thursday 19th at 9am and 4pm UK time.

Image source

Agile Resourcing at ING

In the book I talk a lot about how agile thinking and principles (and not just Agile doing) can be applied in the service of supporting fundamental change in the organisation. To enable a far more responsive, customer-centric way of working. To enable a shift away from the entrenched processes and practices that are holding companies back in the digital-empowered age. Everyone gets how important change (do we call it 'digital transformation' anymore?) is. What most businesses that I talk to really need is pragmatic application.

One of the foundational elements of this application is resourcing. The principle here is that small teams, empowered by the right ways of working and new technologies, can drive big change. My approach with many of my clients is echoed by Eric Ries's phrase 'Think big, start small, scale fast'. You always need to begin transformation programmes with a compelling vision of the company that you want to become, with advocacy from the top of the business, with sufficient impetus to create a positive sense of urgency and kickstart the shift. But the process of transformation to becoming more agile should itself be agile. So you start small, with focused programmes tackling critical business challenges or customer problems, using small multidisciplinary teams to iterate and innovate and build new value but also new ways of working. And you of-course scale fast - using the learnings you get as you go to expand this approach, these ways of working, and to influence culture change, and support a shift in mindset. The number of small, multidisciplinary teams solving customer problems and working in agile ways increases as the programme for change gains momentum, maintaining a good balance between the need to deliver business as usual whilst still generating completely new models and value in new ways. 

A number of large organisations (including some that I work directly with) are utilising this approach to resourcing as a centrepiece of transformation (digital or otherwise). ING Bank is a particularly interesting example since they have been doing it longer than most (their programme has run for several years now), made a significant commitment to agile working from the outset, and so are probably one of the most scaled examples of how this works. Based on the famous Spotify model of Squads, Chapters and Tribes, they have organised in a way (at scale) that enables end-to-end ownership of customer propositions and journeys, focused delivery against customer objectives, heightened levels of employee autonomy and engagement, and organisational responsiveness. I've written up a more detailed case study of what ING are doing here. But they are not the only ones.

In all the talk about transformation there is precious little discussion and insight into the practical application of new ways of thinking (unless, that is, you hire McKinsey for a boat-load of cash). In all the focus on customer experience, and the horizontal challenges that that brings, there is precious little talk about how we need to rethink approaches to asset optimisation and resourcing to tackle these very real problems. I think there are critical opportunities that companies have to do this in the right way, and some very real difficulties that can arise if they don't. Not least is a sensible approach to where you apply these new agile ways of working and where you don't - characterised by the three domains that I talk about here.

But it's hard to under-play how important this. This is potentially one of the most significant shifts in organisation design in decades. I'd welcome a wider discussion and of-course, your own observations and ideas. 

Don't forget there is much more about agile transformation models, thinking and practical application in my book.