Culmination of 3 weeks onsite + 2 weeks offsite, larger scale engagement than planned due to the breadth of scope
We have really enjoyed working with AT on this
Now, we are going to present a summary of our recommendations
This is in effect an overview of our full report, with our more detailed findings and recommendations
We have put breaks in the presentation at what we hope are the most useful points, so we can discuss the material while it is still fresh in everyone’s minds
In particular, companies in the top third of their industry in the use of data-driven decision making were, on average, 5% more productive and 6% more profitable than their competitors
Optimise current
New product development
Insight as a product - becoming a pure insight provider to others
Smarter ways of serving the market: putting smarter analytics in the operating systems to improve customer journeys, and honing insight to share relevant data with new insight customers - e.g. manufacturers
Not every organization is ready to be like Google or Facebook. But leading companies seek to become more data guided to validate intuition and judgement.
Lean Approach
assess, validate, fail
Etsy is an e-commerce platform for small artisans to sell handcrafted items.
Data strategies that are driven by IT are focused on infrastructure, platform, and technologies.
When data strategies are driven by the CEO or CDO, they are guided business priorities and value propositions. “How does becoming a data-led company help us become more competitive?”
MB 3-11
Lean Approach
assess, validate, fail
Discovery Track - used to design and test something new, encourage fail fast and change tack approach
Development Track - during development to validate that what is being built or maintained is meeting the outcomes that we expect
As seen in the away day
Discipline of developing a hypothesis BEFORE taking an action, and setting measurable goals for success or failure.
Discovery Track - can go round several ways
for example - historical data, counter-factual estimation, small scale manual experiments, small-scale Experiments through parallel running
Development track - putting this into production and measuring the outcome before scaling; reviewing the performance of the product on an ongoing basis, to feed back into further discovery for improvement
Within each squad, these Experiments test the proposition frequently on a micro-level, reducing risk and the cost of failure. The business domain strategy, as hypotheses - for example for consumer - is iterated quarterly, and the overall business vision is amended to reflect these learnings.
These squad experiments are prioritised in terms of criticality and risk of failure, on the basis of answering the most important elements of the domain hypotheses.
So, the signals from experiments directly inform domain hypotheses, which in turn deliver or amend overall business goals.
So, the business becomes a leaner organisation, by efficiently directing effort at the most important questions to realise the most valuable opportunities.
Lean Approach
assess, validate, fail
The universe of data science skills is very broad, with the value in the mix of skills that cross the whole business - it’s really about how they are combined
current - small group of people who don’t work in the squads to understand the context of their needs and gain frequent feedback
TRYING to build a one-size fits all system to serve everyone
Move to..
Raw Data accessible to everyone, through a variety of technologies, so that squads can organise and publish the data they need
Building data assets, and enabling AT prod and services SUPPORTED by experiments
Move to a situation where operational systems co-operate where they need without integrating with DW
Exploration happens at the lake, and we serve up just what is needed for each system
Start with Product track exploration because it’s easier to think about
Lean Approach
assess, validate, fail
What defines a data-enabled business is one that is able to exploit this currency to deliver exceptional value…
To be successful as a data-driven company, we believe that Auto Trader needs to:
link current experiments more clearly to support an evolving future business vision
be ruthless in prioritising the highest potential value opportunities, and in developing the new capabilities needed to identify and deliver them
challenge yourselves to put hypotheses and measurement at the centre of everything you do, to simply achieve more
Culmination of 3 weeks onsite + 2 weeks offsite, larger scale engagement than planned due to the breadth of scope
We have really enjoyed working with AT on this
Now, we are going to present a summary of our recommendations
This is in effect an overview of our full report, with our more detailed findings and recommendations
We have put breaks in the presentation at what we hope are the most useful points, so we can discuss the material while it is still fresh in everyone’s minds