Because there’s no better way to start than a quote from one of the most iconic athletes of the decade, as Lebron James once said:
“You can’t be afraid to fail. It’s the only way you succeed. You’re not gonna succeed all the time and I know that.”
In the age of unprecedented information growth and data-driven businesses, it appears most Business Intelligence (BI), Artificial intelligence (AI), Corporate Performance Management (CPM) and other data-related projects will fail, or won’t fully reach their objectives.
Keeping in mind failure is a risk one can only mitigate, the question is:
How can organisations maximise their chance of success when selecting and implementing technology, and in the unfortunate event it goes south, how can they leverage failure to later generate higher value?
1- Engage with all key stakeholders
All data-driven projects, from BI to AI & Machine Learning (ML) aim at breaking silos of information within the business, consolidating data from various sources, building a 360° view across all systems, providing more context to support business-driven decisions. However, this is often seen through the technology lens and organisations also often need to break silos of communication. This can only be achieved with the support of senior management.
The Human factor is obviously critical for success/failure and must start from the very beginning. Engaging with all parties can be compared to an internal sales process:
Create win-win relationships
The first question that needs to be answered is: “What’s in for me?“
You want the Sales team to stop using spreadsheets for their monthly forecast because it’s error-prone, inefficient , and forces the Finance department to spend hours in manual data processing? Well, it’s only going to work if you also understand how it can provide value to them: helping negotiate better deals, answering customers’ questions faster, making more commissions… and balance with the perceived cost of change.
The Finance team needs a flexible budgeting tool because the planning solution managed by IT is too complex for ad-hoc analysis, what-if scenarios and instant queries (especially in times like monthly re-forecast or closing)? How can it be implemented with the support and in conjunction with IT, complementing their effort & investment rather than running under-the-radar, parallel processes?
Many Artificial Intelligence projects failed because they couldn’t link operational and strategic outcomes, provide clear value & ROI to front-line managers and employees.
Identify roles & responsibilities
Mapping key stakeholders, influencers, decision-makers, supporters, detractors… is vital not only to get the project started, but also to keep everyone engaged and accountable for the project outcome. I have seen projects where everyone from IT, Sales, Finance departments, both at operational and management levels were extremely engaged and excited at kick-off, but two months into the project failed to deliver due to the lack of clear responsibilities and accountability.
Common personas in successful projects are:
- The Champion(s), who will get their hands dirty and learn how to translate the business needs into a technical language by mastering the technology. They will be the main interface between the organisation and the technology vendor/partner.
- The Enabler(s), who will create bridges between all stakeholders, keep them engaged and mobilise resources throughout the project.
- The General(s), accountable for the project outcome at a top-management level with the authority to make decisions and keep a strategic focus.
- The Engineer(s), in charge of keeping the project compliant to the company’s IT standards & policy and providing the technical infrastructure.
- The Customer(s), (I am referring to internal customers here: the end-users) who must provide consistent feedback and communicate back on how the project is adding value (or not) at an operational level. Their voice is not always audible or heard but determines the long-term success of any implementation.
2- Define clear outcomes
This can be a tricky step, because the impact of a BI/CPM/AI… roll-out is not always easy to quantify, i.e. “You don’t know what you don’t know”.
Measuring success is comparing the difference between where you were, where you are now and where you are planning to be.
It all starts with a purpose
Projects start with challenges to overcome. Artificial Intelligence and Machine Learning initiatives have a purpose because they will help increase customer satisfaction by 25%, automate the reconciliation process so the finance team can support the CFO with more frequent & in-depth reports, reduce risks of hospital acquired complications by half, lower inventory costs… Not because it can beat a human at Go or looks good on a resume.
Some organisations adopted AI and ML for the wrong reasons, or failed because they lacked to clearly articulate the use case correctly.
Like a new high-tech spaceship, BI, CPM or AI technologies only have a meaning when you know your destination and what the steps are to get there.
A double-edged approach
Success can be measured in both quantitative and qualitative measures. Automating the budgeting process might not increase productivity by an expected 20% ratio or dramatically change the outcome from what it was in the past, however, will help your Finance team spend more time on providing better-quality recommendations, prevent them from calling sick or burning out and improve employee retention. Providing high-tech, fancy instant visualisations and nice-looking dashboards, but less flexible than old Excel spreadsheets might actually bring low value to the end-users.
Short term outcomes and low-hanging fruit is tempting, but must also serve long term goals (Some companies still invest in legacy tools to solve immediate issues, which can hinder long-term initiatives) We can also measure value and set priorities by factoring Risk, Cost, Complexity…
Defining clearly your criteria to measure business outcomes and adjusting them constantly is your compass to success (or failure, which, in this case, will help you learn and adapt).
3- A project is not a one-off
Implementing a new BI platform is like learning a new language: you can only improve and realise how much value you can get out of it. Don’t think it will be over anytime soon: it does not end with the delivery of the initial scope but must build new habits, initiate new projects and raise further questions.
Walk before you run
We all know Apollo 11, but landing on the moon was only made possible because of previous Gemini, Mercury and Apollo missions’ successes, and tragic failures too. (If you like space and lessons of persistence, the Soviet Venera program also speaks for itself…)
Back to the good old motto “Think Big, Start Small, Deliver Quickly”, it is not about technology anymore, it’s about managing change, driving transformation across all parts of the business step by step. And rather than “change” which is a one-time concept, we should switch to a more accurate one: “permanent adaptation”.
It also means no failure is permanent, and not achieving success in the first place does not imply you can’t achieve your goals in the next phase. Preparing for failure is:
- Taking acceptable risks
- Monitoring and documenting your journey
- Learning and adapting from your experience
Don’t be dependent
The relationships between internal and external stakeholders is also a journey. I like to relate it to the dependency cycle of Katherine Symor, later adapted by Vincent Lenhard, where relationships tend to constantly evolve from:
- A dependence stage: The traditional, old-school BI approach where the business depends on IT or an external third party to provide them with analysis.
- A counter-dependence stage: Finance, Sales… running their own parallel spreadsheets in reaction to the lack of flexibility and autonomy, without complying to IT processes (Excel)
- An independence stage: Both processes co-existing separately with no synergy, but it is inefficient and the risk is that this doesn’t meet corporate expectations. Silos still persist.
- An interdependence stage: IT and the business collaborating and leveraging each-others’ efforts (Self-service BI)
This means the way you manage internal stakeholders must change over time, but also how you engage with third-parties (vendors, partners, consultants…) from pure technology experts to trusted advisors who know your business/industry and provide value at another level without unhealthy dependence (technical, IP, commercial…).
4- Keep an open mind
Selecting a solution is not a 100% rational process (we’re just humans after all). But it can get close to it.
We could list all cognitive bias we are subject to (Have a look, it’s very insightful and see if you can relate), but let’s focus on the ones we encounter often in a BI/CPM or AI project:
What worked somewhere else might not work here
Starting with the Law of the instrument: “An over-reliance on a familiar tool or methods, ignoring or under-valuing alternative approaches. – If all you have is a hammer, everything looks like a nail.”
What worked in another environment, even very similar, might not always be applicable to the current one. Keeping an open mind means considering alternative solutions with a scientific approach and keeping out from personal bias.
Try before you buy
Would you buy a car without trying it first? (You might, but it’s taking an unnecessary risk)
Technologies are more and more flexible, agile and easy to use. Most vendors offer a trial version and there is no better way to gauge potential value by running a proof of concept.
Not only it will provide a better look and feel of the final outcome, but it is also a good way to evaluate the quality of your interactions with the vendor or partner. The difference between a POC and a full project is merely the size of its scope.
If you need assistance navigating through Data Management, Business intelligence, Corporate Performance Management or AI & ML solutions, please feel free to reach out to us!
Written by Olivier Bastard – Account Executive at AtoBI
Get in touch with Olivier