Developing an AI Strategy: Analysis & Synthesis
At Veratai, we follow a simple but well-refined recipe for developing AI Strategies called Discover which is designed for SMEs and business units within larger organisations. Discover consists of three tasks: Stakeholder Interviews, Research & Ideation, and Analysis & Synthesis.
This post is the third in our series on developing such an AI strategy for your organisation. The first covered Stakeholder Interviews; the second covered Research & Ideation. This - the final entry in the series - describes the Analysis & Synthesis task.
Where are we in the process?
With the research and interview stages complete, you are in possession of a “longlist” of ideas. These will range from the quick and tactical to significant programmes for change or investment. They’re not fully fleshed out. They are too many. After deeper analysis, not all will be viable. You should, however, be confident that they are both diverse and comprehensive - a consequence of your broad approach to interviews and research.
But a list of ideas does not a strategy make. There’s still work to be done!
Turning ideas into initiatives
It’s time to flesh out the ideas. You’ll do this by turning each of them into an initiative - a concise description of an activity that the business could choose to undertake. The best way to do this is to fill a table with the following information:
What is the “elevator pitch” for this initiative?
What questions or uncertainties do you still have?
What key risks and roadblocks can you foresee along the way?
With reference to what will change, how do things look today?
Is the business well-placed to do this, or not? Justify your answer.
What does the end state look like? (Describe how the business is changed as a result of committing to this initiative.)
Be clear as to whether this idea is this about seizing an opportunity, mitigating a risk or improving efficiency or quality.
Is there a competitive or economic push towards doing this, or not? (If not, there will have to be a “pull” from within the business instead.)
Therefore, what will need to happen to get from today to the end state? Describe a simple and logical sequence that takes the business from A to B.
When the initiative is complete (or in a steady state), what business benefits will be realised? If you can quantify these, then do. If you cannot, stick a placeholder into your document.
It can help to colour the text and add comments to highlight all of the areas of uncertainty and all your unverified assumptions, since you are going to focus on these in the near future.
With the longlist of initiatives well described, you need to filter them and knit them together. To help you do this, you’re going to need one more tool: the “big picture” view.
Painting the bigger picture
At this point, it’s important to take a big step back and ensure that you consider the whole forest - not just the individual trees. The great strength of the focused and detailed way you’ve analysed the business to this point is that your recommendations are going to be highly relevant and well contextualised. The downside is that, as yet, there is no coherency to them. There is also a risk that a significant avenue of exploration has been overlooked.
To resolve this, you’ll want to pause your work on the initiatives and paint the bigger picture within which your strategy will be executed. We normally use a SWOT (Strengths, Weaknesses, Opportunities and Threats) analysis. SWOT is perhaps the mainstay of corporate strategy techniques (and this is not a tutorial on the process - there are many excellent online resources for that) but there are others too: Porter’s Five Forces or the PESTLE analysis, for instance. Each of these tools offers a structured way for you to develop your understanding of the broader business and economic context.
Whatever strategic thinking tool (or tools - several will always be better) you choose to use, your goal is to take the following exam question: “What will AI mean for this business in 3-5 years time?” and use the knowledge you’ve gained to date to project the impacts of the fundamental trends in a structured way.
When you’ve completed this analysis (or analyses), you’ll want to compare it to your initiative list. Specifically, do the following:
Try to explicitly connect the initiatives to the trends, drivers and levers in your strategic analysis. Ask yourself, for example, “Is this initiative a proactive response to this trend?”, or “Does this strength give the business an opportunity to be first to market with this initiative?”
In each case, write specific statements to explain the connections you see. Do this in such a way that the logic and assumptions behind them is clear. This is because you want them to be critiqued and challenged in the next step.
Identify any gaps - for example business weaknesses without any corresponding remedial initiative, or contradictions between your initiatives and market forces. If you find problems, return to your research and try to generate variations on the portfolio of initiatives that will resolve the issues. (There’s no harm in going back to the stakeholders you interviewed in order to chew over these problems with an expert!)
If you’re really struggling to reconcile the initiatives with the strategic analysis, this is an important signal that you should not ignore. What it probably means is that AI is driving fundamental changes in the world to which the business is not equipped to react. If you find yourself in this position, we’d recommend that you raise the issues with your business sponsor and have a frank conversation about them. It may be that both the scope of the strategy and the business’s appetite for change will need to expand significantly.
Stack ranking the initiatives
Assuming you’ve achieved a satisfactory reconciliation between your longlist of initiatives and your chosen strategic framework(s), it’s now time to begin the filtering process.
To begin with, you’ll want to stack rank the ideas yourself. To achieve this, consider the following four aspects of each initiative (which are presented in priority order) and sort them into a list:
Are there multiple, strong connections between the initiative and the most important findings of the strategic analysis?
Are the business benefits of the initiative clear, strong and incontestable?
Can the risks be controlled and the key unknowns resolved early in the initiative’s lifecycle?
Do the costs feel acceptable, given the benefits and do the roadblocks feel surmountable?
Shortlisting the initiatives
You’re now ready to test the thinking behind the initiatives.
This is a good time to schedule a check-in with your sponsor. You will also likely want to speak again to some interviewees about specific initiatives. The objectives of these follow-up meetings are:
To seek challenging but constructive feedback on every aspect of your thinking.
Get a stack ranking (or at least a “top three”) of the initiatives from each stakeholder - along with a justification for their choices.
To get key stakeholders to confirm or re-state the likely business benefits - especially any quantitative elements. (Wherever possible, use their assessment of the likely benefits in your final report, not yours.)
Some pointers for these conversations:
Share your strategic analysis. Does it pass muster?
Present the top ideas from the long-list and ask for a critique. Are the assertions you make strong? What questions and rebuttals come up? Which ideas get people excited? Where you get challenges to your thinking:
Dig into the challenges; find out as much about them as you can.
Then ask, “How else could we achieve this?” - or “What changes would solve these issues?”
Don’t let an initiative drop out just because it is problematic. Try to think of variations that will overcome the challenges and objections.
However, if the logic behind the benefits is faulty, it’s a candidate for deprecation.
Turning initiatives into a strategy
With your initiatives critiqued and improved, it’s time to construct a portfolio that comprises a strong strategy. This step of the process is a bit like cooking a winning meal from a cupboard full of ingredients: the correct selection will combine harmoniously and the sum will be greater than the parts. Our process for this step is as follows:
Start with the best ideas. Prioritise the ideas that both you and your stakeholders ranked high and where the thinking is well developed.
Strategic coverage. Make sure the portfolio addresses the most consequential findings from the strategic analysis.
Reconcile inconsistencies. Then ask yourself: are there clear inconsistencies in the short-list? Inconsistencies could be different underlying assumptions, conflicting end-states (for example, assuming different directions for a product) or inconsistencies in methodology (for example, one idea assumes outsourcing a capability whilst another one assumes building the same in-house). If you find issues, you’ll need to resolve them - somewhere, there will have been deficiencies in the process.
Identify synergies. Check whether there are shared dependencies, synergistic capabilities or competition for resources. Each of these is actually a good thing (even the last one) because it’s telling you that there is harmony between the initiatives. You may need to resolve such conflicts with clever sequencing.
Extract milestones. Break the initiatives down into their major milestones. (Where you’ve identified shared dependencies or capabilities, their establishment makes for a great milestone.)
Articulate benefits. Finally, connect the milestones to the key benefits they deliver on a visual timeline.
Defining the vision
With a portfolio of mutually reinforcing, impactful initiatives in place, take a step back and look at the portfolio as a whole. There should be some clear themes that jump out. These themes are the backbone of the “vision for AI” that you’re going to articulate. The vision is a succinct (single slide) articulation that explains what the business is going to do and why it’s going to do it.
To help clarify this statement, ask yourself what the business looks like when the shortlist is all achieved. What key themes cut across the initiatives? (For example, is the company investing in it’s own AI technology, outsourcing, or partnering?) There should be a consistent logic behind the portfolio - if it feels like a bunch of disparate ideas with different routes to success, then something is wrong. You should be able to demonstrate that they’re mutually supportive and how that key capabilities developed along the way become central to all of them.
Now write the vision down. Make it exciting.
Outline business cases
Traditional corporate strategy often stops short of building business cases for specific initiatives. We have found that this extra step is greatly valued by project sponsors because it supports them in taking concrete decisions off the back of the strategy.
At this stage, the business cases are going to be rough sketches. If you find yourself modelling discounted cashflows and revenue trajectories, then you’re probably adding spurious detail. Instead, we’d recommend keeping things simple. For each milestone, consider the following factors:
Timelines.
Resourcing requirements.
Who needs to be involved.
A “rough order of magnitude” for costs.
The key risks and unknowns (which you already have).
Benefits, roughly quantified. (Remember, you use your stakeholders’ language and articulation of these!)
Articulate the first steps
And now, the super-important step: give your project sponsor a clear recommendation for the first step(s) they should take. Make it really easy to put this strategy into action. To do so, take the first milestones and ask yourself: what is the minimum, meaningful step towards the milestone that will yield a tangible result? Design this “MVP” with three things in mind:
It should be quick and cheap to do with no barriers to getting started.
It should result in something demonstrable that others can see, use and socialise.
It should directly address the biggest risks or uncertainties and provide vital information that clarifies them.
Compile the final report
With all the above in place, you can write the final report. We think the key sections are:
The position statement: how the business views AI and what it means for its future. (Think of this as an “elevator pitch” that you can use to respond to the question: “So what are you doing about AI?”)
The vision: how it will use AI in 1-3 years time (no longer!) when the proposed initiatives are complete.
The journey: a milestone overview demonstrating how the business will get there.
The deep dive: explain the milestones: what benefits they unlock, what capabilities are being developed, how they are mutually supportive and what the big questions need to be answered along the way.
The business cases: explain why this portfolio makes economic sense and what will it mean to commit to it.
The first step: what should the business do today?
Why this process works
And that’s it: our process for developing AI strategies. It seems like a lot of work, but we typically get all of the above done in a month, or less.
To summarise the three blogs in this series, here are ten reasons explaining why we find this process works so well:
By getting into the detail of the business during the interview stage, you develop a deep and specific knowledge of key processes, problems and needs.
You have agreed the likely benefits with your stakeholders before producing the report.
The vision is a comprehensive response to a strategic assessment of the current and future impact of AI on the business.
There is a harmony and mutual reinforcement between the initiatives in the portfolio.
You have painted a clear picture of how to achieve the vision.
You are able to evidence the reasons that each initiative should work, based on your deep dive research.
You are clear and up-front about the key uncertainties, and have designed a first step to address some of these questions.
By stress-testing your initiatives with key stakeholders, you are confident that the proposed portfolio will have the necessary support.
You have provided outline business cases which enable decision-makers to consider the investment requirements and projected benefits, enabling them to take grounded value-based decisions as to what to progress with
You have given a detailed “first steps” brief so that they can get cracking
Final words
So that’s it. Three blogs that cover everything we’ve learned about developing AI strategies. Hopefully it’s been communicated in enough detail that you, dear reader, can embark on your own.
If it all sounds like rather a lot (and it should take several weeks to complete this work), consider starting with a half-day workshop. Gather your key stakeholders together; bolster their number with at least one AI expert and mediate the process of developing one or more strategic analyses. You’ll be surprised at how many ideas and questions are generated. By the end, you should not only have a great launch-point for a strategy project, but you should have the buy-in and excitement of the decision-makers who will have to put it into action.
We wish you the best of luck!