Here are five pre-strategy and organizational discovery steps to incorporate that will help build a fit-for-purpose AI strategy for their organization. It is recommended to lean on each department for each of the steps to build a more holistic view of where your organization stands and identify practical concepts for your strategy.
AI in practice for most companies has fallen short of the hype. Why is that? It comes down to companies adding AI tools without strategy, or a loosely constructed pseudo-strategy. Most companies are attempting to keep up with the AI adoption velocity within their industries, circumventing the natural skepticism when rolling out significant change. Now that the initial wave of aggressive AI campaigns and word-of-mouth pressure has exhausted most industries, CEOs should be considering building more effective AI strategies to fit their operations. Consider the last 18-24 months as the experimentation and test phase.
Here are five pre-strategy and organizational discovery steps to incorporate that will help build a fit-for-purpose AI strategy for their organization. It is recommended to lean on each department for each of the steps to build a more holistic view of where your organization stands and identify practical concepts for your strategy.
As you start to examine potential AI use, keep the following in mind:
Focus on Simplicity and Practicality
The benefits from leveraging AI start in the “simple”, while poor AI strategies are often misconstrued by attempting to identify complex tasks to throw AI at. As you start to build a strategy that fits specifically with your company, hone your focus to simple, and practical, tasks. For instance, try to find items where AI can save 5 to 10 minutes per week, rather than dreaming of how AI could power new market offerings.
Let Them Experiment
AI use runs rampant at every organization, regardless of any official implementation or general AI policies. Even with no AI strategy, this is a great initial step toward impactful AI use across an organization. It starts the process of competence and adoption. You need all levels of your company to build comfortability with AI.
Managers Don’t Have Answers
AI tools are often most impactful on operational activities by the front-line workers, whether they are administrators, software developers, analysts, or others. Building a successful AI strategy will come from insights and detail below department managers, by those executing tasks and creating resources. It is recommended to facilitate through departmental managers, but to generate feedback and input from those being managed.
One Major Risk During Experimentation Phase
One key item business leaders must recognize is protecting proprietary or sensitive data being used by your employees in the public versions of the AI tools. The paid licenses often isolate and protect the AI inputs from use outside your organization, or at least have an option to deactivate sharing of company information with their AI models. However, the public versions may use anything your teams them in training their models. What does this really mean? Don’t let your accounting, finance, engineering, or software development teams use the public versions of Copilot, ChatGPT, Claude, Gemini, etc.
To pair AI with each unique department, it is recommended to gather initial feedback from each division and how they are utilizing or what they may have identified as viable use cases. Internal surveys, conversations or other feedback loops are great to facilitate and aggregate these insights. You may be surprised by the span and scale of the current AI footprint within your organization.
Sample survey questions for all employees:
Sample manager-level questions:
As you’ve empowered your team to execute their roles, departments and people built their own processes to accomplish relevant tasks. These workflows naturally evolve in a myriad of forms and structures. Before leveraging AI with your workflows, you’ll need to map out where AI can be injected.
Instruct every department to create an inventory list of 10-15 scheduled and sporadic repetitive tasks. As your departments define these processes, request each task be evaluated on whether any steps within the process could be replaced by an AI tool. Again, focus on finding 5 – 10 minute time savings, rather than replacing whole processes. To go even further, request departments select a subset of these workflows where they have identified AI might act as a beneficial replacement, and have them map out or diagram those processes.
Sample process inventory list items:
A simple example might be for the HR department to initiate the process to document and approve an unpaid leave request, starting with a manager asking for the current form to use. Mapping this process could identify an AI use case to automate the manual task for the HR staff of attaching and emailing the form to the requesting manager, saving 5-10 minutes per instance.
The core benefit presented by AI is the ability to circumvent manual activities via “intelligent” computing power. Survey your departments and staff with a few simple questions to identify where AI could potentially alleviate time intensive items.
AI tools are not free, and the increase of software licensing each user is not nominal when billed per user / per month. Selecting AI tools should be determined by department or role, and even a quick spreadsheet-based SWAG estimate will provide direction on the operational expenditure (OPEX) impact. Included below are a few data columns to start your estimate, as each department may use different tools or some users may need multiple AI licenses:
Does every customer in your CRM exactly match those in your accounting software? This is the one area that stifles the “magical” aspect of AI everyone was hoping it would help solve complex problems for their company or division. However, AI models were not built with input from your company’s unique and nuanced structure. AI is not smart enough to effectively relate data between siloed systems that do not have a clear-cut relationship.
Ask your IT team whether your company has a Common Data Model (CDM)? From the technical angle, the term Common Data Model (CDM) is used to describe when multiple sources of data contain specific data types and labels to relate data from these different sources. In short, all data across your software some similar structure, so it can be used together.
Sample scenario requiring alignment for customer data:
As you can see, a single customer in the above scenario could have non-matching entries across the system. Per the scenario, the same customer could be listed in the accounting system as “Steel Performance, Inc.”, “SPI” in the CRM, and “SPI-TX” in the shipping system. Again, AI is not smart enough to understand the nuance of your business.
If you want a future where AI tools will aid in reporting, analysis and forecasting between systems, you will need a path to define and implement a CDM across your data systems at the macro level to prepare your data and systems for interoperability. The purpose of this section of this guide is to identify the gaps, as implementing consistency or building mapping tables across systems will require technical implementation.
Now that you have some more detailed understanding of the environment, defining the target outcomes from AI are tenable and a level of ROI can be defined. Again, focus on the simple items first and let your organization’s use organically evolve to more complex initiatives, or emerging opportunities. Focus on the benefits from creating 5% to 10% more capacity within each department. All AI-powered resources and options are maturing, providing greater applicability to business operations for any-sized organization, but developing a fit-for-purpose strategy requires context. Hopefully, these 5 pre-AI strategy actions will help build that context to build your initial strategy and roadmap to harness the growing power of AI for your business.