There have been several reports recently that focus on the slow uptake of AI and project data analytics, despite many in the project management community being outspoken about the benefits AI in can bring.
So what's holding people back, and how can you make a start with using AI for your project management?
At Sharktower, we talk to senior project professionals all the time, and we’re regularly asked the same questions. Things like:
“Our current project data is not good, how can we get any value for data solutions?”
“How difficult are they to implement, how can we create a common understanding of AI?”
“What skills do PMs need to develop in order to remain relevant?”
1. Remove repetitive tasks (so you can focus more on delivering value and outcomes
Mundane manual tasks can take up 30-50% of your time. Scraping data, capturing next actions, copying and pasting, trying to find out what's going on.
Examples of simple automations you can perform with basic project management apps:
- Automatically create tasks from email and Slack conversations
- Collect data from various sources and aggregating them into a single file
- Connect GitHub and Jira for issue management
- Connect Toggl and Harvest for time tracking
- Chase reports and missing data
- Send an email to yourself or your team when a specified action happens
- Add new Calendar events as tasks
Each of these is essentially basic Robotic Process Automation (RPA). You will inevitably have multiple applications in your organisation, so it makes sense to use agnostic tools and experiment with triggers from one system to another, rather than trying to rely on one single vendor.
While not technically RPA, another technique to familiarise yourself with is data visualisation. Power BI, for instance, uses integration and automation to connect multiple data sources and model and visualise your data.
Microsoft Power BI gives you analytics and insights from £7.50 per user/per month
2. Experiment with advanced techniques to improve predictability
There are many areas in project planning that have to be estimated or predicted, and - as with with any estimate - there is a risk of getting it wrong.
When estimating how long a project will take, for instance, there’s always an element of guesswork, especially for new change or innovation projects. The risk is under or over-estimating project completion times. To help mitigate the risk, you probably take the estimates for how long each project task will take and come up with a best and worst-case scenario for when the project will complete.
By using a range of possible values like this, instead of a single guess, you can create a more realistic picture of what might happen, but you still can’t predict how likely each scenario is. Techniques like Monte Carlo analysis can help you do exactly this. This is a type of computational algorithm and you can experiment with it yourself by using resources like RiskAMP’s Monte Carlo Simulation Engine for Microsoft Excel.
A conditional risk model from one of RiskAMP's sample spreadsheets
What Monte Carlo simulation does is analyse all the potential scenarios and give you probabilities on when the project will complete, for example, 2% chance of completing the project in 12 months; 15% chance of completion within 13 months, 95% chance of completion within 15 months, etc.
Like any forecasting model, the simulation will only be as good as the estimates you make. It's important to remember that the simulation only represents probabilities and not certainty. Nevertheless, Monte Carlo simulation can be a valuable tool when forecasting an unknown future.
3. Apply Machine Learning to model project complexity
Machine Learning techniques are excellent at finding patterns and outliners in data, often previously unseen. Using pattern recognition and historical data, along with network analysis of connect dependencies, AI is proven to provide better scheduling and is able to quickly model the impact of different scenarios.
However, projects are complex systems to model, and even the smallest of projects has an endless network of cause-and-effect impacts. In particular, behavioural aspects of different stakeholders are difficult to model but critical to project success.
CAUTION: Many projects have typically suffered from poor historical data which is heavily biased, subjective and therefore unreliable for training robust models. This shouldn’t stop us using the data we have today to gain new insights and indicators, but we need to work on improving project data capture and quality.
Allocate time to learn and experiment, then expand.
Automation is an enormous time-saver, but it can take experimentation to get it working correctly for your needs.
Once you begin to benefit from user-defined workflow automation you can start to go a bit further. AI-driven bots can do more advanced tasks such as understand context, classify documents, mine processes and structure data. Unattended bots can automatically carry out defined processing without your observation, even monitoring your work to suggest further tasks they can manage.
Got a question?
If you're thinking of getting started with AI in your project delivery, get in touch via email and we'll be happy to discuss ways you can adopt a more data-driven approach. Or you can request a 1-2-1 demo of Sharktower.
Get started with a free trial of Sharktower
- Go to the trial request page
- Submit your details (including your work email address - we’ll never share your data with anyone else).
- We’ll get back to you ASAP to schedule a call to determine your priorities and ensure Sharktower can support you at every stage of the process.
Want to know how to make better use of data for project delivery? Read our article 'The project data dilemma: six questions answered'.