What factors does an AI take into account when auto-scheduling a rota?

July 31, 2023

When discussing our artificial intelligence auto-scheduling tool, a common question is “what factors does an AI take into account when auto-scheduling a rota, and where does it get the data from?”

Where does the auto-scheduling tool get its data from?

Most of the data that the AI needs is stored in the WFM system. This is often ShopWorks, but our AI is independent of ShopWorks and can build rotas for other WFM systems. Your workforce management software is the best place to access the data for a couple of reasons:

  • There are already processes to keep this data up to date.
  • The WFM system is a repository for digital versions of the business logic that your organisation uses to manage staff scheduling and time and attendance.

The AI does have some data of its own, this is normally limited to “weightings”, “fairness”, and other specific instructions that the AI needs to know how to build the rota.

How does the AI get the data it requires?

When you press auto-schedule, the AI sends a message to lock the rota, this prevents any further changes to the data. The AI then requests a “payload” of data from the Workforce Management system, this is essentially an up-to-date copy of all the data it needs to build that rota.

Once the AI has built a rota, it sends the data back to the WFM system, which then populates the rota and unlocks it allowing access.

The process takes between a few seconds and a few minutes, depending on the complexity of the rota being built.

How does the AI get data for scheduling

Is there such a thing as a perfect rota?

Our AI will build the optimal rota based on your criteria. There may not be a single solution that meets your demand and budget and every member of staff’s preferences, and there may only be a best fit based on the weightings we have used to drive the AI. 

What factors does an AI take into account when building a rota?

The AI needs to (and does) consider all of the factors that a human manager would consider.

what factors does an AI take into account when auto-scheduling a rota

These factors include:

Weightings: This data allows us to configure all of the priorities for the AI. Is it more important that you hit budget than meet your demand per role? If so, we give a higher weight to the budget. For some factors, such as WTD compliance, we will weight it so that no rota can be built that isn’t 100% compliant. Agreeing on the weightings is the key part of our discovery and implementation process for auto-scheduling via AI and is how we ensure that the rotas our AI builds meet your business requirements.

Demand per role per venue: Before building a rota, you need to know how many staff you need by role in 15, 30 or 60 minute slots. This is called a demand curve, and if you are using our Scheduling.Ai, you will probably use our demand Forecasting.Ai as well. Weightings can be set to optimise customer service or manage costs more efficiently. If you use our Demand.Ai, it can produce two curves at a given probability level and the Scheduling.Ai will be instructed to use either the upper or lower curve depending on your preferences. This is not as complex as it sounds, let’s look at an example. Suppose we predict how many staff you need to a probability of 97.5%, our AI could say that there is a 97.5% probability that you need between 10 and 11 cashiers on your checkouts between 10:00 am and 11:00 am on a given Saturday. In this case, if you had asked us to optimise for customer service, we would schedule 11 cashiers.

Minimum staff levels: You may not have enough customer demand for a single member of staff but your rules may say that you need a minimum of two staff on site whenever a store is open. The AI needs to know this.

Budget: The AI can pick up your staff budget from ShopWorks and can compare the total cost of a rota it has built against your budget.

Skills: We might need to consider skills before building a rota, i.e. is this person qualified for this role or task?

Authority: Does this person have the authority to do this role, such as assistant manager or shop opening?

Tasks: In ShopWorks, there can be multiple tasks within a shift, and they can be matched to skills. Our recommendation engine will allocate tasks to the most suitable staff members. The Scheduling.Ai has to ensure that you have enough qualified people at all times to deliver all available tasks.

Shifts already booked: The payload includes any shifts that have already been scheduled and won’t overwrite them. 

Leave and absences already booked: Once leave has been booked and approved in ShopWorks the AI will build a rota around them.

Pay rules: the AI needs to know the pay rules, if it builds a shift that runs into overtime or premium payment, it needs to calculate the exact cost.

Contracted hours: our AI is extremely keen to make sure all contracted hours are worked before it starts to allocate overtime.

Employee preferences: Employees can document their preferences in our availability module and the AI will try its best to accommodate these. It also does it in a fair way and has no bias towards any group of employees.

Employee availability: Availability is also held within the availability module of ShopWorks, it is likely to have a higher weighting than preferences, i.e. if the AI is told that a person isn’t available on Sunday’s it won’t rota them. For this reason, many of our customers prefer to have managerial input into employee availability.

Compliance rules (WTD): ShopWorks stores the rules around WTD and other similar rules in the compliance module. The AI is weighted to ensure that it never builds a rota with a WTD breach.

Compliance rules (role-based): The AI can also be weighted to ensure that role-based rules are always followed. For example, a first aider is on-site if the venue is open and trading.

Local rules (Union): Many companies have agreed compliance rules that are in excess of WTD directive regulations which are unique to them. For example, we have a customer who has agreed that each staff member gets two days off per week, and they are always consecutive. The AI takes this into account when building a rota.

Opening times: The AI needs to know when each department opens and closes each day, and it gets this from ShopWorks.

Overtime rules: This isn’t about the cost calculation, more about how much overtime can be offered, how long can a single shift be etc.

Fairness Rules: the AI should allocate unsociable shifts, overtime, hours to zero hours employees and other factors fairly and even. 

Consistency rules: These include starting all shifts in a week within a few hours of each other to give some consistency to the week, such as early or late shift patterns.

Shift length rules: The AI needs to know the maximum and minimum shift lengths.

Bespoke configurations and rules: As you can see from the list above, the AI considers a host of key factors. If you feel we have missed something, we can often add these in at the configuration phase. Each implementation of auto-scheduling is configured and tested to be unique per customer.

Solve Your Biggest Operational and HR Challenges Using Proven AI Solutions that Deliver ROI

You don’t need development skills or data science engineers to implement our solutions. They are quick and easy to integrate with your existing tech stack.

Latest blog posts

Scheduling Glossary

Many Scheduling software tools employ Artificial Intelligence and Machine Learning to automate their processes, resulting in a lot of jargon, acronyms, abbreviations and technical terms. In this scheduling glossary, we...

read more