Cognitive Rosters allows you to use valuable data from your business to predict, build and fill rosters with the click of a button. Cognitive Rosters follows on from Predictive Workforce, with Tanda building Vacant or Filled Shifts for you.
Cognitive Rosters is currently an Add-on Feature in Tanda, so will need to be enabled in your account before you can use it.
Vacant vs Shift Filling
When you ask for Cognitive Rosters to be enabled, Tanda Support will ask you which type you want enabled:
Cognitive Creator: uses the Predictive Workforce Shifts to build Vacant Shifts on your Roster
Cognitive Creator - Shift Filling: Use the Predictive Workforce Shifts to build and fill Shifts with staff names on your Roster
The only difference is Shift Filling will automatically assign staff to the shifts built on your Roster. You can switch between the two at any point, just let us know.
Once it's enabled, you can ask Tanda Support to help set it up for you, or you can follow the instructions below:
Connect and Configure Sales & Revenue Data
The first part of setting up Cognitive Rosters is to ensure you have connected your Sales & Revenue data. If you haven't done this already, the following article will step you through how to do this: Import your Revenue Data
The next step is to configure the imported data. You may have already done this if you've been using Predictive Workforce, otherwise you can follow the two steps below:
Go to the Data Stream management page, located under Settings > Add-ons > Sales & Revenue > click the blue link "Data Stream":
1. Data Stream Cards linked to Teams - this is explained in the Import your Revenue Data article, however specifically for Cognitive Rostering you just need to ensure you have Data Stream Cards linked to your Teams that you want to roster against:
In the example above, we have Restaurant Bookings and Sales data feeding into Tanda and we have linked them with our Team: Front of House.
2. Set Rostering Ratios - this tells the rostering model how many staff you require per data source, set in 15 minute increments.
In the example below:
- For every 1 table booking in 15 minutes, we want 1 employee rostered or
- For every $500 of sales in 15 minutes, we want 1 employee rostered
You need to have at least one Data Stream Card with a ratio set, however you can adjust the figure if it's not quite right once you're in the Roster.
Check Predictive Workforce is now enabled
Once you have configured the steps above, go to Rosters > Current Roster:
In the sidebar, select Day view and select a day:
In the sidebar again, ensure you have ticked the Team(s) that you linked your Data Stream Cards to. In this example, we only linked data to Front of House, so we only have that Team ticked:
You should see Predictive Workforce appear in the middle of the screen:
At this point, if your Roster doesn't show like it does above, you might not have configured your Data Stream Cards correctly or set your Ratios. Revisit this article, or contact Tanda Support on the details at the bottom of this article for assistance.
Configure Roster Data
Before building your shifts, you can apply some settings to ensure the prediction best suits your business:
Manage Demand Data
To the left of the Predictive Workforce box, you have Manage Demand Data:
This allows you to set which historical time data is used to assist in forming the basis for the prediction. You can either select an Average of a particular Day and the number of weeks or a Specific Date in the past.
- If you use the Day average, below this you can then set how many weeks worth of data you want to use to make the prediction.
- If you use a Specific Date, you can use the calendar date picker. This is useful if you want to use an event where you know it was busy to model against the current roster i.e State of Origin nights, Boxing Day etc.
Move the slider across to the hours that your business requires staff, to only show these times in the Predictive Workforce box:
In the example above, the business is open from 11am to 11pm.
Click the icon to enable this option and you can then enter a percentage by which you think staff numbers will increase or decrease on this day. For example, if you enter 10%, your staff numbers prediction will increase by 10%. You can also enter a negative, if you think your staff numbers will be less than predicted, you would enter -10 in the field:
In the table view of the Predictive Workforce box, next to the Team Required (below called Front of House Required), you will see a settings cog, click on here to apply settings to your business. If you can't see this option, it means Cognitive Rostering hasn't been enabled in your account. Have a look at the top of the article again for how to enable it:
You can apply settings to Staff, Shift Length, Open/Close and some Advanced settings. Each tab has an explanation as to how the setting applies so have a read through to familiarise yourself.
One setting of particular importance is:
- Min shift length: you should change this to the minimum shift length required by your Award. If you're not sure you can contact Fair Work or have a look at the Award document in Tanda under Payroll > Payroll Manager.
Interpret the Prediction
Now you have configured the data you can look at what the Predictive Workforce information means.
Still in the table view, at the top you can see your Data Sources, so for this example we have Restaurant Bookings and Sales and their data split across the day:
You will also see the Ratio that you set in the Data Stream management page reflected here. If you don't think you've set the Ratio quite right, want to change it or play around with how changes affect predicted shift numbers, just change the number and you'll notice the Staff Required number change.
The row below is based on the Team selected in the filter on the left, how many staff are required across the day.
For the example below, at 11am, the business requires 3 staff in Front of House to meet demand based on the imported data:
You'll also see the Build! button. This is another indicator that Cognitive Rostering is turned on in your account, again if you can't see this you'll need to contact Tanda to have it enabled. We'll talk you through building shifts just below.
When you start building shifts or making changes, you'll notice the shift boxes may change colour. If the box is:
- Red - you still need to roster this many staff to meet demand
- Grey - you are staffed just right
- Blue - you are over-staffed
At any point you can see what your predictive information looks like by switching over to the graph view:
Building Shifts: Cognitive Creator
Once you've applied your Ratios and Settings and you're ready to build your shifts, you can click Build! at the end of the Team row:
The Vacant predicted shifts will then be build below:
From here, click on Vacant to assign an Employee to each shift and repeat until you've filled all your shifts.
Just keep in mind, Predictive Workforce is there to guide you, but you still have full access to make any amendments as you normally can in Rosters:
- You can delete shifts, change the Start and Finish times, change break lengths and assign specific break times.
- You can also add in extra shifts. It may update the box above to show in blue that you are overstaffed but it won't stop you from rostering the employee on.
If you have more than one Team, just press Build! at the end of each row.
Building Shifts: Cognitive Creator - Shift Filling
If you have Shift Filling enabled, when you click Build! at the end of the Team row, your shifts will be created and filled with Employees:
As before, you can change any of the information, including swapping out staff by clicking on their name and selecting another employee.
Updated Predictive Workforce
Once you've built your shifts, you will notice the Predictive Workforce box updates.
The graph will update with a solid rostered staff line:
For the table view, you don't need to stress about having all the boxes grey to show you're staffed just right, it's not always possible to roster perfectly, as you might need to roster staff longer than needed to meet Award minimum shift conditions.
You can also hover over the individual boxes to show why it is showing the status it is. In this example, we have 6 staff rostered at 8pm which is perfect, but then based on the data at 8.15pm, we only need 5 staff.
When you click Build! you might see the following error message on the left in the Manage Demand Data box:
Your Business Hours have been set inside the hours that data has been imported. You might close the restaurant at 11pm but served customers until midnight, therefore in order to see those shifts you just need to set the Business Hours to match the Required Shifts. From there, once you can see all the shifts you can change the finish time, if you don't want to be rostering staff that late.
Comparing the data
You can use the graph to compare your data to your rostering.
Once your current day Sales data has imported, you can see whether the prediction was accurate or whether there were events that occurred that day that need to be taken into account for the same day next week and you could alter your ratio or settings.
You can select from a variety of sources in the panel for these comparisons e.g. Team vs Actual Sales, Expected Sales vs Total Staff or Expected Sales vs Actual Sales.
Once you've built one day, click on the next day in the week to start the process for the next day:
You can change your Ratios and Demand Data for each individual day, or keep them the same, depending on how your business operates.