Home · Solutions · Other solutions
Solution · Other solutionsManagers set goals where they already work, and HR watches completion live
Goals and ratings in Teams, not a hunt through the HR system
A manager's own team data, goals, check-ins and ratings history, arrives as a card in Microsoft Teams; goals are drafted, confirmed by the manager and written back to the HRIS.
Executive summary
Goals live in email, ratings history is buried in the module, and HR chases completion three times a cycle.
The performance module stays exactly where it is.
Managers spend cycle time on conversations, because the data behind them is one question away in a chat they already have open.
HRIS performance module, goal fields only; Power BI completion model
Business problem
Performance
Performance modules are built around the HR process, not around a manager's week. A people manager opens the module three times a year under a deadline, and between those visits everything they need, goals, progress notes, previous ratings, the comment somebody left at mid-year, sits behind a navigation path they relearn each time. So the data moves out of the system and into a notebook, a private spreadsheet, a chat thread.
Two groups pay for that. Managers lose hours per cycle to navigation, exports and assembly before any conversation happens. HR business partners lose weeks per cycle to completion reports pulled by hand and to reminder rounds addressed, each round, to roughly the same names.
Nothing here is a technology failure, which is why it survives. The module is the system of record and will not be replaced, the vendor's own app is not where managers work, and the cost is spread so thinly, a couple of hours per manager per cycle, that it never appears as a line anyone owns. Multiply thin by 310 managers and three cycles and it stops being thin.
How it works today
The pattern below turns up in organisations with a competent HRIS that still cannot close a cycle on time.
- SystemHR opens the goal-setting window in the HRIS and a bulk email goes to every people manager
- PersonThe manager asks the team to send goals by email, because that is faster than explaining where the fields are
- WaitingGoals arrive over two weeks in five different formats and sit in the inbox until a quiet afternoon
- PersonThe manager, or an assistant, retypes them into the module, late or not at all
- SystemAn HR business partner exports completion status to a spreadsheet and builds the chase list by unit
- PersonThree rounds of reminder emails go out, largely to the same managers each round
- Risk of errorAt year end, ratings are written against goals that were never recorded, and mid-year notes are reconstructed from memory
- WaitingCalibration stalls while business partners reconcile the gaps by hand
Why the current process costs more than it appears
Time that disappears before anyone measures it.
- Manager hours are the visible loss, and the expensive kind: people paid to run teams spend cycle time on navigation and exports rather than on the conversation the cycle exists for.
- Decisions rest on thinner evidence than anyone intends. A rating assigned without the mid-year note is a rating assigned from memory, and a promotion argued from recollection is harder to defend six months later.
- Goals written in eight personal styles cannot be compared inside a unit, so calibration spends its first half agreeing what the goals meant instead of deciding who delivered.
- Business partners who should be coaching difficult conversations spend cycle weeks chasing completion, the least skilled work they do all year.
- Where completion is invisible until the window shuts, HR cannot intervene while intervention is still cheap, and the same units run late every cycle for the same reason.
Cost of inaction
Each new people manager learns the module from the one who trained them, which is how avoidance gets inherited rather than corrected. Twelve months from now the three cycles will have run the way they always run: goals by email, three reminder rounds, a scramble before calibration, and a quarter of the business partners' year on a chase list nobody wanted to build.
What the table cannot price is defensibility. Ratings made from memory hold up right until the first serious challenge to a promotion or a pay decision, and the only question then is what the record shows. A record assembled from inboxes answers it badly.
A plausible organisation with realistic proportions. The figures are there to be recalculated on your data; they are not a client result.
An insurance company with 2,900 employees running performance in Workday alongside Microsoft 365; three cycles a year, goal setting in the first quarter, a mid-year check-in and a year-end review.
310 people managers supported by twelve HR business partners, each manager preparing three times a year for a team of six to ten.
Goals arrive by email and are retyped into the module; completion is tracked in an exported spreadsheet; three reminder rounds per cycle; mid-year notes live in personal files.
About 150 minutes per manager per cycle on navigation, exports and preparation, plus roughly three working days per business partner per cycle on reports and chasing.
A conversational agent in Microsoft Teams reads the manager's own team data from the HRIS, shows a status card with one-tap nudges, drafts goal wording from the manager's notes, and a UiPath Robot writes only the goals the manager has confirmed; a Power BI page shows completion by unit, refreshed daily.
In the modelled case a little over half the manager preparation time returns to managing, the reminder rounds disappear, and completion becomes visible while the cycle is still open. Every figure here is modelled; none of it is a client measurement.
Proposed solution
The performance module stays exactly where it is. What moves is the place a manager touches it. We build a conversational agent that lives in Microsoft Teams, either as a UiPath conversational agent deployed as a Teams app or, where the client standardises on the Microsoft stack, as a Microsoft Copilot Studio agent. It reads the HRIS through UiPath Integration Service and is scoped at every request to the hierarchy the asking manager owns, resolved from the HRIS and Microsoft Entra ID.
The entry point is a card, not a menu. A team card arrives in Teams with one row per report: goal status, check-in status, the last two ratings, and a button that nudges the employee with a direct link. Plain-language questions work in the same chat: who has not entered goals, what was agreed with a named report at mid-year, how this team stands against the unit.
Writing is where we are deliberately conservative. When a manager types rough notes, the agent proposes wording in the organisation's own goal format; the manager edits it on screen and confirms; only then does a UiPath Robot write the goal to the HRIS, through a service account whose rights end at goal fields. Ratings and comments are displayed, never generated or altered. Each read, draft and write carries a timestamp and an actor.
Business partners get the same data as a Power BI page refreshed daily: completion by unit, by manager, by cycle stage. That page replaces the export and the three reminder rounds, and turns chasing into a targeted conversation with the four or five managers who genuinely need one.
Microsoft Teams Adaptive Cards and agent hosting; UiPath conversational agents deployed as a Teams app; UiPath Orchestrator queues, audit trail and credential store; UiPath AI Trust Layer model allow-list, PII masking and conversation audit; Power BI semantic model and workspace sharing
The agent's tools and guardrails, the team status card and its nudge action, the goal-format prompt and the confirmation step, the write robot and its field-level scope, the completion model and the business partner view
The Workday performance module through the UiPath Integration Service Workday connector, or SAP SuccessFactors over OData; Microsoft Entra ID for hierarchy and group scoping
How the automated process works
- SystemHR opens the cycle in the HRIS; the agent picks up the window dates and the current goal format from configuration
- AutomationEvery people manager receives a team status card in Microsoft Teams with goal and check-in status per report
- PersonThe manager asks the agent a question, opens a report's history, or types rough goal notes into the chat
- AutomationThe agent drafts goal wording in the organisation's format and puts it on screen for editing
- PersonThe manager edits and confirms each goal; nothing reaches the record before that confirmation
- SystemA UiPath Robot writes the confirmed goal to the HRIS through a scoped service account and logs the write
- AutomationMissing goals produce a one-tap nudge to the employee and a daily refresh of the completion model
- PersonThe business partner opens the Power BI page and talks to the units that are behind, while the window is open
Human-in-the-loop model
Automation handles
- Team status cards, plain-language answers and ratings history, pulled from the HRIS at the moment of asking
- Goal wording drafted from the manager's own notes in the organisation's format
- Writing confirmed goals to the HRIS and logging every write with its actor and timestamp
- Nudges to employees and the daily completion view by unit and manager
People decide
- The performance conversation itself, and whether a goal is the right goal
- Confirmation of every goal before it is written; an unconfirmed draft never reaches the record
- Ratings, calibration outcomes and pay decisions, entered by managers in the HRIS
- Cycle design, the goal format, and which late units get a conversation rather than a reminder
Before and after
Systems and integrations
Everything below runs on licences and systems you already hold, or would need anyway.
Inputs
- HRIS performance module (goals, check-ins, ratings history)
- Microsoft Entra ID manager hierarchy
- the organisation's goal format and cycle calendar
Automation layer
- UiPath Agents
- UiPath Orchestrator
- UiPath Robots
- UiPath Integration Service
- UiPath AI Trust Layer
Target systems
- HRIS performance module, goal fields only
- Power BI completion model
Human touchpoints: the team status card in Microsoft Teams; the agent chat; the goal confirmation step; the business partner completion page
Technologies used
answers a manager's questions over their own team's performance data and drafts goal wording from notes
Areads goals, check-ins and ratings history from the performance module through its API
Awrite confirmed goals back to the HRIS, run nudges and the completion extract, hold the queue and the audit trail
Athe team status card, the nudge action and the manager's whole interaction surface
Aresolves the manager hierarchy so every answer is scoped to the team the asker actually owns
Amodel allow-list, PII masking and conversation audit over the drafting step
Adaily completion view by unit, manager and cycle stage for HR business partners
Aalternative host for the same agent where the client standardises on Microsoft
AIllustrative economic model
A model, not a promise.
Managers are the expensive input here, at €78 an hour fully loaded, so the arithmetic prices their time and nothing else. 310 people managers across three cycles a year works out at about 77 preparations a month, and only 0.6 of each preparation is the navigating, exporting and assembling that the assistant removes, so that share is folded into the volume and the calculator counts 47 preparations a month rather than 77. Business partner time sits outside this table and appears in the next one. Nothing below was measured at a client.
Run the numbers on your data
An illustrative estimate from your own inputs. It models released capacity; it is not a promise of savings.
Business benefits
- Managers spend cycle time on conversations, because the data behind them is one question away in a chat they already have open
- Goals are entered inside the window and in one format, which makes calibration a comparison rather than a translation exercise
- Mid-year notes and ratings history sit in front of the manager when the year-end rating is written, so the decision rests on the record rather than on recall
- Business partners stop assembling completion reports and start coaching the small number of managers who are genuinely behind
- The organisation can see, while a cycle is running, which units are managing performance and which are filling in fields
- Every goal write, nudge and data access is timestamped, so a question about when a goal was set has an answer independent of anyone's memory
The management view
- Completion by business unit is visible daily while the window is open, not in a report assembled after it shuts
- Business directors see their own managers' status without raising a request with HR
- Goal quality becomes comparable across units because wording follows one format instead of eight personal styles
- The performance record stops depending on which managers copy their notes into the system
Board-level KPIs
Security and governance
Trust in automation is built on the audit trail, not on a promise.
- Scope is resolved per request, not per session: the agent answers only about the hierarchy the HRIS and Microsoft Entra ID agree the asker manages, and each access is logged
- The write path is deliberately narrow: one robot, one service account, goal fields only, no route to ratings, salary or comments; its secret is held in the Orchestrator credential store or Azure Key Vault
- Ratings and mid-year comments are displayed to the manager, never generated, edited or rewritten by the agent
- Drafting runs under the UiPath AI Trust Layer with a model allow-list, PII masking and conversation audit in the EU region of UiPath Automation Cloud; on a Microsoft Copilot Studio build, Microsoft Purview audit and Entra Agent ID cover the same ground
- Works-council consultation on manager-facing AI is a discovery deliverable with a named owner, not a box ticked later
Why now
Pay-equity and transparency expectations have turned documented, comparable goals into a governance question rather than an HR preference, and a record assembled from inboxes does not answer it
Performance modules now expose goals, check-ins and ratings through APIs, so the system of record can stay precisely where it is while the experience moves to where managers already work
Scoped conversational agents over enterprise data are mature enough for an HR context under an explicit trust layer, and the modelled €9,200 a month of manager time is spent every month the decision waits
Relevant executive roles
Cycles close inside the window with goals that compare across units, and completion is visible while there is time to act on it
Business partners move from assembling chase lists to coaching the managers who find the conversation hard
Their managers' status is visible without a request to HR, and most of a manager's preparation time each cycle goes back to managing
Common questions and objections
They should, and they do not, which is the observable fact this starts from. The agent reads and writes the same record; it changes where a manager stands when they do it, not what the system of record is.
The model drafts wording from the manager's own notes, in your format, and the draft is inert until the manager edits and confirms it on screen. Ratings are never generated, and the write path reaches goal fields and nothing else.
A manager sees exactly what the HR system already shows them about their own team, and each access is logged. That is more transparency than today's situation, where the same data is copied into private spreadsheets nobody audits.
When this is not the right solution
- Managers already complete cycles inside the module and on time, in which case the problem sits somewhere other than access
- The performance module is being replaced or the cycle redesigned within the year; build once the new model is fixed
- The manager population is small enough for HR to support each person individually, where a conversation beats an agent
- Works-council agreement on manager-facing AI cannot be obtained, and drafting is the part that would have to be dropped
A question for the next management meeting
Pick any rating this company awarded last cycle: how long would it take us to produce the goal it was set against and the mid-year note behind it?
Implementation approach
The first week looks the same at every client: we look at the data.
We deliver
- Analysis of the performance data model, the cycle calendar, the goal format and last cycle's completion
- Agent design: tools, guardrails and the rule that ties every answer to the requester's own hierarchy
- The goal-format prompt and the confirmation step, tuned on real goal text from your own units
- HRIS integration and the write robot, with a service account limited to goal fields
- The team status card, the nudge action and the daily completion model in Power BI
- A pilot with one business unit for one cycle, read-only first, then drafting and writing
- Monitoring of answer accuracy and write errors, with a runbook for the HR team
We need from you
- API access to the performance module and a test tenant with representative data
- An HR process owner who owns the goal format and the cycle design
- One business unit's managers for the pilot, plus last cycle's completion report
- Whatever works-council consultation your agreements require for manager-facing AI
Stages
Discovery
Data model, cycle calendar, goal format and last cycle's completion, with HR and two directors
Design
Agent scope, guardrails, the team card, the confirmation step and the security model
Build
Agent, HRIS integration, write robot and the completion view in your own environment
Pilot
One unit, one cycle: read-only first, then drafting and writing, measured on completion and accuracy
Rollout
Remaining units and later cycles, with close support while each window is open
Optimisation
Prompt and guardrail tuning, new questions, extensions to the reporting
Quick win. Effort is driven by how clean the performance data model is, how many goal formats the organisation genuinely uses, and whether works-council consultation must precede any AI-assisted draft.
The goal fields are empty in February and they are still empty in November.
Send us last cycle's completion report by unit and the dates the three reminder rounds went out. We come back with the Teams card that would have replaced them and a one-page view of where the cycle actually stalls.
Replay last cycle's completion with usThe neighbouring process usually has the same problem
Thousands of survey comments unread, recognition lost in a thread, communities nobody can find.
View solution Other solutionsClose rates, targets and the forecast on one trusted pageForecast_Q3_v7_FINAL_CFOcomments.xlsx has nine tabs and three definitions of committed.
View solution Other solutionsThe same right answer to every HR policy questionTwo advisors answer the same handbook questions all year, from different versions of the handbook.
View solutionIndustries we deliver this in most oftenManufacturing & industryRetail & e‑commercePublic sectorServices & ITShared services