Home · Solutions · Supply chain

Solution · Supply chain

Every open purchase order gets a confirmed date in SAP, and planners hear about slips first

Order confirmations and delivery dates chased automatically

Robots chase suppliers for confirmations, read the replies, write the confirmed dates into SAP and warn planners in Teams; buyers only handle the disputes.

DepartmentalMicrosoft TeamsHuman in the loopAI where it earns its place
2,800open purchase-order lines a month need a follow-up at this illustrative supplier. Buyers chase them by hand, and SAP learns the real date days later.

Executive summary

Challenge

Buyers spend their mornings chasing suppliers for dates, and planners plan on dates that are already wrong.

What changes

Each morning an unattended UiPath robot selects from SAP S/4HANA every open purchase-order line that meets a chase rule.

Business value

Buyers spend their mornings on declines, splits and suppliers that need a conversation, not on routine chasers.

Systems involved

SAP S/4HANA purchase-order confirmations (ME22N / BAPI); SharePoint confirmation log; Power BI report

Business problem

Supply assurance

A purchase order is a request, not a promise. Until the supplier confirms quantity and date, the ERP plans on the date the buyer wished for. Some open lines are always unconfirmed, late or quietly moved, and the only way to know is to ask: run an open-item list, write emails, wait, read the answers and, if time allows, type the new date into SAP.

Buyers lose the first two hours of the day to this. Planners distrust the dates in SAP, keep their own shortage lists in Excel and call the buyers, who call the suppliers. The plant director sees premium freight, overtime and the occasional stopped line; the customer sees a missed delivery on the scorecard.

At scale the lines worth chasing outgrow the team, so buyers triage by instinct and small suppliers are assumed fine. Confirmations reach SAP inconsistently, MRP replans on a mixture of real and fictional dates, and nobody can say which suppliers slip after confirming. That knowledge sits in nine heads and nine mailboxes.

How it works today

  1. PersonOn Monday a buyer runs the open purchase-order list in SAP (ME2N), exports it to Excel and colours lines with no confirmation, a date later than requested or a delivery date already past
  2. PersonFor each coloured line the buyer writes a chaser from a personal mailbox, or calls if the part is critical
  3. WaitingReplies trickle in over two to five working days, mixed with quotations and invoices
  4. PersonThe buyer reads each reply, decides what "first half on the 14th, rest the week after" means and updates ME22N, or just notes it in the Excel
  5. AutomationThe fifteen largest suppliers send EDI order responses that SAP books unread; the other four hundred do not
  6. Risk of errorDates never typed into SAP stay as requested dates, so MRP, the shortage list and the customer promise rest on numbers a supplier has already contradicted
  7. PersonOn Friday the planners bring their own shortage list to the meeting and the buyers reconcile it with their mailboxes from memory
PersonWaitingAutomationRisk of error

Why the current process costs more than it appears

The budget shows headcount, not what it is spent on.

  • Chasing is only the visible cost. A reply misread or filed unread becomes an expediting call, a premium freight booking or a replanning round a week later.
  • Wrong dates in SAP are paid for downstream: MRP schedules production on material that will not arrive, planners build a parallel truth in Excel, sales promises dates the plant cannot keep.
  • Instinct decides who gets chased, so small suppliers are chased last and surprise most; the one that slips on every third order is never identified because nobody keeps count.
  • Supplier reliability lives in buyers' memory and leaves with them, and a customer scorecard does not distinguish a supplier's failure from an email nobody read.

Cost of inaction

Twelve months of chasing by hand≈ €134,400
Three years of the same Monday export≈ €403,200
If lines needing follow-up grow to 3,600 a month, each year≈ €172,800

Expediting stays "normal", premium freight remains a cost line nobody questions, and the planners' Excel survives: the routine does not get worse in a way anyone notices, only more expensive in places booked elsewhere. The next supplier or plant means another buyer.

The larger exposure is the one your customers measure: a late delivery announced two weeks earlier in an unread email counts on the scorecard exactly like an unforeseeable one. And the knowledge of which suppliers slip stays in the buyers' heads, and leaves with them.

Illustrative scenario

A plausible organisation with realistic proportions. The figures are there to be recalculated on your data; they are not a client result.

Organisation

A tier-2 automotive supplier, three plants in Poland and Slovakia, 1,400 employees, SAP S/4HANA, Microsoft 365 E3; nine buyers in central purchasing, twelve production planners.

Volume

About 6,000 purchase-order lines a month, of which 2,800 need a follow-up: unconfirmed after two working days, confirmed later than requested, or past due without a goods receipt. The fifteen largest suppliers confirm by EDI; four hundred others reply by email in Polish, German and English, from a one-word "OK" to a paragraph about a holiday shutdown.

Current process

Monday exports from ME2N, chasers from personal mailboxes, replies interpreted by the buyer, confirmations entered in ME22N when time allows; planners keep a separate shortage list and the Friday meeting reconciles the two.

Bottleneck

Around eight minutes of buyer time per chased line, about 373 hours a month; a confirmed date reaches SAP three to seven days after the supplier sent it, if at all.

Solution

Robots chase each morning from the shared purchasing mailbox, one consolidated message per supplier, and escalate on a fixed ladder; table replies are parsed by rules, free text by UiPath Communications Mining; confirmed dates reach SAP the same day; late dates alert the planners in Teams; buyers see only uncertain replies, declines and disputes; Power BI ranks suppliers weekly.

Potential outcome

In the modelled case, routine chasing falls to the exception share (20 to 25% of chased lines in the first months), confirmed dates reach SAP the day the supplier replies, and planners learn of a slip within the hour rather than on Friday. Illustrative, not a client result.

Proposed solution

Each morning an unattended UiPath robot selects from SAP S/4HANA every open purchase-order line that meets a chase rule: unconfirmed two working days after release, confirmed later than requested, or past due without a goods receipt. The rules are yours, per supplier and material group, in a SharePoint list purchasing edits. One chaser per supplier per day leaves the shared purchasing mailbox through the Microsoft Outlook 365 connector, in the supplier's language and signed by the buyer, with a table of PO line, material, quantity, requested date and an empty "confirmed date" column.

A filled-in table is handled by rules: the robot checks the PO line and quantity and writes the confirmation. A reply in prose ("half on the 14th, the rest after the holiday, and the price has gone up") is read by UiPath Communications Mining, a machine-learning capability of UiPath IXP trained on your own past replies: it labels the intent (confirmed, new date, partial, declined, question) and extracts dates and quantities with a confidence score. Above your threshold the robot writes the confirmation to the purchase-order item in SAP as an order acknowledgement and archives the reply; below it, or for any decline or price question, the reply becomes a buyer task in UiPath Action Center, completed in Microsoft Teams with the proposal pre-filled.

A confirmed date later than the requirement date triggers an alert in the planners' Teams channel with the PO, material, both dates, coverage in days and the buyer's name. Every chaser, reply, score and SAP change is logged to a SharePoint list behind a Power BI report, published weekly by subscription and as a tab in the purchasing channel. Generative AI stays out: the replies are short and repetitive, and a supervised model with a threshold is easier to govern than a prompt.

Native capabilities used

UiPath Communications Mining (IXP) labels and field extraction with confidence thresholds; UiPath Orchestrator queues, time triggers and audit; UiPath Integration Service connectors for Microsoft Outlook 365, Microsoft Teams and Microsoft OneDrive & SharePoint; UiPath Action Center actionable notifications in Microsoft Teams; Power BI subscriptions and Teams tabs

What we build

Chase rules and escalation ladder, chaser templates, the reply-table parser, the Communications Mining taxonomy trained on your replies, the SAP confirmation update and its safeguards, shortage alerts, the SharePoint log, the Power BI report and the runbook

Custom integration

SAP S/4HANA open-line selection and confirmation updates through UiPath SAP activities (BAPI/OData), with SAP GUI activities for ME22N where no released interface exists

How the automated process works

  1. AutomationA time trigger in Orchestrator starts the robot each morning; it pulls open purchase-order lines from SAP and applies the chase rules
  2. AutomationOne consolidated chaser per supplier leaves the shared mailbox; second and third chasers follow the escalation ladder, the third with the buyer in copy
  3. AutomationReplies are collected from the shared mailbox; table replies are parsed by rules, free text goes to Communications Mining for intent and date extraction
  4. SystemAbove the confidence threshold the robot writes the confirmed date and quantity to the purchase-order item in SAP and archives the reply with its score
  5. PersonUncertain replies, declines, partial confirmations and any mention of price reach the buyer as an Action Center task in Teams, with the interpretation pre-filled
  6. AutomationA confirmed date later than the requirement date triggers an alert in the planners' Teams channel with the material, both dates and coverage in days
  7. AutomationSuppliers that ignore three chasers are listed to the buyer for a call; every chaser and reply is logged to the SharePoint list behind the Power BI report
AutomationSystemPerson

Human-in-the-loop model

Automation handles

  • Which lines to chase, when and in which language, by rules purchasing owns
  • Sending and escalating chasers, reading table and free-text replies, scoring its confidence
  • Writing confirmed dates and quantities to SAP, with the reply archived as evidence
  • Shortage alerts to planners and the weekly supplier reliability report

People decide

  • Replies the model is unsure about, and every decline, partial or price-related answer
  • Whether to accept a later date, split the order or source elsewhere, given the coverage
  • Which suppliers move to a call or a formal escalation after ignoring three chasers
  • The chase rules, thresholds and templates, owned by purchasing rather than IT

Before and after

BeforeAfter
Buyer time per chased line~8 minseconds for the routine share; minutes for disputes
Confirmed date visible in SAP3–7 days after the reply, if entered at allsame day, with the reply archived
Planner learns of a slipFriday shortage meetingTeams alert within the hour
Share of lines chased by a person100%modelled 20–25% (uncertain, declined, disputed)
Supplier reliability databuyers' memory and Excelweekly Power BI report per supplier

Systems and integrations

Every entry can be checked in vendor documentation. The evidence class is stated next to each one.

Inputs

  • SAP S/4HANA open purchase-order lines
  • shared purchasing mailbox in Outlook
  • supplier replies (table and free text)
  • EDI order responses already booked in SAP

Automation layer

  • UiPath Orchestrator
  • UiPath Robots
  • UiPath Communications Mining
  • UiPath Integration Service
  • UiPath Action Center

Target systems

  • SAP S/4HANA purchase-order confirmations (ME22N / BAPI)
  • SharePoint confirmation log
  • Power BI report

Human touchpoints: planners' Teams channel; Action Center tasks in Teams for buyers; weekly Power BI report in the purchasing channel

SAP S/4HANA open purchase-order linesUiPath OrchestratorUiPath RobotsSAP S/4HANA purchase-order confirmationsplanners' Teams channel

Technologies used

UiPath Robots + Orchestrator

select open lines, send and escalate chasers, write confirmations, log every step

A
UiPath Communications Mining (IXP)

reads free-text replies: intent, dates and quantities, with a confidence score

A
UiPath Integration Service (Microsoft Outlook 365 connector)

sends chasers from the shared mailbox and collects replies

A
UiPath Integration Service (Microsoft Teams connector)

posts shortage alerts to the planners' channel

A
UiPath Action Center in Microsoft Teams

buyer tasks for uncertain interpretations, declines and price questions

A
SAP S/4HANA (BAPI/OData via UiPath SAP activities)

open-line selection and confirmation updates on the purchase-order item

A
SharePoint list + Power BI

confirmation log and the weekly supplier reliability report

A
Averified product capability (vendor documentation)

Illustrative economic model

The arithmetic is open, so it can be argued with.

Illustrative model
2,800 PO lines a month × 8 minutes of chasing and updating= 373 h / month
373 h × €30 fully loaded hourly cost≈ €11,200 / month
× 12 months≈ €134,400 / year
Buyer capacity released per year (illustrative)≈ €134,400

Phone calls are excluded, and so is everything downstream, which is one reason to read the table as an illustration rather than a client measurement. 2,800 is the number of purchase-order lines a month needing at least one follow-up, not the total released. Eight minutes per line covers selecting it, sending the chaser, reading the reply, updating SAP and a share of repeat chasers. €30 is a fully loaded hourly cost for a buyer in Central Europe. The model counts buyer capacity only; premium freight, expediting and stoppages avoided are left out, although usually larger.

Run the numbers on your data

hours released per month
of annual capacity released

An illustrative estimate from your own inputs. It models released capacity; it is not a promise of savings.

Business benefits

  • Buyers spend their mornings on declines, splits and suppliers that need a conversation, not on routine chasers
  • Confirmed dates reach SAP the day the supplier replies, so MRP, the schedule and the customer promise rest on what the supplier actually said
  • Planners learn of a slip within the hour, with coverage in days, and can replan or source elsewhere before the line is affected
  • Every open line is chased by the same rule, so small suppliers no longer surprise the plant because nobody had time to write to them
  • Supplier behaviour becomes a number: who confirms within 48 hours, who slips after confirming, who needs three chasers

The management view

  • The share of open purchase-order value with a supplier-confirmed date becomes a figure the COO reads each morning, not a feeling
  • Chasing discipline no longer depends on who is on holiday; the rules run daily and the ladder is the same for every supplier
  • Supplier reviews and dual-sourcing decisions rest on a log rather than on memory
  • Suppliers, materials and volume can grow without adding buyers for follow-up work

Board-level KPIs

share of open PO lines with a confirmed date in SAPconfirmation within 48 hours by supplierdate slips after confirmationpremium freight cost per monthbuyer hours per open line

Security and governance

The automation holds exactly the rights it needs, and not one more.

  • Reading open lines and writing confirmations on purchase-order items is the whole of what the robot's SAP account may do: no PO creation, no price or quantity changes
  • Mailbox access covers the shared purchasing mailbox only, through a dedicated connection; personal mailboxes are never read
  • SAP and mailbox secrets sit in Orchestrator's credential store backed by Azure Key Vault; the supplier replies, the trained model and the log those credentials open stay in your Microsoft 365 tenant, with UiPath Automation Cloud running from its EU region
  • Every confirmation written by a robot carries the source reply, label, confidence score and timestamp: an auditor can see why a date changed
  • Retraining, new labels and threshold changes are reviewed by purchasing and versioned; personal data in replies is limited to supplier contact names under your retention policy

Why now

01

OEM customers score tier-2 suppliers on delivery performance and charge back line stoppages and premium freight; the wrong date in SAP is where most surprises start, while the routine meant to prevent them consumes the modelled €11,200 of buyer capacity a month

02

Experienced buyers are hard to hire and harder to keep; a team that spends its mornings on chasers loses the people it most wants to retain

03

Reading free-text replies no longer needs a custom NLP project: Communications Mining is a standard capability of UiPath IXP, and the Outlook 365, Teams and SAP connectors are off the shelf

Relevant executive roles

COO

Production plans on dates a supplier has confirmed, and slips reach the planners before they reach the line

Procurement Director

Buyers work on suppliers and disputes instead of chasers, and supplier reliability becomes measurable

Plant Director

Premium freight and expediting are decided with days of notice, and the Friday shortage meeting gets shorter

CIO

A governed, auditable SAP integration through standard interfaces replaces personal mailbox rules and Excel trackers

Common questions and objections

Suppliers ignore our emails now; why would they answer a robot?

The chasers are not anonymous: they come from the purchasing mailbox, signed by the buyer, one consolidated message per supplier per day instead of a dozen. The ladder applies to every supplier every time, and the reliability report gives buyers a fact for the next supplier review.

Do we really need machine learning to read a reply?

Not for all of them. Suppliers who fill in the reply table are handled by plain rules, and that share grows once they learn the table gets them left alone. The model is for prose replies, where the alternative is a person reading each one; anything uncertain goes to a buyer.

We are moving suppliers to a portal next year.

The chase logic does not change. Portal confirmations become one more structured channel for the rules, and the robot keeps chasing the long tail that never adopts the portal, where most late surprises come from.

When this is not the right solution

  • A disciplined weekly routine and a good chaser template are cheaper when fewer than a few hundred lines a month need a follow-up
  • Confirmation control is not maintained in SAP, so MRP ignores confirmed dates anyway; that master-data decision comes first
  • Most suppliers already confirm by EDI or portal, and the remaining tail is small enough to chase by hand

A question for the next management meeting

Of the delivery dates our planners are working from this week, how many were confirmed by a supplier, and how many are simply the dates we asked for?

Implementation approach

Delivery runs in stages, so it can be stopped at any point.

We deliver

  • One month of open lines and one month of the chasing mailbox, sorted into chase categories, reply styles, languages and exception types
  • Chase rules, escalation ladder and chaser templates per supplier group and language, in a list purchasing can edit
  • The Communications Mining taxonomy, trained and validated on your replies, with agreed thresholds
  • SAP selection and confirmation update with safeguards: no quantity, price or new-line changes by the robot
  • Shortage alerts agreed with the planners, Action Center tasks in Teams, the SharePoint log and the Power BI report

We need from you

  • Six months of supplier correspondence from the purchasing mailboxes and the matching purchase-order history from SAP
  • A purchasing process owner and one planner per plant for rules, thresholds and alerts
  • A service account for SAP limited to purchase-order confirmations, and the shared mailbox
  • Confirmation control settings in SAP that let MRP use confirmed dates, or the decision to switch them on

Stages

Discovery

Open-line profile, reply styles, exception types and current confirmation discipline

Design

Chase rules, escalation ladder, label taxonomy, thresholds, alert criteria, security model

Build

Selection and chasing robots, reply parsing, SAP update, Teams touchpoints, log and report

Model training

Communications Mining trained on your replies until the agreed precision per label

Pilot

One plant or supplier group, buyers in parallel, exception handling and acceptance

Rollout

Remaining plants and supplier groups, hypercare, tuning of rules and thresholds

Departmental. Effort depends on the number of plants and purchasing organisations, the variety of reply styles and languages, and how consistently confirmation control is maintained today.