CDQ & Nestlé Touchless Project
Summary of the questions and answers from the session
User Adoption and Change Management
- What was the biggest hurdle to overcome when aligning with the users and stakeholders?
The most challenging part was to achieve consistency in applying data standards across markets. Different functions (supply chain, finance, etc.) had to be aligned, which took significant amount of time.
- How is the reaction of the users not being able to edit the data?
At first, approvals were deactivated for a month so that the users could validate outcomes. This eased the transition, and the approach is now used as best practice when rolling out to new markets.
- Which pushbacks have you experienced from the customers and how did you handle them?
The main concern was the fear of change: customers worried about fewer fields and automation. We overcame this through open communication and by showcasing the benefits.
- Do you (and how) differentiate between leads and customers or do you put each customer request into the ERP?
It varies by market, BR itself does not differentiate.
- Is this both for the customer to create a new customer and to make changes to existing customers?
The process only applies to creation at the moment, however we plan to include change requests at some point in the future.
Process and Automation
- Are your customers legal entities or individuals?
Our customers indeed cover only customer and vendor data, not individual data.
- Is there a manual vetting in the process?
No, there is no manual vetting in the process if the data meets proper criteria.
- Could you please describe in more detail the difference between touchless creation and manual creation?
Manual creation can involve multiple workflow steps, while touchless creation executes all checks programmatically. If rules are met, the record is created automatically. Only failed checks go to manual review.
- We have a complex approval process in place before a customer can be created. Is that on your end not needed?
In Brazil, the process is simpler: anyone can register via the web portal. No external approval process is required.
- What will happen in case a customer becomes insolvent? Which process is used to block the customer?
Normal blocking and deletion process is currenly in place, although we are testing a report to get alerts when a Tax ID has become inactive to trigger the deletion in our systems.
- Are you planning to set up this kind of process for suppliers too?
Yes, this process is also planned for suppliers.
- Is there a workflow for when there's a need to approve something? And if yes is there any report to check where request is at that moment?
Only the last step would come into play and only if one of our checks fails (completeness, duplicate or data standard). That's the case for roughly 15% of the volume that needs to be reviewed by someone, and 85% is autoapproved. We do have a PowerBI report but the LDOs also can see their request in their MDG inbox.
- Do you consider the cessation of activities of partner?
We haven’t yet, but it's definitely on our radar. When a Tax ID has been marked as inactive we are working on triggering a blocking CR.
Data Quality and Compliance
- Do you have any checks if the customer is who he says he is?
Yes, we ensure authenticity of customer information, depending on local data sources available. For exapmle, in Brazil data is consumed directly from government databases, which are our trusted source of truth.
- Are you checking the data? I mean, everybody can enter anything, so how do you know that the company is legal and the information provided is true?
Yes, we do check against trusted sources. For example, in Brazil we rely on official government data sources, which we trust as accurate.
- Was the duplicate check done in Power Apps or in MDG? (other than Tax ID)?
Duplicate checks are currently handled in MDG. The plan is to implement validations in Power Apps in the future.
- Would automated enrichment and verification of customer data also be possible in Europe, given very strict regulations on data protection?
Yes, it is possible. CDQ, based in Switzerland/Europe, already applies automated enrichment and verification, respecting local data protection regulations.
- Who do you believe is the actual custodian of the master data?
The data is owned by the market, the data stewards own the standards and IT owns the tools to create that data.
Technology and Integration
- How do you handle the Company code and Sales org data in SAP? Are they derived automatically?
Yes, they are derived automatically via predefined rules. In Brazil, the rules needed to be mature and stable before rollout, since they had often changed earlier.
- What is the IBM part and the SAP part in the entity resolution process? Are you using AI models?
IBM focused on developments outside of SAP, mainly Power Apps. We are using AI only for OCR to indicate what we want to extract from a document.
- I see that you have a PowerApp in place, looks like the entry channel. Is this self-developed? Is this the only entry channel for customer requests?
In Brazil, both a web portal and Power Apps are used (the latter internally). In other markets, requests still go through MDG, but the goal is to unify entry channels.
- I noticed that Power Apps was used for data input. Is there a reason why you didn’t go through MDG directly instead?
MDG was considered too complex by users. Power Apps offers a simpler UI and allows independent development outside of SAP, while maintaining full integration capabilities. PowerApps and MDG are connected via API.
- Which MDG-deployment was integrated in this process? (On-prem based on S/4, private cloud, public cloud, ... ?)
Cloud deployment was used.
- What you see as the role of MDG if you move the rules & validations to Power BI?
MDG remains the central governance tool. While some validations may move to Power Apps, MDG ensures audit, approval tracking, and data governance, especially in markets where APIs are limited.
- Where / How OCR and AI are used in the process?
In France (launching October), customer/vendor creation requires a bank document (PDF). We're using OCR to extract data such as IBAN, bank name, account holder: we make use of GenAI with Argos tool to process this data using prompts.
Financial and Operational Aspects
- Comparing the old and the new process how many percent of the process costs for creating one customer do you save with this process?
While the exact numbers in terms of cost savings were not calculated, the process drastically reduces the time people spend on the record - from 40 mins on average to 5 mins - which in itself translates to money savings.
- Considering the project costs and the process costs for the new process, can you say from what amount of customers this approach is worthwhile? Where is break-even?
Break-even was achieved, though no exact customer volume was mentioned.
- How are the finance related info updated? Like credit limit, risk category, payment term, and what´s with IBAN?
Finance-related info (credit limit, risk category, terms, IBAN) is updated automatically but outside the master data process. Rules vary by market, and credit risk is maintained after record creation.
- Can you also use it for credit management?
No, credit management is outside the master data process.
