'Time is life' - how Pfizer uses Celonis to deliver medicines faster with process-first automation

When Conor Riordan talks about order automation at Pfizer, it’s not just about efficiency metrics. It’s realistically about life and death. As the pharmaceutical giant's Director of Customer Service put it during a presentation at Celonis' Celosphere conference in Munich this week: 

Time is life in our business.

This thinking shapes everything about Pfizer's approach to process improvement and automation. The company, which employs 81,000 people globally and manufactures 1.4 billion packs of medicine annually, has spent six years building a process intelligence capability with Celonis that now spans every market where it does business. The goal isn't just operational efficiency - it's getting medicines to hospitals, pharmacies, and physicians as quickly as possible.

Currently, 75% of Pfizer's deliveries arrive same day or next day. The company's target is 95% within two years. For hospitals that order medicines three or four times daily and don't carry inventory due to space and cash constraints, those hours matter. Riordan explains:

The more touches and the more manual entry, the slower the process, the slower it's going to take product to get to hospitals, to get to pharmacies, to get to our physicians.

This philosophy underpins Pfizer's CX for Change Program, which aims to change 1 billion lives by 2027 through better customer experience and faster delivery. But what makes Pfizer's approach relevant for other organizations is its disciplined focus on process standardization before automation - a methodology that has driven the company from 58% order automation two years ago to 85% today, with a target of 95% by the end of next year.

The process-first mandate

Pfizer's approach is built on two principles: run standard processes globally, and only automate standardized processes. This sounds obvious, but as Riordan notes, the discipline required to maintain this approach is considerable. He says:

We lead with process, not with automation. Unless we have standard processes, unless we understand what we're trying to do, unless we have visual management boards, we will not automate the process, because all we're doing at that point is adding a lot of costs to try and automate a broken process very quickly.

The company has deployed what it calls IMEX (Integrated Manufacturing Excellence) methodology into commercial operations, creating process-centric teams that map standard processes using swim lanes, work instructions, and integrated visual management boards. Metrics drive continuous improvement programs, and gemba walks ensure markets adhere to these processes.

Pfizer also audits key data elements that underpin these processes on a periodic basis - service level agreements, batch sizes, forecast accuracy - ensuring systems run on curated data the company trusts. Only after this foundation is in place does the company consider AI and automation. Riordan was honest about the consequences of getting this sequence wrong, citing his involvement in a blockchain program several years ago:

We led with use case, and we spent an awful lot of money - I mean, a lot of money - and we finished up about 12 to 18 months later with no solution, because we started with the use case trying to find the problem. That's not the way to do it.

This aligns with the broader message Celonis has been delivering this week at Celosphere - that successful AI deployment requires understanding operational processes first, before applying technology solutions. As Celonis argued in its keynote, process intelligence provides the missing operational context that AI needs to actually work in the enterprise.

A $1.2 billion weekly operation

The scale of Pfizer's operation is challenging. The company runs a single instance SAP system that processes approximately $1.2 billion in revenue per week, having migrated to S/4 HANA last summer.

As a traditional B2B company, Pfizer deals with five different distribution channels: wholesale customers, retail markets including hospitals and pharmacies, physicians, and government contracts. The wholesale business runs predominantly on EDI with strong automation rates. But once Pfizer moves into retail, hospitals, pharmacies, and physicians, processes become more manual. Riordan notes that public hospitals in Europe still frequently order via fax, with Pfizer receiving thousands of fax orders annually that need to be processed and entered into SAP.

To address this, Pfizer has assembled a technology stack that includes Conexiom for sales order automation (which can read orders in any format, including multiple languages), SAP Commerce Cloud (branded as Pfizer Prime) for online ordering, HighRadius for cash application, SAP Service Cloud for ticket management, OpenText for document management, and Qualtrics for customer feedback.

But it's Celonis that sits at the center of Pfizer's ability to understand, measure, and improve these processes across its operations.

Measuring what matters

Pfizer has built a tiered metrics system (T1 through T5) that runs from individual business processes up to the Senior Vice President of Global Supply Chain's monthly scorecard. Within this system, Pfizer's customer service integrated balanced scorecard lives within Celonis, tracking two metrics: on-time in full delivery and automation rate.

The scorecard uses only two colors: green and red. There is no amber. As Riordan explains: 

Red is good - red means we have to do something. It means there's an action and we need a call to action. We don't have ambers. That's the rule. It's either on track or not on track.

The Celonis dashboards reveal the automation rate across the organization - currently showing 17.7 million activities out of 21 million automated. But more importantly, they show where automation breaks down. In one example Riordan shared, invoice creation and delivery creation were 100% automated, while delivery blocks sat at zero percent automation, requiring human intervention every time.

This visibility matters because delivery blocks are often symptoms of broken processes rather than problems themselves. Sometimes they're triggered by price checks, export licenses, or other issues. The question Celonis helps Pfizer answer is: why can one market operate fully automated while another can't?

For example, Pfizer's best-in-class market can process an order from creation to delivery in 20 minutes (limited only by a background job that runs every 15 minutes). Other markets take 16 hours for the same process. When both markets use the same SAP system and standard business processes, the fix isn't automation - it's understanding why certain markets are putting delivery blocks on orders in the first place.

This is where Pfizer's gemba walks come in. Three weeks ago, Riordan and his team were in Colombia and Mexico conducting gemba walkthroughs, helping them compare their processes to Argentina - a similar market with significantly better performance - and facilitating conversations between markets about how they run processes differently. Riordan notes:

It's not all about technology. It's about getting the experts to talk to each other to say how they're running the process better.

The central support model

One insight from Pfizer's implementation is how the company has structured Celonis adoption. Rather than expecting individual markets to become experts in the platform, Pfizer maintains a Center of Excellence with regional teams that support markets - a common approach amongst Celonis install base and an approach that is advocated for by the vendor itself. 

When a market's automation rate is below target, they come to the central team for help. The team uses Celonis dashboards to identify opportunities, then works with the market to deliver improvement projects. But the markets themselves don't need deep technical expertise in process mining. As Riordan explained during the Q&A:

Sometimes the tool can be - within customer service units - just a bit too tricky for them to actually use as a day-to-day tool. So we support them from a regional perspective in terms of giving them the insights.

This approach recognizes that while Celonis provides powerful analytical capabilities, making those capabilities accessible requires interpretation and translation by specialists who understand both the technology and the business context.

Humans, not transactions

Underlying Pfizer's automation drive is a shift in how the company wants its customer service organization to spend time. Currently, customer engagement represents just 14% of activities, with the remainder split across processing sales orders, master data maintenance, billing, disputes, reporting, and continuous improvement.

Pfizer's target is to flip this ratio - getting customer engagement up to 80% while automation and AI handle transactional work. This is what Riordan calls "market model elevation" - moving people from doing transactions on systems to engaging with customers.

The company is deploying agentic AI across multiple use cases to support this shift: scanning sales orders to check prices against contracts, guiding customer creation and licensing processes, automating credit releases based on recurring patterns, handling dispute calculations based on contract terms, and providing conversational analytics rather than static reports.

But again, the sequence matters. Riordan emphasizes:

You've got to start with the problem and then find a use case, if one exists, to support fixing that problem. There's huge pressure on everyone: 'What's your agentic AI strategy?' That's the wrong question. The question we're trying to solve is: How do we improve our metrics? How do we improve our customer experience? The technology we select to hit those metrics and experience is what's important.

When good technology meets poor adoption

A telling moment  in Riordan's presentation came when talking about technology adoption challenges. Pfizer implemented SAP Commerce Cloud to allow hospitals to order online - a logical solution given the volume of manual ordering he mentioned previously. The technology worked perfectly. Adoption was zero.

Hospitals flatly refused to key orders into Pfizer's system. For pharmacies ordering one or two items, Commerce Cloud was ideal. But hospitals ordering 20 items at a time simply wouldn't use it. This forced Pfizer to pivot to Conexiom, which could read orders in any format the customer chose - Excel, PDF, email, fax - without requiring customers to change their behavior.

Riordan describes this as:

A classic example of just because the technology works doesn't mean you're going to get adoption, especially when you're dealing with customers.

With Conexiom, orders are scanned in and then become visible on Commerce Cloud, giving customers what they need without forcing workflow changes. This hybrid approach has driven automation rates from 58% to 85% across Europe in two years.

Throughout the presentation, Riordan emphasized that technology selection must serve business outcomes rather than the other way around. When asked about defining system roles within the organization, he explained that Pfizer has a Solution Architect responsible for determining how systems integrate - both from a technology and data perspective. More importantly, the architect ensures that AI and automation live within core operational systems (SAP, Conexiom) rather than being bolted on through Celonis.

For Pfizer, Celonis remains focused on process mining and intelligence - helping the company understand where to deploy AI and automation, not necessarily executing it. As Riordan notes: 

If SAP can do the AI, we will do it in SAP. If Conexiom does it, we will do it in Conexiom.

The company is exploring AI within Celonis for interrogating process maps and asking questions, but not for building AI workflows to support core processes.

Finally, for Riordan, everything is about the focus on customer experience. As he concludes:

All of this doesn't make sense if our customers don't see the results. We're laser-focused on capturing feedback from our customers through our Qualtrics surveys, but also freeing up our people to speak to customers, to understand how we can support them in terms of running their business and not just being a transactional partner that's basically supplying product to them.