In This Article
Executive Summary
A biotech of about 100 people needs reliable laptops, accounts, networks, and security, and it also needs a handful of GxP systems kept in a validated state. Those are two different kinds of work. Most companies at this size choose between three ways of covering them: a fully in-house team, a managed service provider (MSP), or a hybrid of a small internal team with an MSP and specialist partners. The right biotech IT support model is the one that puts each kind of work with the party best able to do it and prove it was done.
Regulators do not care which model you pick. They care that the company stays accountable. EU GMP Annex 11 says formal agreements must exist with any third party that maintains a computerized system, and it adds that internal IT departments “should be considered analogous.” CISA makes the same point for security: outsourcing to an MSP does not absolve an organization of its risk management duties. So the question is less “who does the work” and more “who can show an inspector the evidence.”
This article maps the work, compares the three models on what each handles well, sets out what an MSP must be able to show before it touches GxP systems (qualified staff, change control, access management, support for validated systems, and a quality agreement), and gives a practical way to decide. It is about the support model. The leadership question is a separate one, covered in our earlier piece on the fractional CIO.
What IT Support Means at 100 People
When leaders at a growing biotech say “IT support,” they usually mean the help desk. Someone cannot log in, a laptop will not connect to the conference room screen, a new hire starts Monday and needs an account. That work is real and it matters, but at 100 people it is only part of the job. A better starting point is to list every kind of recurring IT work the company depends on, then decide who should do each one.
We find it useful to sort the work into six kinds. The lines between them are not perfect, and some tasks belong in two places, but the sorting makes the support model conversation much more concrete.
The Six Kinds of IT Work
| Kind of work | Examples at a 100-person biotech | GxP impact |
|---|---|---|
| 1. End-user and device support | Help desk tickets, laptop setup and replacement, printers, conference rooms, mobile devices, onboarding and offboarding | Usually indirect, except offboarding, which removes access to GxP systems |
| 2. Infrastructure and identity | Microsoft 365 or Google Workspace, the identity provider, single sign-on, networks, Wi-Fi, file storage, cloud tenants, backup | High where GxP applications depend on it (Annex 11 expects infrastructure to be qualified) |
| 3. Security operations | Endpoint protection, patching, monitoring and alert response, email security, vulnerability scans, incident response | High for any system holding GxP records |
| 4. Business application administration | Finance and ERP, HR systems, contract management, collaboration tools | Low, unless the system holds regulated records |
| 5. GxP system administration | User provisioning in the eQMS, document management, LMS, eTMF, safety database, lab systems; configuration changes; periodic access reviews | Direct |
| 6. GxP compliance and validation | Computer system validation (CSV) or computer software assurance (CSA), periodic reviews, supplier assessments, change impact assessments, inspection support | Direct |
A seventh kind of work, IT leadership and governance (strategy, budget, vendor standards, board reporting), covers all six. It is a real need, but it is a different decision from the support model. This article sets it aside.
Where GxP Changes the Picture
Kinds one through four look much like they do at any professional services firm of the same size. Kinds five and six are what make a biotech different. They exist because the company creates or holds records that fall under good practice regulations (GxP is the shorthand for good manufacturing, clinical, laboratory, and other “good practice” rules). At 100 people, a clinical-stage company may have a quality management system, a document management system, a learning management system for GxP training, a trial master file, and access to systems run by its CRO and CDMO partners. Some companies also run a QC or development lab with instrument software.
The regulations that govern those systems reach well past the application itself. EU GMP Annex 11 opens with this line: “The application should be validated; IT infrastructure should be qualified.”3 FDA’s 21 CFR 11.10 requires controls that include limiting system access to authorized individuals and a determination that people who develop, maintain, or use electronic record systems have the education, training, and experience to do their assigned tasks.6 ICH E6(R3), the revised good clinical practice guideline, lists user management, password management, firewall settings, antivirus, security patching, system monitoring, and penetration testing among the security controls a responsible party should consider for computerized systems used in trials.10
Read those lists again and notice who usually does the work. Patching, firewall rules, identity, and backup are core MSP services. That means the help desk and infrastructure team, whoever they are, touch the controls that keep GxP systems in a defensible state. The support model is a quality decision as much as an operations one.
The practical test. For each of the six kinds of work, ask two questions: who does it today, and where is the evidence that it was done? If the answer to the second question is “in the MSP’s ticketing system, and we have never seen it,” that is the gap to close first, whatever model you choose.
Model One: Fully In-House
In a fully in-house model, the company employs its own IT staff for most or all of the six kinds of work. At 100 people, that usually means a small team: an IT manager or director, one or two support and systems people, and perhaps a GxP systems or validation specialist who reports to IT or to quality. Outside parties are used only for specific products (the identity provider, the eQMS vendor) and not for day-to-day operations.
What It Handles Well
An in-house team knows the company. It knows that the lab runs a specific instrument PC that cannot take Windows updates without a change record, that the head of clinical operations travels every other week, and that the board deck goes out the Thursday before each meeting. That knowledge is hard to buy.
- On-site presence. Labs, freezers, instrument PCs, badge readers, and conference rooms need hands. An internal team is there when they fail.
- Training records are yours. The people who administer GxP systems are your employees, trained under your quality system. When an inspector asks for the training record of the person who changed a configuration, you pull it from your own LMS.
- Direct control of change. The team works inside your change control process because it is part of your organization. No contract has to define the handoff.
- Lower friction with quality. When IT and quality share a building and a calendar, the handoffs that slow validation (requirements, test review, release approval) tend to move faster.
Where It Struggles
The weaknesses are mostly about depth and coverage. A team of two or three cannot be expert in identity management, endpoint security, cloud architecture, network engineering, and GxP validation at the same time. It also cannot cover a 24-hour security alert queue. When one person is on vacation or leaves, a large share of the company’s IT knowledge goes with them.
Security operations is the most common gap. Monitoring alerts around the clock, responding to an incident at 2 a.m., and keeping detection tools tuned are specialized, continuous jobs. Few 100-person companies can staff them internally, and most that try end up buying a managed detection service anyway, which makes them a hybrid in practice.
Annex 11 also offers no shortcut for an in-house team. In the section on suppliers and service providers, after requiring formal agreements with third parties, it adds a short sentence: “IT-departments should be considered analogous.”3 In other words, an internal IT department needs the same clarity about responsibilities that a contract would give an outside provider. The 2025 consultation draft of a revised Annex 11 goes further. It names “the internal IT department” alongside vendors and service providers and expects approved procedures with internal IT that cover much of what a contract with an MSP would cover.4 That text is still a draft, and the January 2011 Annex 11 remains the version in force, but the direction is plain. Being in-house does not remove the need to write down who does what.
A Word on Staffing Ratios
Leaders often ask for a rule of thumb: how many IT staff should a 100-person company have? Benchmarks exist. Computer Economics (now part of Avasant) publishes IT staffing ratios for more than a dozen job functions, including measures such as users per help desk staff member by organization size.14 The detailed figures are sold to subscribers and are not public, so we do not quote them here. More to the point, a general benchmark does not account for GxP work. A cross-industry help desk ratio says nothing about how many hours your access reviews, periodic reviews, and change impact assessments take.
A better approach is to size by workload. Count the things that drive effort and estimate the hours each one takes:
- Monthly help desk tickets, and the share that need someone on site
- Devices under management, and how often they are replaced
- New hires and departures each month (each one is an onboarding or offboarding task across every system)
- The number of GxP systems, and for each one the frequency of access reviews, periodic reviews, and vendor releases
- Change records raised per quarter that touch GxP systems or their infrastructure
- Security alerts that need a human decision
Six months of ticket data and a list of GxP systems will give you a far better estimate than any ratio. It will also show which kinds of work are steady enough to staff internally and which are spiky or specialized enough to buy.
Model Two: A Managed Service Provider
In this model, an outside company provides most of the six kinds of work under a contract with service levels. The MSP runs the help desk (usually remote, sometimes with scheduled on-site visits), manages devices and identity, operates security tools, and may administer business applications. Some MSPs focus on life sciences and offer GxP services. Many do not.
What It Handles Well
CISA’s guidance for MSP customers puts the appeal plainly: IT services can be complex and time-consuming for many organizations to manage on their own, and MSPs can help reduce costs and support efficient operations for organizations of all sizes.1 For a 100-person biotech, the specific strengths are:
- Depth across specialties. An MSP has network engineers, identity specialists, and security analysts on staff. You get access to those skills without hiring each one.
- Coverage. Help desk hours can extend across time zones, and security monitoring can run around the clock.
- Standard tooling. Mature MSPs bring their own device management, patching, backup, and ticketing tools, already set up and documented.
- Easy scaling. Adding 30 people after a financing round is a contract change, not a hiring campaign.
Where It Struggles
The same CISA document is just as plain about the downside. It states that outsourcing the management of networks, cloud infrastructure, applications, and devices to an MSP “does not absolve an organization from risk management responsibilities,” and that an MSP relationship can introduce new risks by adding third-party attack surfaces.1 A 2022 joint advisory from the cybersecurity authorities of the United States, United Kingdom, Australia, Canada, and New Zealand warned that threat actors can use a vulnerable MSP as an initial access point into many customer networks at once.2
For a biotech, three more limits matter:
- GxP depth varies widely. An MSP that serves law firms and dental practices may run an excellent help desk and still have no idea what a validated state is, why a Windows update on a lab PC needs a change record, or what an audit trail review involves.
- Lab support is often thin. Remote-first MSPs are built for laptops and cloud services. Instrument PCs, lab networks, and equipment that cannot be patched on the normal schedule need someone who understands both the instrument and the quality rules around it.
- Records live in their systems. Tickets, change logs, and access requests sit in the MSP’s tools. Unless the contract gives you access and retention rights, you may not be able to produce them when an inspector asks.
The Independence Advantage
There is one place where an MSP can help a small biotech meet a regulatory expectation that is hard to meet in-house. FDA’s data integrity guidance recommends that “the system administrator role, including any rights to alter files and settings, should be assigned to personnel independent from those responsible for the record content.”8 PIC/S guidance on data integrity says much the same. It notes that system administrators should normally have no involvement or interest in the outcome of the data, gives the example that QC supervisors should not administer their own lab systems, and suggests that IT administrators outside the quality and production functions typically fill the role.9
At 100 people, finding an administrator who is independent of the data can be hard. PIC/S recognizes this and allows smaller organizations to have a nominated person in the quality unit or production hold administrator access, with conditions: a separate restricted account for routine work, and administrator activity recorded and approved in the quality system.9 An MSP with properly trained staff offers another route. Its administrators have no stake in a batch result or a trial data point. That benefit holds only if the MSP’s staff are trained on your procedures, use named accounts, and work under your change control. Without those conditions, an independent administrator is simply an uncontrolled one.
Watch for shared MSP accounts. Many MSPs use a single technician account per customer, or a shared account inside each tool, so any of their staff can pick up a ticket. FDA’s guidance explains that when login credentials are shared, a unique individual cannot be identified through the login.8 CISA advises that MSP accounts should not be assigned to administrator groups by default, should be limited to the systems the MSP manages, and should be disabled when not in use.1 Ask how the MSP handles this before you sign, not after the first inspection.
Model Three: The Hybrid
The hybrid model keeps a small internal team, often one to three people, and buys the rest. An MSP covers commodity work and around-the-clock coverage. Specialist partners cover work that needs deep expertise but not full-time attention, such as validation, lab systems, or a security assessment. The internal team owns the parts that require company knowledge, quality system access, and accountability.
This model is common for a reason. Deloitte’s 2024 Global Outsourcing Survey, which drew on insights from more than 500 executives globally, found that 70% had selectively insourced work previously held by a third party over the previous five years, while 80% planned to maintain or increase their investment in third-party outsourcing.15 Those findings are not specific to biotech, but they describe the pattern we see: companies are not choosing all-in or all-out. They are pulling specific work back in house while continuing to buy the rest.
The Shape of a Hybrid Team
Small Internal IT Team
Owns the service relationships, GxP system administration, onboarding and offboarding sign-off, on-site lab and office support, and the company’s IT procedures. Works inside the quality system and holds the training records.
Managed Service Provider
Runs the help desk queue, device management, patching, identity operations, backup, and security monitoring. Works to your change control for anything that touches GxP infrastructure, under a contract and quality agreement.
Validation and GxP Systems Partner
Supports CSV or CSA work on new systems and major releases, periodic reviews, supplier assessments, and inspection preparation. Engaged by project or on a small retainer.
Other Specialist Partners
Lab informatics and instrument integration, penetration testing, network design for a new site, or a one-time cloud migration. Engaged when the need arises.
Who Does What
The hybrid works only when every task has one clear owner. CISA recommends that customers and MSPs agree a shared responsibility model that sets out which party applies patches, maintains hardware, and trains employees, and which responsibilities are shared.1 The table below is a starting point we adapt with clients. “R” means responsible for doing the work and “A” means accountable for its being done and evidenced.
| Task | Internal IT | MSP | Specialist partner | Quality |
|---|---|---|---|---|
| Help desk tickets (non-GxP) | A | R | ||
| Onboarding and offboarding across all systems | A | R (identity, devices) | Confirms GxP training before access | |
| Patching of servers and endpoints | A | R | ||
| Patching of lab instrument PCs | A, R | Supports on request | Advises on impact | Approves change |
| Change to GxP infrastructure (for example, a network or backup change) | A | R under your change control | Approves | |
| User provisioning in GxP applications | R | A (system owner approves) | ||
| Periodic access review of GxP systems | R (prepares) | Provides identity data | A | |
| Validation of a new GxP system | R (system owner support) | R (executes) | A | |
| Security monitoring and first response | A | R | Informed if GxP data is affected | |
| Backup and restore testing | A | R | Reviews evidence for GxP systems |
Your own version will differ. Some companies put GxP application administration with quality rather than IT, which can work if the administrator is still independent of the data, as discussed above. What matters is that the table exists, that each party has signed off on it, and that it matches the contract.
Where Hybrids Fail
Hybrids fail where one party’s work hands off to another. We see three patterns often enough to call them out:
- The internal person becomes a ticket router. A single internal IT lead spends the day forwarding requests to the MSP and chasing them, with no time left for GxP administration or vendor oversight. The fix is a clear intake process in which users go straight to the MSP for routine requests.
- Nobody owns the gap between two vendors. The MSP manages the identity provider. The eQMS vendor manages the application. A departed employee’s single sign-on account is disabled, but a local account in the eQMS was never removed. Each party did its part. The company still has an access finding.
- Specialists are engaged too late. A validation partner is called a month before go-live and finds that requirements were never approved and the vendor’s documentation was never assessed. The work that should have run alongside the project now runs after it.
Keep the internal team small, but not empty. We are cautious about a “zero internal IT” version of the hybrid for a GxP company. Someone inside the company has to hold system owner responsibilities, approve changes in the quality system, and explain the arrangements to an inspector. ICH E6(R3) says the sponsor should have access to relevant information, such as SOPs and performance metrics, to select and oversee service providers.10 That oversight needs a named person with time to do it.
What a GxP-Capable MSP Must Show You
Whether you choose a full MSP model or a hybrid, the MSP will touch infrastructure and security controls that GxP systems rely on. Before it does, it needs to show you five things. These come straight from the requirements already in force, not from a wish list.
1. Qualified Staff
Annex 11 says all personnel “should have appropriate qualifications, level of access and defined responsibilities to carry out their assigned duties.”3 Part 11 requires a determination that people who maintain electronic record systems have the education, training, and experience for their tasks.6 For an MSP, that translates into specific questions:
- Which named staff, or which defined team, will work on your GxP-supporting infrastructure?
- What GxP training have they had, and can they show records?
- Will they be trained on your procedures (change control, deviation reporting, access management), and who will keep those training records?
- What happens when staff rotate? How quickly is a new technician trained before touching your systems?
Where a specialist partner advises on systems used in manufacturing, processing, packing, or holding of drug products, FDA’s rule on consultants also applies. 21 CFR 211.34 requires such consultants to have sufficient education, training, and experience to advise on the subject, and requires records of their name, address, qualifications, and the type of service they provide.7 This record is easy to overlook for validation and IT partners.
2. Change Control
Annex 11 is short on this point: any changes to a computerized system, including system configurations, should only be made in a controlled manner under a defined procedure.3 The difficulty for an MSP is that its business runs on automation. Patches roll out on a schedule. Firewall rules change through a ticket. Backup policies are updated across all customers at once.
The MSP does not have to stop automating. It has to agree, in writing, which of your systems and infrastructure components are in GxP scope, and treat changes to those components differently. In practice that means:
- A list of in-scope components (servers, network segments, storage, backup jobs, identity settings, lab PCs) that both parties maintain
- Standard, pre-approved changes (routine patches tested in a defined way) separated from changes that need your impact assessment and approval
- An emergency change path for security fixes, with documentation completed afterward
- Change records you can see and keep, either in your system or exported on a schedule
EU GMP Chapter 7 adds an expectation that MSPs often miss: the contract acceptor should not make unauthorized changes, outside the terms of the contract, that may adversely affect the quality of the outsourced activities.5 If your MSP changes its own tools or hosting provider in a way that affects your environment, you need to know before it happens.
3. Access Management
Access management is where the handoffs between an MSP and the company show most clearly. Annex 11 requires that creation, change, and cancellation of access authorizations be recorded.3 Part 11 requires limiting system access to authorized individuals.6 PIC/S expects administrator rights to be strictly controlled, least-privilege roles, and user lists that can be used in periodic reviews.9 The joint MSP advisory adds the security view: customers should ensure contracts require multifactor authentication on all MSP accounts used to access customer environments, and should identify and disable accounts that are no longer in use.2
Ask the MSP to show you, not tell you:
- How each technician gets a named account in your environment, and how it is removed when they leave the MSP
- How privileged access is granted (standing, or just in time for a specific task) and logged
- How your offboarding request flows through their process, and how quickly access is removed from every system they manage
- What reports they can give you for your periodic access reviews
4. Support for Validated Systems
An MSP rarely validates your GxP applications, but its work keeps them in a validated state or takes them out of one. Annex 11 expects periodic evaluation of computerized systems covering incidents, problems, upgrade history, performance, reliability, security, and validation status.3 Much of that information comes from the MSP’s records. The MSP should be ready to:
- Provide infrastructure qualification evidence for what it manages, or support you in creating it. ISPE’s GAMP Good Practice Guide on IT infrastructure control and compliance sets out a risk-based approach to qualifying and controlling infrastructure platforms, including outsourced and third-party data centers.12
- Report incidents that affect GxP systems in a form your quality team can assess. Annex 11 expects all incidents to be reported and assessed, with root cause identified for critical ones.3
- Run and document backup restore tests for GxP data on a defined schedule.
- Take part in business continuity planning, since Annex 11 expects continuity arrangements for systems that support critical processes to be documented and tested.3
FDA’s February 2026 guidance on computer software assurance is helpful here. It is CDRH and CBER guidance written for device manufacturers, which drug companies often use by analogy. It recommends a risk-based approach to evaluating the vendor of software or service, and its approach allows companies to draw on validation activities performed by other entities, such as developers, suppliers, and cloud service providers.11 An MSP with good documentation can reduce your validation effort. One with poor documentation adds to it.
5. A Quality Agreement
Annex 11 requires formal agreements with third parties that “provide, install, configure, integrate, validate, maintain (e.g. via remote access), modify or retain” a computerized system or related service, and those agreements “should include clear statements of the responsibilities of the third party.”3 PIC/S expects quality agreements with service providers that include specific provisions for data integrity, including a requirement to notify the contract giver of data integrity failures.9 A standard MSP master services agreement does not do this. It covers price, service levels, liability, and termination. It is not written for GxP.
The usual answer is a separate quality agreement (or a GxP schedule to the main contract) that quality signs. The next section covers what goes in it.
A good sign in an MSP proposal. The MSP asks to see your list of GxP systems, your change control SOP, and your access management SOP before it prices the work, and its proposal names a GxP-trained service lead. An MSP that does not ask these questions is pricing a generic service.
Writing the Agreement
Two documents usually govern an MSP relationship at a biotech. The commercial agreement (a master services agreement plus statements of work and service levels) is owned by finance, legal, and IT. The quality agreement is owned by quality and IT together. They need to agree with each other, and the quality agreement should say which document wins if they conflict on a GxP matter.
What the Regulators Point To
EU GMP Chapter 7 was written for outsourced GMP activities in general, and its principles carry over to IT services. It says the contract giver is ultimately responsible for ensuring processes are in place to control outsourced activities, that the contract should specify each party’s responsibilities and communication processes, and that the contract acceptor should not subcontract any of the work without the contract giver’s prior evaluation and approval.5 That last point matters for MSPs, many of which use subcontracted after-hours support or a separate security operations provider. ICH E6(R3) makes the same point for clinical work: sponsor oversight extends to activities further subcontracted by the service provider.10
The 2025 consultation draft of Annex 11 is the most detailed public text on this subject. It is not final and should not be cited as a requirement, but it is a useful checklist. It says that when a regulated company relies on a service provider, it “remains fully responsible for these activities,” should oversee the provider against agreed service levels and key performance indicators, and should have a contract that covers nine topics.4
| Topic in the 2025 draft Annex 11 (section 7.5) | What it means for an MSP agreement |
|---|---|
| Activities and documentation to be provided | List the services in GxP scope and the records the MSP will produce (change records, incident reports, backup test results, access reports) |
| Company procedures and regulatory requirements to be met | Name the SOPs the MSP must follow, such as change control, deviation reporting, and access management |
| Regular, ad hoc, and incident reporting and oversight, including SLAs and KPIs | Set GxP-relevant measures, such as time to remove access after departure and time to report an incident affecting a GxP system |
| Conditions for supplier audits | Give your quality team, or an auditor acting for you, the right to audit with reasonable notice |
| Support during regulatory inspections | Commit the MSP to provide records and, if requested, a knowledgeable person during an inspection |
| Resolution of issues from operations, audits, and inspections | Agree how findings become corrective actions, and who tracks them |
| Communication of quality and security issues | Define notification timelines for security incidents and data integrity failures |
| Exit strategy that keeps the regulated company in control of its data | Require handover of records, configurations, credentials, and documentation at the end of the contract |
| Release of new versions and the chance to test before release | Agree advance notice of changes to MSP tools or platforms that affect your environment |
Security Terms That Belong in the Contract
The 2022 joint advisory recommends that customers make sure their contractual arrangements specify that the MSP implements the baseline security measures it describes.2 It also recommends that contracts transparently identify ownership of security roles and responsibilities, and detail how and when the MSP notifies the customer of an incident affecting the customer’s environment.2 The NIST Cybersecurity Framework 2.0 treats supply chain risk as a governance matter, including the expectation that security requirements are built into contracts with suppliers and that relevant suppliers are included in incident planning, response, and recovery.13
In practice, the security schedule should cover multifactor authentication on all MSP accounts, logging of MSP activity in your environment with logs you can access, incident notification timelines, backup and restore testing, and the right to receive or review the MSP’s own security assessments.
The Exit Clause Nobody Reads Until They Need It
The most common gap we see is the end of the relationship. Companies change MSPs after a financing round, an acquisition, or a bad year of service. If the contract does not require the MSP to hand over change history, ticket records, configuration documentation, and administrator credentials, the new provider starts with no history and the company may lose records that support its validated state. NIST CSF 2.0 includes this in its supply chain category: risk management plans should include provisions for activities that occur after a partnership or service agreement ends.13 Write the exit terms when the relationship is new and both sides are cooperative.
Records retention outlasts contracts. GxP records, including the change and access records that support a system’s validated state, often need to be kept for years after an MSP contract ends. Make sure the agreement either transfers those records to you on a schedule or obliges the MSP to keep them for your retention period and produce them on request.
How to Decide
No model is right for every 100-person biotech. The decision depends on a few facts about the company, most of which can be gathered in a few weeks.
Six Questions That Drive the Choice
- How much GxP work does the company own? A company whose only GxP systems are a SaaS eQMS and an eTMF hosted by its CRO has a very different load from one running a QC lab with instrument software and a LIMS. More owned GxP systems push toward more internal capability or a strong GxP specialist partner.
- How much work needs hands on site? Labs, multiple offices, and physical security systems need presence. A remote-first MSP may not be able to provide it at an acceptable response time.
- How fast will the company change? If headcount may double after the next financing, or the company is approaching a first commercial launch, the model needs to scale. An MSP scales faster than hiring.
- What is the security exposure? Public companies, companies holding patient data, and companies with valuable intellectual property need monitoring and response that a small internal team cannot staff around the clock.
- Who will oversee the arrangement? Every model needs a named person inside the company who can manage providers, approve GxP changes, and speak to inspectors. If that person does not exist, fix that first.
- What is the company’s tolerance for key-person risk? A one-person internal team is a single point of failure. A hybrid with a documented MSP layer reduces that risk.
A Simple Way to Read the Answers
| Company profile | Model that usually fits | Watch out for |
|---|---|---|
| Clinical stage, mostly SaaS GxP systems, no internal lab, one office | MSP-led hybrid: one internal IT and GxP systems lead, an MSP for operations and security, a validation partner on call | The internal lead spending most of the day routing tickets |
| Clinical stage with a research lab and some GxP lab work | Balanced hybrid: two or three internal staff including on-site lab support, MSP for infrastructure and security, lab informatics partner as needed | Lab PCs falling outside both the MSP’s scope and the internal team’s change control |
| Approaching commercial launch, or with in-house GMP manufacturing or QC | Larger internal team with GxP systems administration in house, MSP for security monitoring and commodity work | Growing the internal team without written procedures for the internal IT department, which Annex 11 treats as analogous to a third party |
| Very early, few or no owned GxP systems | MSP with a clear plan for when GxP systems arrive | Signing a long MSP contract with no GxP terms, then trying to add them later |
Headcount alone does not decide the model. Two companies of 100 people can need very different support depending on what they make, where they work, and which systems they own.
Running the Selection
Whichever direction the answers point, the steps to get there are similar.
Map the Work
List the six kinds of work, the systems in each, and the current owner. Pull six months of ticket data. Mark every system and infrastructure component that is in GxP scope.
Decide What Stays Inside
Choose the tasks that need company knowledge, on-site presence, or quality system authority. Those define the internal team. Everything else is a candidate to buy.
Write Requirements Before Talking to Vendors
CISA suggests procurement start with a requirements list gathered from every department that uses the services, covering security and continuity as well as function.1 Add the five GxP requirement areas from this article.
Assess Providers on Evidence
Annex 11 says the competence and reliability of a supplier are key factors in selection, with the need for an audit based on risk.3 Ask for sample change records, training records, and access reports from another life sciences customer (with names removed), not just a capabilities deck.
Sign Both Agreements Together
Finalize the commercial agreement and the quality agreement at the same time, with the responsibility table attached to both. Agreeing GxP terms after signature is much harder.
Transition Under Change Control
Moving from one provider to another changes who administers GxP infrastructure. Treat the transition as a change: assess impact, train the new staff on your procedures, move records, and confirm access lists before go-live.
Review Performance and Compliance
Chapter 7 expects the contract giver to monitor and review the contract acceptor’s performance.5 Hold quarterly reviews that cover GxP measures alongside ticket statistics, and include the MSP in your supplier audit schedule.
Signals That the Model Needs to Change
A support model chosen at 100 people will not fit at 250. Watch for these signs:
- Access review findings that trace back to gaps between the MSP and application vendors
- Change records for GxP infrastructure that are missing, late, or written after the fact
- The internal IT lead spending most of the week on tickets rather than on GxP administration and provider oversight
- A new site, a new lab, or a first GMP activity that the current MSP has no experience supporting
- An inspection or audit request for records that took days to obtain from the MSP
Any one of these is worth a conversation. Two or more usually mean the split of work needs to be redrawn.
Conclusion
The choice between in-house, outsourced, and hybrid IT support is often framed as a question of price and service levels. For a biotech, it is also a question of evidence. Every model can work, and every model can fail an inspection. What separates them is whether the company knows which party does each kind of work, whether that party is trained and controlled under the company’s procedures, and whether the records reach the company when it needs them. Annex 11, Part 11, ICH E6(R3), and the data integrity guidance all point the same way: the regulated company stays accountable, and internal IT is held to the same clarity as an outside provider. In our experience, the hybrid model fits most 100-person biotechs best, because it keeps GxP ownership and on-site knowledge inside while buying depth and coverage outside. It works only if the handoffs between parties are written down.
Sakara Digital works with pharma and biotech organizations deciding how to structure IT support around their GxP systems, from mapping the work and drafting the responsibility table to assessing MSPs and writing the quality agreement. If you are weighing a change to your support model and want an independent view of the options, we are happy to have that conversation.
For Further Reading
For Further Reading
- The Business Case for a Fractional CIO in a Clinical-Stage Biotech
- Access Reviews That Actually Catch Problems in GxP Systems
- The Quality Agreement Clauses That Break When You Digitize
- Cloud Vendor Audits for GxP SaaS: Using SOC 2 Reports Without Overrelying on Them
- What a Biotech Under 200 People Should Not Build In-House
- Talent Strategy for Pre-Revenue Biotech: When to Hire vs. Contract
References & Sources
- Cybersecurity and Infrastructure Security Agency (CISA), National Risk Management Center. “Risk Considerations for Managed Service Provider Customers.” CISA Insights, September 2021. https://www.cisa.gov/sites/default/files/publications/cisa-insights_risk-considerations-for-msp-customers_508.pdf
- CISA, NSA, FBI, NCSC-UK, ACSC, CCCS, and NCSC-NZ. “Protecting Against Cyber Threats to Managed Service Providers and their Customers.” Alert AA22-131A, May 11, 2022. https://www.cisa.gov/news-events/cybersecurity-advisories/aa22-131a
- European Commission. “EudraLex Volume 4, Annex 11: Computerised Systems.” January 2011 (in force June 30, 2011). https://health.ec.europa.eu/system/files/2016-11/annex11_01-2011_en_0.pdf
- European Commission. “EudraLex Volume 4, Annex 11: Computerised Systems.” Stakeholder consultation draft, 2025 (not final). https://health.ec.europa.eu/document/download/40231f18-e564-4043-94de-c031f813d38b_en?filename=mp_vol4_chap4_annex11_consultation_guideline_en.pdf
- European Commission. “EudraLex Volume 4, Chapter 7: Outsourced Activities.” Revision 1, June 2012 (in operation January 31, 2013). https://health.ec.europa.eu/system/files/2016-11/vol4-chap7_2012-06_en_0.pdf
- U.S. Code of Federal Regulations. “21 CFR 11.10: Controls for Closed Systems.” Electronic Code of Federal Regulations. https://www.ecfr.gov/current/title-21/chapter-I/subchapter-A/part-11/subpart-B/section-11.10
- U.S. Code of Federal Regulations. “21 CFR 211.34: Consultants.” Electronic Code of Federal Regulations. https://www.ecfr.gov/current/title-21/chapter-I/subchapter-C/part-211/subpart-B/section-211.34
- U.S. Food and Drug Administration. “Data Integrity and Compliance With Drug CGMP: Questions and Answers.” Guidance for Industry, December 2018. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/data-integrity-and-compliance-drug-cgmp-questions-and-answers
- Pharmaceutical Inspection Co-operation Scheme (PIC/S). “PI 041-1: Good Practices for Data Management and Integrity in Regulated GMP/GDP Environments.” July 1, 2021. https://picscheme.org/docview/4234
- International Council for Harmonisation. “ICH E6(R3) Guideline for Good Clinical Practice.” Step 4, January 6, 2025. https://database.ich.org/sites/default/files/ICH_E6%28R3%29_Step4_FinalGuideline_2025_0106.pdf
- U.S. Food and Drug Administration. “Computer Software Assurance for Production and Quality Management System Software.” Guidance for Industry and FDA Staff, February 3, 2026. https://www.fda.gov/media/188844/download
- ISPE. “GAMP Good Practice Guide: IT Infrastructure Control and Compliance (Second Edition).” August 2017. https://ispe.org/publications/guidance-documents/gamp-it-infrastructure-control-compliance
- National Institute of Standards and Technology. “The NIST Cybersecurity Framework (CSF) 2.0.” NIST CSWP 29, February 26, 2024. https://nvlpubs.nist.gov/nistpubs/CSWP/NIST.CSWP.29.pdf
- Computer Economics (Avasant). “IT Staffing Ratios: Benchmarking Metrics for 17 IT Job Functions.” https://computereconomics.com/it-staffing-ratios
- Deloitte. “Global Outsourcing Survey 2024: Multidimensional Sourcing, Orchestrating the Extended Workforce Ecosystem.” 2024. https://www.deloitte.com/global/en/issues/work/global-outsourcing-survey.html








Your perspective matters—join the conversation.