CDMOHUBCDMOHUB
Pricing
CDMOHUB

Connecting biotech and pharma teams with verified manufacturing partners worldwide.

Platform

  • Find CDMOs
  • CDMO Matchmaking
  • Expert Matchmaking
  • Modality Guides
  • Supplier Maps
  • Join as Partner

Company

  • Blog
  • Briefings
  • About Us
  • Contact

Legal

  • Terms of Service
  • Privacy Policy

© 2026 CDMOHUB. All rights reserved.

CDMO Hub, Inc. Registered in Delaware, USA

    CDMOHUBCDMOHUB
    Pricing
    CDMOHUB

    Connecting biotech and pharma teams with verified manufacturing partners worldwide.

    Platform

    • Find CDMOs
    • CDMO Matchmaking
    • Expert Matchmaking
    • Modality Guides
    • Supplier Maps
    • Join as Partner

    Company

    • Blog
    • Briefings
    • About Us
    • Contact

    Legal

    • Terms of Service
    • Privacy Policy

    © 2026 CDMOHUB. All rights reserved.

    CDMO Hub, Inc. Registered in Delaware, USA

    CDMOHUBCDMOHUB
    Pricing
    CDMOHUB

    Connecting biotech and pharma teams with verified manufacturing partners worldwide.

    Platform

    • Find CDMOs
    • CDMO Matchmaking
    • Expert Matchmaking
    • Modality Guides
    • Supplier Maps
    • Join as Partner

    Company

    • Blog
    • Briefings
    • About Us
    • Contact

    Legal

    • Terms of Service
    • Privacy Policy

    © 2026 CDMOHUB. All rights reserved.

    CDMO Hub, Inc. Registered in Delaware, USA

    CDMOHUBCDMOHUB
    Pricing
    1. Home
    2. /Guides
    3. /How to Choose a CDMO
    Buyer Guide

    How to Choose a CDMO

    Choosing a contract manufacturer is the highest-consequence sourcing decision most drug programs make. A wrong pick surfaces late — during tech transfer, a pre-approval inspection, or a launch-quantity scale-up — when switching costs are at their peak. This guide is the selection framework we see disciplined sourcing teams actually use, in the order they use it.

    Last updated August 4, 2026

    On this page

    1. 01Define the requirement before you shortlist
    2. 02Test capability and capacity fit — separately
    3. 03Read the regulatory and inspection record
    4. 04Audit the quality system, not the tour
    5. 05De-risk the tech transfer — where programs actually fail
    6. 06Structure the commercial terms for the program you will actually run
    7. 07Where to look — and what geography actually costs you
    8. 08Red flags

    01Define the requirement before you shortlist

    Most bad CDMO selections are decided before a single supplier is contacted, by starting from a list of names instead of a specification. Write these down first — they are what turns a vague search into a filter, and they are what a serious CDMO will ask for on the first call anyway.

    Molecule and dosage form
    Small molecule, peptide, monoclonal antibody, cell or gene therapy, oligonucleotide, ADC — and the presentation (oral solid, sterile injectable, lyophilized vial, prefilled syringe). This single pair eliminates most of the market immediately, because the equipment trains do not transfer between them.
    Development stage and the next milestone
    Preclinical tox supply, first-in-human, Phase II, registration batches, or commercial launch. A partner optimized for fast clinical supply is often the wrong partner for a validated commercial process, and vice versa. Be explicit about the next milestone, not the end state.
    Scale, now and at peak
    Batch size and annual volume for the current milestone AND the realistic commercial case. The most expensive failure mode is a partner who can make your Phase I material and cannot make your launch quantities — the scale ceiling should be validated up front, not discovered later.
    Target markets and the regulatory consequence
    US, EU, Japan, China, or rest-of-world. Markets determine which regulatory filings, inspection regimes, and site registrations the facility must already carry. Adding a market late usually means adding a site.
    Handling constraints
    Potency and containment (OEB band), controlled substances, cold chain, single-use versus stainless, and any segregation requirement (e.g. beta-lactam, cytotoxic, live viral). These are facility attributes — they cannot be negotiated into existence.
    Timeline and what is actually fixed
    Separate the dates driven by clinical or regulatory commitments from the dates that are preference. Capacity is booked in advance; knowing which of your dates are real determines how much leverage you have and where you can trade.

    A shortlist built from this specification is defensible. A shortlist built from brand recognition is not.

    02Test capability and capacity fit — separately

    Capability and capacity are different questions, and conflating them is one of the most common sourcing errors. Capability asks whether the site can perform the operation at all. Capacity asks whether it can perform it for you, in your window, at your volume. A CDMO can be a genuine expert in your process and still be the wrong answer because the relevant suite is fully booked for eighteen months.

    Test capability with evidence, not with a capabilities deck: ask which specific suite and equipment train would run your product, how many campaigns of this type it has run in the last twenty-four months, and whether the process was developed there or transferred in. Then test capacity separately: ask what is currently booked in that suite, what the realistic slot is, and what a reservation would cost. A partner who answers the second question vaguely is telling you something.

    Below are the capabilities most frequently listed across the CDMO Hub directory, with live counts. Use them as a starting filter — then verify the specific claim at the specific site.

    • Process Development963
    • Analytical Services873
    • API Manufacturing523
    • Formulation Development621
    • Aseptic Processing329
    • Solid Dose Manufacturing365

    Live counts of CDMOs listing each capability in the CDMO Hub directory.

    03Read the regulatory and inspection record

    A facility’s inspection history is the single most objective quality signal available to a buyer, and it is largely public. Treat it as a required diligence step rather than a nice-to-have, because it is the one input that does not come from the supplier’s marketing.

    For US-relevant sites, the FDA classifies the outcome of each inspection into three categories. **NAI** (No Action Indicated) means no objectionable conditions were found. **VAI** (Voluntary Action Indicated) means objectionable conditions were found but the agency is not prepared to take or recommend action — these are common and are not automatically disqualifying. **OAI** (Official Action Indicated) means regulatory or administrative action is warranted; an OAI classification on a site you intend to use is a material finding that requires explanation and evidence of remediation. Ask for the establishment inspection report, the Form 483 observations, and the company’s responses — a confident partner shares them.

    Look at the pattern, not a single data point. Repeat observations across consecutive inspections in the same system — particularly data integrity, aseptic practice, or investigations and CAPA — indicate an unresolved systemic issue rather than an isolated finding. Also check recency: a clean record from an inspection five years ago says much less than a recent one, especially if the site has changed ownership, expanded, or added a new product type since.

    Outside the US, the equivalent signals are EU GMP certificates and any non-compliance reports published in EudraGMDP, the national competent authority’s findings, PMDA status for Japan, and membership of the PIC/S scheme. Confirm that the certificate covers the specific dosage form and operation you need — site-level certification does not imply every activity at that site is covered.

    Finally, verify at the site level, never the corporate level. A group can hold an excellent overall record while the specific facility that would make your product has a different history. Inspection records attach to establishments, and so should your diligence.

    04Audit the quality system, not the tour

    Every site visit looks impressive. The quality system is what determines whether the site performs on a bad day, and it is assessed through documents and behavior rather than through the walk-through. Ask for these specifically:

    Deviation and CAPA record
    Volume, aging, and closure rate over the last twelve months. A low deviation count is not automatically good — it can mean under-reporting. What matters is that deviations are found, investigated to genuine root cause, and closed on time. Ask to see the overdue list.
    Change control
    How changes are evaluated for regulatory impact and how customers are notified. You need contractual certainty that a change affecting your product cannot be made without your agreement — this belongs in the quality agreement, not in goodwill.
    Data integrity controls
    Audit trails enabled and reviewed, access control and segregation of duties, and no shared logins on GxP systems. Data integrity is the most common theme in serious regulatory findings across the industry, so probe it directly.
    Supplier and raw material qualification
    How the CDMO qualifies its own suppliers, and whether single-source materials in your process are identified with a mitigation. Your supply chain risk includes theirs — inherited, whether or not you looked.
    The quality agreement itself
    Who decides batch disposition, who owns investigations, notification timelines, audit rights including for-cause access, and how long records are retained. Negotiate this in parallel with the commercial agreement; leaving it to the end concedes the leverage.
    Personnel stability
    Turnover in the quality unit and in operations, and whether the people in your kickoff meeting are the people who will run the campaign. Continuity in the QP or quality head is a strong signal; churn there is a leading indicator of trouble.

    05De-risk the tech transfer — where programs actually fail

    Tech transfer is where selection decisions are proven right or wrong. The capability existed, the contract was signed, and the program still slips two quarters because the process did not behave the same way in the receiving site. Assess transfer risk explicitly, as its own criterion, before signing.

    Ask how the CDMO runs a transfer: whether there is a defined protocol with acceptance criteria agreed in advance, who the named technical lead is, whether engineering or demonstration batches are planned before GMP material, and how analytical methods are transferred and co-validated. Method transfer is the step most often underestimated — a process that transfers cleanly can still stall for months on an analytical method that will not reproduce in the receiving lab.

    Probe the gap between the originating and receiving equipment. Differences in mixing geometry, heat transfer, filtration area, hold times, or single-use versus stainless contact materials are where a robust process quietly becomes a variable one. The right answer to "will this scale?" is a comparability plan, not reassurance.

    Ask for references from transfers of a comparable modality and scale — and ask specifically about one that went badly. How a partner describes a difficult transfer, what they attribute it to, and what they changed afterwards tells you more about how they will handle your problem than any successful case study.

    Agree up front what happens when the transfer misses its criteria: who pays for repeat batches, how the timeline is re-planned, and what the exit path looks like. Exit terms are cheapest to negotiate when neither side expects to need them.

    06Structure the commercial terms for the program you will actually run

    Price per batch is the least important commercial term. What determines total cost and flexibility is the structure around it:

    Capacity reservation and take-or-pay
    Reserving a slot secures the timeline but commits you to pay for it whether or not your program is ready. Match the reservation window to the confidence in your own milestones, and negotiate the reschedule right explicitly.
    Minimum order quantities and batch economics
    A minimum batch size that exceeds your clinical demand means paying for material you will discard. Model cost per usable unit rather than cost per batch — the ranking often changes.
    Failed-batch and yield risk
    Who bears the cost of a batch that fails specification, and does that answer change between engineering, clinical, and validation batches? This is one of the largest hidden cost swings in a contract.
    Raw material and pass-through handling
    Whether materials are procured by you or the CDMO, what margin is applied, how price changes pass through, and who owns the inventory. Long-lead materials should be identified and ordered against an agreed forecast.
    Intellectual property and process ownership
    Establish clearly what improvements developed during the work belong to you and what remains the CDMO’s background IP. Ambiguity here is what prevents a later transfer to a second source.
    Exit and second-source rights
    The right to transfer the process to another site, the documentation you receive, and the support obligation on exit. Negotiate this at signature — it is nearly impossible to obtain later, and its absence is what converts a supplier problem into a supply crisis.

    For commercial programs, plan for dual sourcing or at least a documented second-source path. Single-sourcing a commercial product is a business continuity decision, and it should be made deliberately rather than by default.

    07Where to look — and what geography actually costs you

    Geography affects far more than unit price. It determines travel time for person-in-plant and audits, time-zone overlap for the daily problem-solving that a transfer requires, shipping lanes and cold-chain risk, import and export licensing, and which regulatory authority inspects the site. A lower quoted price with a twelve-hour time difference and a two-day travel commitment is not obviously cheaper once a program hits a problem.

    Manufacturing capacity is genuinely global, and concentration varies sharply by modality — sterile fill/finish, API synthesis, and cell and gene therapy have different maps. The live facility distribution across the CDMO Hub directory is shown below.

    • United States724
    • Germany336
    • India326
    • China305
    • Italy213
    • France184

    Manufacturing facilities recorded in the CDMO Hub directory, by country. Explore the full facilities map.

    08Red flags

    None of these is automatically disqualifying on its own. Each one warrants a direct question, and a partner who cannot answer it comfortably has told you something useful.

    • Reluctance to share inspection history

      Form 483 observations, responses, and EU GMP certificates are ordinary diligence requests. Hesitation, heavy redaction, or "our policy is not to share that" is the strongest single signal on this list.

    • Capability confirmed only at corporate level

      The answer names the group, not the site, suite, or equipment train that would run your product. Ask which building. If that cannot be answered specifically, the capability may not exist where you need it.

    • No named technical lead

      Business development is responsive and technical contact is scarce. The relationship you actually depend on is with the process and quality people; if you cannot meet them before signing, you are buying a sales process.

    • Every answer is yes

      A partner who claims to do every modality, every scale, and every dosage form, on your timeline, with no trade-offs, is either not listening or not being straight. Good CDMOs decline work that does not fit them.

    • Vague or absent capacity answers

      Unwillingness to state what is booked in the relevant suite, or a slot that keeps moving between conversations, usually means you are being fitted around a larger customer.

    • Quality agreement deferred

      "We will sort the quality agreement after signature" concedes the terms you most need. Deviation ownership, notification, and audit rights get harder to negotiate once the program is committed.

    • Recent unexplained change

      A change of ownership, a site closure, a leadership exit in the quality unit, or a major expansion — not disqualifying, but each changes the risk profile and should be discussed openly rather than discovered.

    Where you see ranked lists or Evidence Scores on CDMO Hub — best-of lists, country pages, and company intelligence profiles — the ranking is computed from the same public regulatory record covered in section 03, never from vendor submissions. Read the full methodology for every input, weight, and eligibility gate.

    CDMO selection checklist

    0 / 11

    Tick what a candidate supplier has confirmed in writing.

    Future: send this checklist as an RFI to shortlisted CDMOs directly from your dashboard.

    Frequently Asked Questions

    What is the difference between a CDMO and a CMO?+
    A CMO (contract manufacturing organization) manufactures to a process you provide. A CDMO (contract development and manufacturing organization) also develops the process — formulation, analytical methods, scale-up — and then manufactures it. If your process is not yet defined or not yet robust at scale, you need development capability, which is what the extra D denotes. Many organizations market themselves as CDMOs while their real strength is on one side of that line, so probe which one it is for your specific dosage form.
    How long does CDMO selection take?+
    For a straightforward clinical program with a well-defined process, a disciplined selection typically runs about two to three months from specification to signed agreement: shortlisting, technical exchange under CDA, proposals, a quality audit, and contract negotiation. Commercial programs, novel modalities, and anything requiring a facility audit with remediation take longer. The step most often underestimated is the quality agreement. Selection time is also not the constraint that usually matters — capacity lead time is, so start early enough that the slot you want still exists.
    Should I choose a large CDMO or a smaller one?+
    It depends on where you sit in their customer mix rather than on absolute size. Large networks offer multi-site redundancy, a deep regulatory track record, and a clear commercial scale path, but a small program can be deprioritized when a larger customer needs the suite. Smaller specialists often give more senior attention and flexibility, with concentration risk and a lower scale ceiling. The practical test is to ask where your program would rank among their accounts, and what happens to your slot when a bigger customer has an urgent need.
    What does an FDA inspection classification actually mean?+
    The FDA classifies each inspection outcome as NAI (No Action Indicated — nothing objectionable found), VAI (Voluntary Action Indicated — objectionable conditions found, but no action recommended), or OAI (Official Action Indicated — regulatory or administrative action is warranted). VAI is common across the industry and is not by itself disqualifying. OAI at a site you intend to use is material and requires evidence of remediation. Read the pattern across inspections rather than a single result: repeat observations in the same system, especially data integrity or aseptic practice, matter more than an isolated finding.
    How many CDMOs should I put on a shortlist?+
    Three to five for a technical exchange, narrowing to two for audit and proposal. Fewer than three removes your negotiating position and your fallback if diligence disqualifies one. More than five dilutes the depth of evaluation each receives and consumes internal time that is better spent auditing properly. If your requirement is unusual enough that fewer than three qualified candidates exist, that is important information about your timeline and leverage — and a reason to engage earlier.
    Do I need to audit the facility in person?+
    For any GMP manufacturing that will support a filing or a commercial product, yes. Remote or paper audits are reasonable for early screening and for low-risk services, but they cannot assess facility flow, segregation, housekeeping, or how staff actually behave under normal operation. If travel is impossible, use a qualified third-party auditor and read the full report rather than the summary. Also secure for-cause audit rights in the quality agreement so you can return when something goes wrong.
    What is the most common reason a CDMO relationship fails?+
    Tech transfer that does not reproduce the process in the receiving site — most often through analytical method transfer or an equipment difference that was not treated as a variable. The second most common is a mismatch discovered late between the partner’s real capacity and the program’s scale-up needs. Both are visible during selection if capability and capacity are tested separately, transfer risk is assessed as its own criterion, and the scale ceiling is validated with evidence rather than assumed.

    Go deeper by modality

    • Peptide Manufacturing guide
    • Monoclonal Antibody (mAb) Manufacturing guide
    • Cell & Gene Therapy Manufacturing guide
    • Browse all CDMOs
    • Manufacturing facilities map
    • CDMOs by certification

    Ready to shortlist?

    Our AI Matchmaker analyzes your project requirements and matches you with qualified manufacturing partners from the CDMO Hub directory. Free for buyers.

    Try AI MatchmakerCreate Free Account
    CDMOHUB

    Connecting biotech and pharma teams with verified manufacturing partners worldwide.

    Platform

    • Find CDMOs
    • CDMO Matchmaking
    • Expert Matchmaking
    • Modality Guides
    • Supplier Maps
    • Join as Partner

    Company

    • Blog
    • Briefings
    • About Us
    • Contact

    Legal

    • Terms of Service
    • Privacy Policy

    © 2026 CDMOHUB. All rights reserved.

    CDMO Hub, Inc. Registered in Delaware, USA