The Vendor That Passed UAT and Then Fell Over
You are PM for a Phase 2 study using an ePRO vendor selected 7 months ago. Contract value $410K, fixed-fee, with a go-live milestone 5 weeks out. The vendor passed user acceptance testing on schedule. In the last 3 weeks: (1) the vendor's project manager resigned and was replaced by a junior coordinator; (2) two change requests you submitted (a new diary reminder logic and an added site-level report) have gone unanswered for 12 days; (3) your data manager reports that test data from UAT is still sitting in the production environment; (4) the vendor missed the last two weekly status calls without notice. Your study's first patient first visit is 6 weeks out and the ePRO system must be live 1 week before that. The contract has a 30-day cure period for material breach and no explicit service-level credits.
1. Assess the vendor's health and decide what you do in the next 10 business days.
Read the worked answer
Assessment: this is a leading-indicator failure, not a milestone failure. The go-live milestone is still nominally on track, which is exactly why it's dangerous - every signal here is a predictor of a late, poor-quality delivery, and none of them have yet hit a date you can point to.
Score the signals:
- PM resignation + junior replacement - loss of institutional knowledge at the worst time; the new coordinator likely cannot make decisions, which explains the silence.
- 12 days of unanswered change requests - a decision-making vacuum, not a scheduling problem. Unanswered CRs also mean scope is drifting without a change order, which is a commercial risk to you later.
- UAT data in production - a data integrity and validation failure. In a regulated system this is a finding: production must be segregated from test, and the validation evidence for go-live is now questionable. This is the most serious item.
- Two missed status calls without notice - communication breakdown; the vendor is either overloaded or de-prioritizing you.
Next 10 business days:
- Day 1-2: Written notice to the vendor's account lead (not the junior coordinator) requesting a live call within 48 hours. Agenda is fixed: named replacement PM with relevant experience, written status of both CRs with dates, a remediation plan and dated evidence for the production-environment issue, and a revised go-live plan. Put it in writing; email is fine but copy your contracts/procurement contact.
- Day 2-3: Ask your data manager for a written description of the production data issue and its validation impact, so you can characterize severity rather than relay a rumor.
- Day 3-5: Hold the call. Do not accept "we're working on it." Require dated commitments and a named escalation contact. Confirm whether the CRs are in or out of the fixed fee - silence on CRs often becomes a disputed invoice later.
- Day 5-10: Stand up a contingency in parallel, not instead. Identify whether the protocol can tolerate a paper/alternate ePRO fallback for the first weeks, and get a rough quote and lead time from a second vendor so you know your real options and their cost. Do not tell the vendor you are shopping yet; you want the information, not the confrontation.
- Document every missed call and unanswered CR with dates. You are building the record you would need if you invoke the cure period.
On the contract: you have a 30-day cure period for material breach and no SLA credits. That means your leverage is the cure notice itself and the remaining payments, not a penalty clause. Don't fire the notice on day 3 - a cure notice starts a clock you can't easily stop, and 30 days from now is past your go-live. Hold it as the escalation of last resort if the day-5 call produces no dated plan.
2. Your medical monitor says the production-data issue is "the vendor's problem, not ours." Do you accept that framing?
Read the worked answer
No. The vendor owns the fix; you own the consequence. Those are different things and conflating them is how PMs get blindsided.
The production-environment contamination is the vendor's defect to remediate. But the downstream questions are yours:
- Is the system validated for use? If test data sat in production, the evidence supporting the validation is compromised. You cannot go live on a system whose validation you can't defend, regardless of who caused it. Ask the vendor for a written re-validation plan and the evidence that production is clean before go-live.
- Is any real subject data affected? If production has been used for anything live - even a pilot site - you need to know whether subject data is mixed with test data. That is a potential data integrity issue reportable under your quality system, and it is squarely the sponsor's responsibility to assess, not the vendor's.
- What does the site see? If sites can log in and see test records, they may act on them. That is a safety-adjacent issue.
So the correct response to the medical monitor is: "Agreed the vendor fixes it. I need you to tell me whether the contamination affects data integrity for any enrolled or screened subject, and I need the vendor's re-validation evidence before I'll sign off go-live." Route it through your quality/QA function so the assessment is documented rather than a hallway opinion.
The general principle: in a vendor relationship, the contract allocates responsibility; it does not allocate risk. The risk of a vendor failure lands on the study - on timelines, on data quality, on subjects. Your job as PM is to keep the risk on your radar even when the responsibility sits with someone else.