PMP 2026 | 93 | Ethics: Contracts, Suppliers, and Data
Integrity is one of those words that everyone uses, but in project management it has a very specific meaning. Integrity means consistency of actions, words, and records. What you promise must match what you do, and what you do must match what you record. Nowhere is this more important than when contracts, vendors, and data are involved. These elements cross organizational boundaries, where trust can either flourish or collapse. Contracts bind two or more organizations to a set of terms. Vendors often act as extensions of your team but operate with their own incentives. Data, meanwhile, is shared across systems and people who may not all understand the risks. In each of these, integrity is what prevents small lapses from escalating into costly disputes.
Welcome to Certified, the P M P prepcast. This is episode ninety three of season one. Today we are exploring integrity in contracts, supplier relationships, and the use of data.
Clear commitments and accurate performance evidence can reduce uncertainty between organizations, but integrity does not guarantee lower bids or less oversight. The project manager should coordinate truthful information, appropriate authority and fair application of the actual agreement. A trusted relationship still needs usable evidence and defined responsibilities. The aim is to make performance and decisions understandable, not to replace verification with goodwill.
The same principle applies to data. Data controls, privacy policies, and retention rules all exist because information flows between organizations. If your team cuts corners, by creating shadow copies of production databases, for example, you may save a day of testing but risk a breach that damages both companies. Here, integrity means saying no to quick wins that compromise rules. It also means documenting access consistently so audits show the same story as the actual behavior. As steward of both contracts and data, the project manager is not merely a coordinator but the guardian of consistency.
Define deliverables and acceptance conditions clearly enough for the parties to understand what is being purchased and how it will be evaluated. When a request changes a contractual commitment, use the agreement's authorized change mechanism. Work already within delegated contractual discretion may follow a different route. The project manager coordinates the evidence and the appropriate decision owner rather than inventing authority. Acceptance should reflect the actual agreed conditions and supporting evidence. A confident statement that it looks fine does not establish a condition that remains unverified.
An informal understanding can create ambiguity about whether work was requested, authorized or accepted. Use the agreement's actual mechanisms and retain appropriate evidence of the decision. Clear terms and records can help resolve a dispute, but cannot guarantee one will never arise. When records conflict with the delivered result, investigate and report the gap rather than assuming the tidier account must be correct.
Choose and manage contract incentives with attention to the behavior they may encourage. Time-based, unit-based and outcome-based arrangements can each be appropriate in a suitable context, and each can create problems when poorly designed. Consider uncertainty, measurability, quality and how much control each party has over the outcome. Identify a perverse incentive with evidence rather than assuming that paying for hours is inherently unethical. Fee adjustments and changes to terms belong to the authorized contract process.
When incentives are misaligned, integrity demands that you raise the issue. If you notice a vendor emphasizing volume over quality because of payment terms, you cannot ignore it. Responsibility means you disclose the risk and recommend corrective adjustments. Respect means you do so transparently, without accusation, and by referencing contract clauses. Documenting these adjustments in the contract file ensures that any fee credits or incentive changes are visible for audits. By aligning incentives to outcomes, not shortcuts, you demonstrate integrity and protect long-term value.
Explore contract integrity through an original fictional service-desk transition. A supplier is paid partly against an agreed measure of resolved support cases. During the pilot, it reports fifty closures in a week. Twenty of those cases reopen because users cannot complete the original task. The supplier says the target was achieved because its system recorded fifty closure events. The buyer says the work was not resolved. Before accusing either party, examine the contractual definition, the measurement method and the underlying case history.
Twenty reopened cases out of fifty reported closures is forty percent in this small example. That result can indicate a problem worth investigating, but it does not by itself establish fraud or a fee adjustment. Some reopened cases might concern new requests, while others may demonstrate that the original condition was never satisfied. The agreement's definition and actual evidence matter. A project manager should not invent a penalty from an unfavorable dashboard or accept the dashboard as conclusive proof of success.
Trace several cases from initial request through attempted resolution, user confirmation and reopening. Determine whether the reporting method counts events, distinct cases or genuinely verified outcomes. If it counts repeated closure events for the same case, a high total may conceal unresolved work. Explain that limitation openly. Changing the metric prospectively may require an authorized contractual decision, while correcting an inaccurate report under the existing definition is a different action.
Suppose the original agreement explicitly requires verified resolution and the evidence shows that the twenty cases were prematurely closed. Follow the agreed correction and commercial process through the authorized owner. Suppose instead that the agreement poorly defines closure and both parties used different reasonable interpretations. Clarify the ambiguity and pursue an appropriate authorized resolution rather than rewriting the past definition unilaterally. Ethical conduct requires truthful evidence and fair handling even when the contract itself is imperfect.
A better future measure should consider what each party can actually control. A supplier may resolve its part of a problem while waiting for a buyer-controlled access decision. Penalizing it for every delay regardless of responsibility could create another unfair incentive. Conversely, excluding all waiting cases could encourage the supplier to classify difficult work as blocked. Define the boundaries, evidence and exception treatment through the proper process. No metric is ethical merely because its title contains the word outcome.
The final performance account should distinguish the original claim, the verified findings, any corrected report and any authorized change to future measurement. Preserve the case evidence according to the agreement and information controls. Do not quietly delete the reopened cases to improve a score. Do not circulate identifiable user details more broadly than needed to explain performance. Contract integrity and data integrity must be maintained together.
Data integrity also matters when artificial intelligence supports project work. Use information and tools authorized for the purpose, limit access appropriately and verify important outputs against reliable records. A plausible generated summary is not evidence that a test passed or a supplier met a requirement. Protect restricted information in prompts, uploads and retained outputs. Sanitization and redaction require verification; their labels do not guarantee that data is safe to share. The accountable person remains responsible for the resulting report or decision.
Consider a fictional supplier-assessment report drafted with an approved artificial-intelligence tool. The project analyst uploads only information authorized for that use and asks for a summary of the evidence. The generated draft says that all recovery tests passed, but the source package contains only a test plan and a statement that execution is scheduled. The wording is plausible and confident. It is still unsupported because an intended test is not a completed test result.
The reviewer should compare each material assertion with the actual source record. For the recovery claim, locate the result, date, configuration, scope and responsible approval if those are required by the assessment. When the evidence is absent, say that the result has not been established from the supplied material. Do not soften the unsupported statement to probably passed and leave it in a decision report. Confidence in wording cannot fill an evidence gap.
The tool may still be useful for organizing verified material, identifying questions or drafting a clearly labeled summary. The accountable reviewer decides whether the output is accurate enough for the intended purpose. Keep the sources and material corrections traceable under the organization's process. An approved tool does not make every input appropriate or every output correct. Approval for one purpose also does not automatically authorize using the same service with restricted supplier information for another purpose.
Now add a proprietary supplier appendix that the buyer may inspect but may not redistribute. Access to that appendix does not establish permission to upload it into a separate service. Check the relevant contractual and information rights before using it. An authorized on-site review, a restricted summary or another agreed evidence arrangement may be possible. If the necessary verification cannot be performed within the available rights, report that limitation and seek an acceptable route rather than inventing a reassuring conclusion.
The same discipline applies to an unfavorable generated statement. Suppose the draft claims that the supplier failed a control, but the source only says the control was outside the review scope. Outside scope is not the same as failed. Correct the distinction before the report reaches decision makers and preserve the legitimate scope limitation. Fairness requires challenging unsupported negative claims as well as unsupported positive ones. Automation does not transfer responsibility for either kind of error away from the reviewer.
Suppose a team proposes copying restricted production data to an unapproved spreadsheet for testing. Do not proceed with that unauthorized use. Work with the data, security and testing owners to obtain an approved approach, which may use suitable synthetic data, verified sanitization or a controlled environment. Confirm that the data still supports the intended test and that residual exposure is acceptable under the rules. A formal exception is needed only if the actual policy requires one. Record the authorized use and retention arrangements accurately.
Manage supplier relationships through fair evaluation, relevant disclosure and accurate performance records. Observe the applicable gift and conflict rules rather than relying on hidden favors. Retain the evidence the organization is entitled and required to hold, with appropriate access controls and source identification. Some supplier material may remain proprietary or restricted. Do not assume that every performance file must be copied into a shared repository. The evidence arrangement should support verification while respecting contractual and information rights.
Agree how the parties will access required evidence and preserve its provenance. An authorized reference, inspection record or controlled copy may be appropriate depending on the agreement. Record limitations when material cannot be disclosed directly, and seek an acceptable verification route rather than claiming it has been independently checked. A shared record can reduce misunderstanding, but it does not guarantee that all parties agree or that the source is correct. Review the evidence as well as the storage arrangement.
At its core, integrity across contracts, vendors, and data means that the story told in conversations matches the story told in documents and the story revealed by evidence. If those three diverge, trust collapses. For example, a completion claim alongside missing acceptance evidence requires clarification of what is complete, what remains unverified and whether the statement misleads. A detected data copy requires assessment of its authorization and controls. Report the actual gap and correct any misleading account rather than assuming that every incomplete record or copy proves dishonest conduct. P M I’s ethical standard is that your words, your actions, and your records must align. That consistency is what defines professionalism in complex, cross-boundary projects.
An integrity problem can create contractual, operational or reputational consequences, but the outcome depends on the facts and applicable process. Ethical conduct does not guarantee that a difficult project retains everyone's confidence. Keep the account accurate, respect information rights and address the actual gap. Those responsibilities remain even when the result is unfavorable.
The same principle runs through agreements, incentives and information: an accurate record should describe an authorized action and the evidence for its result. A polished file cannot repair an unsupported commitment or an improper data use. Next we apply that distinction to a supplier change, a testing request and an audit. Notice which owner can decide and what remains to be verified in each situation.
Consider a supplier who describes a late product change as minor and says it has no impact. The assurance may be useful input, but assess whether the evidence and decision route required by the actual agreement have been satisfied. Do not infer approval authority from urgency or from the supplier's confidence. If the necessary verification and authorization are already complete, the next action may instead be coordinated implementation.
Obtain the supplier's required impact evidence and involve the authorized contract owner. Compare the proposed change with the actual terms, acceptance conditions and relevant financial mechanism. A point of total assumption is useful only where the incentive arrangement makes it applicable. Approve, reject or modify the request through the proper authority, then update only the commitments and records affected. Keep forecasts honest while a decision is pending. An integrity review does not assume that every requested change must produce a contract amendment and new baselines.
For the testing scenario, the team wants to move restricted production information into an unapproved spreadsheet. Stop that proposed unauthorized use and consult the relevant data and testing owners about an acceptable alternative. Evaluate approved synthetic or appropriately protected data for fitness as well as privacy. Document any required authorization, exception, access and retention conditions. Do not describe the dataset as safe solely because identifiers were removed. Both the permitted use and the effectiveness of the controls matter.
Verify that the actual test process follows the agreed controls. Access records can support that review, but the absence of an obvious unauthorized copy in a log is not proof that every copy was controlled. Check storage, sharing and retention arrangements as appropriate to the use. Report a limitation or incident through the responsible route. The ethical task is an honest account of the evidence and the action, not a promise that approved-looking records eliminate risk.
Develop the testing-data case through an original fictional warranty application. The team needs to test date calculations, duplicate claims and unusual address formats. A developer proposes using a copy of the live customer database because it contains realistic examples. The proposed environment and recipients are not approved for that information. The immediate project response is to stop the unauthorized proposal and work with the responsible data, security and test owners on a permitted approach, not to assume realism overrides the restrictions.
Begin by specifying what the test actually needs. Date-boundary behavior can often be represented by deliberately constructed cases, while duplicate detection needs controlled relationships between records. A synthetic dataset may provide those features without reproducing real customers, but its suitability must be assessed. Randomly generated rows could miss the precise edge cases that matter. Privacy protection and test adequacy are separate questions that both require an answer.
Suppose a permitted sanitized dataset is proposed instead. Removing names alone may leave account numbers, free-text comments or combinations of attributes that reveal people or restricted commercial information. The qualified owners should verify the transformation and residual exposure under the actual rules. The project manager should not certify the data safe merely because a filename contains sanitized. Record the authorized purpose, recipients, environment and any restrictions that travel with the data.
Also examine the entire workflow, not only the initial transfer. Testers may export failure examples into tickets, attach files to messages or retain local copies for later debugging. Those actions can create new locations and recipients outside the original approved arrangement. Establish permitted handling and retention with the responsible owners and verify that the actual process follows it. A controlled starting location does not automatically control every later copy.
At the end of the test, follow the applicable retention and disposal requirements. Do not delete relevant records indiscriminately if another duty requires preservation, and do not retain sensitive data indefinitely because it might be useful. The appropriate legal, data or records owner resolves conflicting obligations. The project manager coordinates completion and captures the evidence required by the agreed process. An accurate disposition record should say what was done and what remains, rather than simply declaring all data gone without verification.
Cooperate with legitimate audits and investigations through the authorized scope and information controls. Provide accurate relevant evidence and identify missing or restricted material honestly, using approved access or disclosure arrangements where needed. Do not conceal an unfavorable result or alter history to make it appear cleaner. Correct errors through the applicable traceable process and preserve relevant originals where required. An audit can identify improvement needs, but confidentiality and evidence handling remain responsibilities during the review.
An honest account of a gap supports a valid review even when it leads to an unfavorable finding. Do not promise that auditors or regulators will forgive it because a corrective plan exists. Explain what is known, the impact, the response and any evidence still needed. Concealment would add a further integrity problem. The organization needs an accurate basis for action, and the reviewer needs to understand the limits of the material provided.
When comparing answers about contracts or data, inspect the substance of the proposed action. A written amendment can still be unauthorized, and a formal access request can still seek an impermissible use. A verbal emergency decision may be permitted if the actual procedure and recording requirements are met. Backdating and deliberate concealment misrepresent the facts. The useful question is whether the commitment or information use is authorized and accurately documented, rather than whether the option sounds formal.
A practical routine is to define commitments clearly, evaluate performance against the actual conditions and use the appropriate authority for changes. Examine incentives for unintended effects without condemning a contract type by its name. Use data only under the permitted purpose and controls, and verify important conclusions, including artificial-intelligence outputs. Retain or reference supplier evidence within the relevant rights and restrictions. Report material gaps honestly through the proper channels. These actions connect ethical intent with verifiable work.
For this lesson, integrity means that the action, explanation and record are consistent with what actually occurred. That is a working teaching description rather than a quotation from a formal glossary. A clean-looking record does not establish that an action was authorized, and a sincere statement does not substitute for evidence. Test both the action and its account. This keeps the discussion focused on responsible conduct rather than cosmetic paperwork.
Finally, examine a fictional audit request that asks for supplier evidence the project does not hold. The project manager has a summary from an authorized assessor but not the underlying proprietary records. Explain the source, scope and limits of that summary, and arrange any further access through the agreed route. Do not describe the project team's review as independent inspection of records it never saw. The distinction matters even if the assessor is highly credible.
A legitimate reviewer may need additional evidence, and the organization may face an unfavorable finding if the evidence is insufficient. That consequence does not justify altering the description of the review. Work with the authorized owners to provide what can properly be disclosed and identify unresolved gaps. A corrective plan can address future evidence access, but it does not prove that past verification occurred. Honest cooperation includes admitting the limits of the available record.
Across these cases, ask whether the commitment, action, measurement and account refer to the same thing. A closure event may not mean resolution, a test plan may not mean a passed test, and permission to inspect may not mean permission to redistribute. These are practical distinctions rather than abstract moral slogans. They help a project manager connect ethical responsibility to the actual work crossing organizational boundaries.
A final supplier-data variation concerns the end of an engagement. The supplier's staff no longer need access to the project environment, but a continuing warranty obligation remains. Removing operational access and preserving an authorized route for warranty support can both be necessary. Do not leave broad access active merely because the supplier may someday need to help. Equally, do not destroy required evidence or make legitimate support impossible without coordinating the transition.
The responsible access, contract and operating owners should define what ends, what continues and how future access is approved. Verify the actual change rather than relying only on a completed offboarding ticket. The project manager can record that the requested access review was performed and identify any remaining exceptions with owners and dates. This keeps the closure account aligned with reality while respecting the continuing agreement. A contract's end, an account's removal and the end of every obligation are three different events that should not be collapsed into one checkbox.
Use chapters twenty-four and twenty-five for procurement, chapter thirty-three for reliable reporting and chapter thirty-seven for compliance. Episode one hundred four develops artificial-intelligence decisions with the book's case evidence. In each, identify the relevant authority, information rights and verification need. Do not select a formal-looking action without checking those conditions. That is the transferable reasoning this ethics lesson adds to the technical material.
A cross-organizational project needs reliable commitments and an honest account of their fulfillment. Preserve appropriate evidence, respect information rights and disclose material limitations to the responsible recipients. These practices support trust without guaranteeing that a contract succeeds or an audit has no findings. The professional responsibility is to act within authority and make the result understandable. That holds even when the evidence is inconvenient.
For more certification study resources, books, and podcasts, visit Bare Metal Cyber dot com. Thank you for listening to Certified, the P M P prepcast.
