> ## Documentation Index > Fetch the complete documentation index at: https://docs.trubuild.io/llms.txt > Use this file to discover all available pages before exploring further. # Evaluation governance > Access levels, human sign-off, and the audit record behind every TruBuild evaluation and award. TruBuild is designed for defensible tender evaluation. The AI does not make the decision. It gives your evaluation team a faster first pass, a cleaner review surface, and a complete record of what people accepted, changed, queried, and awarded. This matters in construction procurement because the award recommendation is usually challenged: by an unsuccessful bidder, an owner-side auditor, a funder, or your own committee six months later. Governance is what lets you answer "why this contractor, at this price, on this evidence" without reconstructing spreadsheets and email threads. Use this page as the baseline governance model for enterprise tender teams. ## Governance principles | Principle | How to apply it in TruBuild | | ------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Least-privilege access** | Give evaluators the lowest access level that lets them do their job: Viewer for review-only stakeholders, Commercial for cost teams, Technical for scoring evaluators, Full for the package lead who signs off. | | **Fix the evaluation basis before scoring** | Confirm contractors, weights, reference documents, and criteria before anyone runs a commercial or technical round. | | **Treat AI output as draft evidence** | Anomaly flags, extracted totals, technical scores, and query drafts are inputs to human review, not conclusions. Assigned evaluators review them before committee use. | | **Preserve each round** | Run a new commercial or technical round after every material clarification cycle instead of overwriting prior results. Every round keeps its own timestamps and status. | | **Record human judgment** | Use overrides, notes, and award comments to explain why the team accepted or changed the machine output. Overrides persist in the round record. | ## Access levels Every package member has one of four access levels. The level controls what they can create, change, and approve. | Access level | Can do | Cannot do | | -------------- | ----------------------------------------------------------------------------------------------------------- | ------------------------------------------------------ | | **Full** | Create rounds, upload files, edit and delete overrides, export, award the package | Nothing restricted | | **Commercial** | Create commercial rounds, upload commercial files, edit commercial data, export | Award the package, open the technical evaluation area | | **Technical** | View and work in the technical evaluation area: proposal review, criteria scoring, technical clarifications | Award the package, open the commercial evaluation area | | **Viewer** | View all results in both areas, download exports | Create rounds, upload files, edit overrides, award | The commercial and technical separation is strict in both directions: Commercial access cannot open technical evaluation screens, and Technical access cannot open commercial ones. On owner-led tenders this keeps pricing visibility from biasing technical scoring and vice versa. TruBuild enforces this in the interface, not just in policy. Restricted actions such as **Run New Round**, **Create and run evaluation**, and **Award package** appear disabled for users without the required level, with a **Read-only access** tooltip explaining why. Access levels come from your role assignments: a package role overrides a project role, and organization owners and admins always have full access. See [Roles and permissions](/get-started/roles-permissions) for how to map procurement, commercial, and technical teams onto these levels, including the probity-driven separation of commercial and technical teams on owner-led tenders. Owner-side reviewers, lenders' monitors, and internal audit usually only need Viewer access. They can open every dashboard, ranking table, and scorecard, and download the comparison workbook, without any risk of changing the evaluation record. ### Who can trigger what | Action | Button | Required access | | ----------------------------------- | -------------------------------------------------- | --------------------------- | | Start a commercial round | **Run New Round** or **Create and run evaluation** | Full or Commercial | | Export the comparison workbook | **Export Comparison** | Full, Commercial, or Viewer | | Edit or delete a line-item override | Pencil or trash icon on the row | Full or Commercial | | Award the package | **Award package** | Full only | Technical evaluation actions (proposal review, criteria scoring, and technical clarifications) require Full or Technical access. See [Technical evaluation](/platform/technical-eval) for that workflow. ## Round integrity A round is a controlled snapshot of one workflow (commercial or technical) inside a package. Commercial and technical rounds are created independently but share the same package context: contractors, documents, weights, tender queries, and the award record. Each round records its creation time, its last update time, and every override applied to it. Each round moves through four statuses. The status tells you whether the results in front of you are safe to review. | Status | Meaning | Governance note | | ------------- | ----------------------------------------- | -------------------------------------------------------------------------- | | **Setup** | Round created, evaluation not yet started | Nothing to review yet | | **Analyzing** | Extraction and comparison in progress | Do not brief the committee from partial results; wait for Ready | | **Ready** | Results available for review | Human review starts here | | **Failed** | The run did not complete | An error message and retry option appear; the failed attempt stays visible | Switch between rounds with the **Select round** dropdown on the commercial pages. Prior rounds stay intact when you run a new one, so the difference between initial pricing and post-clarification pricing remains auditable. For round naming, triggers, and the full multi-round sequence, see [Rounds and clarifications](/platform/rounds). ## Human review record Governance depends on separating what the bidder submitted from what your team judged. TruBuild keeps both visible. ### Received vs normalized The vendor comparison table has a bid mode toggle: * **Received** shows the original bid exactly as extracted from the contractor's submission. * **Normalized** shows the comparison after your team's overrides are applied. Always confirm which mode a screenshot, export, or committee figure came from. A commercial position based on normalized figures is only defensible if the overrides behind it are explained. ### Overrides and evidence On any line item in the vendor comparison, evaluators with Full or Commercial access can: * Click the pencil icon to adjust a vendor amount or rename an item, with the change persisted in the round record. * Click the trash icon to remove an override and revert to the received value. * Add notes or comments explaining the judgment. * Open the source document link to see the page in the contractor's submission that the value came from. Values TruBuild filled in where a bidder left a gap are dimmed and marked **Imputed**, so evaluators never mistake an assumption for a submitted price. Exclusions and unpriced items carry badges and flow into the submission quality scoring, which keeps scope gaps visible rather than buried in a spreadsheet footnote. An unexplained override is the weakest point in an evaluation record. If your team adjusts a rate or amount, add a note stating the reason: arithmetic correction, clarified rate from a PTC response, or agreed normalization. Auditors read overrides first. ## Commercial controls Commercial evaluation should answer one question: **which bidder is commercially most advantageous after exclusions, unpriced items, anomalies, and clarifications are understood?** The **Pricing Integrity** and **Submission Quality** scorecards grade each bidder from A to F. Treat any low grade as a prompt for review and clarification, not a conclusion: a bidder graded low on Submission Quality may simply have priced against a different BOQ revision, and that is exactly the kind of finding the record should show a human investigated. | Control | Recommended owner | Evidence in TruBuild | | ------------------------------------ | -------------------- | ------------------------------------------------------------------------------ | | Reference BOQ accepted | Commercial lead | BOQ upload marked complete in round setup | | Vendor files complete | Commercial team | Vendor file count per contractor in round setup | | Comparison round created | Commercial lead | Round name, round number, status, timestamps | | Pricing anomalies reviewed | Commercial evaluator | Pricing Integrity scorecard, overpriced and underpriced division detail, notes | | Unpriced and excluded items reviewed | Commercial evaluator | Submission Quality scorecard, exclusion and unpriced badges, PTC questions | | Final commercial position approved | Procurement manager | Exported comparison workbook and award comments | ### Commercial review checklist * The contractor list is confirmed and locked. * Your organization's default currency matches the tender currency; an admin sets it in organization settings. * The reference BOQ is uploaded (and the PTE, if you want a benchmark row and deviation analysis). * Each contractor has the expected priced submission files. At least one vendor must have files before the round can run. * Globally unpriced contractors (marked **Unpriced** in the bidder ranking) are investigated before they distort the comparison. * Divisions flagged as overpriced or underpriced in the Pricing Integrity card are reviewed first. * Exclusions and unpriced items in the Submission Quality card are turned into commercial PTCs. * The bid mode is checked: know whether you are reviewing **Received** or **Normalized** figures. * Material overrides carry notes explaining the change. ## Technical controls Technical evaluation should answer one question: **which bidder best satisfies the scope, criteria, and owner requirements at the agreed technical weighting?** | Control | Recommended owner | Evidence in TruBuild | | ------------------------------- | ------------------- | ------------------------------------------------------------- | | Criteria accepted | Technical lead | Reviewed criteria tree | | Contractor proposals complete | Technical team | Uploaded proposal documents by contractor | | Initial technical round created | Technical lead | Round kind, round number, round name | | AI scores reviewed | Assigned evaluators | Accepted scores, overrides, reviewer comments | | Clarification impact assessed | Technical lead | Rerun or carry-forward selections by contractor and criterion | | Score review completed | Technical lead | Technical review marked complete on the latest round | Completing the technical score review is not optional paperwork. TruBuild blocks the award until the review of the latest technical round is complete, which guarantees a human signed off on the scores that support the recommendation. ## Tender query controls Pre-tender and post-tender queries are not administrative noise. They are part of the award evidence. Each query carries its own status, owner, designated approver, due date, and audit history, so a committee can trace who drafted a response, who approved it, and when it went to the tenderer. Three rules keep the register defensible: * **Nothing goes out unapproved.** Responses move through **Pending Approval** before they can be sent, so a designated approver signs off every answer that reaches a bidder. * **Categories are owned.** Use the Technical, Commercial, Contractual, and General categories consistently, so the commercial and technical leads each run an accountable queue. * **Closures are explained.** A closed query should carry a meaningful close reason (Resolved, Withdrawn, Duplicate, Merged, or Other), because an auditor reads an unexplained closure as an unanswered question. See [Tender queries](/platform/tender-queries) for the full status model, approval workflow, and reporting. ## Award sign-off The award is the single most audited action in the process, so TruBuild gates it hardest: only Full access can award, and the technical review must be complete first. On the award page you select an eligible contractor (only contractors ranked in both the technical and commercial evaluations qualify), review the technical and commercial summary for that contractor, and record award comments. Treat the comments as mandatory in practice: they are the committee's justification and the auditor's first stop. If the technical review on the latest round is incomplete, the **Award package** button stays disabled until it is done. The step-by-step award walkthrough, with examples of strong and weak award comments, is in [Award readiness](/platform/award-readiness). ## Audit trail TruBuild records governance-relevant actions automatically, with the acting user and a timestamp. Organization administrators can review the activity log in the **Activity** section under **Settings**, and each evaluation round keeps its own record of overrides and exports. | Action | What is recorded | | --------------------------- | ----------------------------------------------------------------- | | Commercial round created | Round number, round name, asset, status | | Technical PTC round created | Round type and the clarification source used | | Package awarded | Awarded contractor, award comments, timestamp | | Files uploaded | Upload category (BOQ, PTE, vendor files) per event | | Overrides applied | Persisted in the evaluation round record itself | | Exports generated | Exporting user, timestamp, and scope (vendors, rounds, divisions) | Two properties make this record useful in a dispute: * **It is passive.** Evaluators do not have to remember to log anything. If a round exists, its creation is recorded; if an export left the system, its scope is recorded. * **It is scoped.** Export logging captures which vendors, rounds, and divisions a workbook contained, so you can establish exactly what a committee or external reviewer saw and when. Export logging is the piece teams most often overlook. When a comparison workbook is circulated outside TruBuild, the export record is your evidence of what version left the system. Retain the exported `.xlsx` files alongside the committee pack. ## Evidence exports Each export is a fixed snapshot for the tender file. The filename embeds the asset or package name and the round or date, so a workbook found on a shared drive months later still identifies which evaluation state it came from. | Deliverable | Format | Filename pattern | Governance use | | ---------------------------- | ------ | -------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------- | | Commercial comparison | Excel | `_Commercial_Comparison_.xlsx` | The commercial position for one round, with summary metrics, division-level detail, and a warnings sheet where anomalies were detected | | Technical evaluation results | Excel | `_TechnicalEvaluation_Results.xlsx` | Contractor rankings and per-round scores with evidence references | | PTC questions | Excel | `_PTC_Questions_.xlsx` | Clarification questions, vendor responses, and the status of each question, proving the clarification loop was closed | | Evaluation report | Word | `_EvaluationReport_.docx` | Committee-facing narrative: executive summary, rankings, detailed scores, appendices | When you export the commercial comparison, you choose which vendors to include, rate or amount display, whether to include the PTE, and whether to export **Received** or **Normalized** figures. State those choices in the committee pack; the audit trail logs the same scope against the export event, so the pack and the log corroborate each other. Export before every committee session, not after. A committee that reviews a live screen has no fixed record of what it saw; a committee that reviews a named workbook does. ## Award readiness Before a package is awarded, the procurement manager should be able to answer: * Which contractor is selected, and is it ranked in both evaluations? * Which technical round and commercial round support the recommendation, and are both marked **Ready** and reviewed? * Which material PTCs and tender queries remain open? * Which figures were manually overridden, and does each override carry a note? * Is the committee figure based on **Received** or **Normalized** bids? * Why did the team choose the selected contractor over the nearest alternatives? See [Award readiness](/platform/award-readiness) for the final committee checklist. ## Related Commercial workflow details for cost managers and QS teams. Technical workflow details for evaluators and discipline leads. Round naming, triggers, and multi-round sequencing. The committee-ready award record and residual risk register. The controlled query register: statuses, approvers, and reporting. Organization, project, and package roles behind the access levels.