How to Organise RFQ Files Before Estimating

If you need to organise RFQ files before estimating, the fastest reliable method is to keep the original package untouched, rename only your working copies, sort files by type and purpose, verify revisions, quarantine duplicates and superseded versions, and hand over one clean estimator pack with an issues log. Consistent naming, metadata, revision control, and audit trails are established good practice in records and construction information management, even if your team is not formally working to ISO 19650.

Poor file organisation costs estimators time twice: hunting for the right documents, then defending assumptions that could have been resolved during intake. Revision control is the bigger commercial risk — UK BIM Framework guidance based on ISO 19650 states that revisions and versions should be tracked, status should be assigned through metadata, and associated exports such as PDFs should be named differently from native source files, with an audit trail when updates occur. If the PDF drawing and the native model are both in the package, they should never be treated as interchangeable without checking revision and source.

Start by creating a received snapshot. Save the email, ZIP, portal download, or shared-drive export into an Originals folder and do not rename those files. The National Archives notes that original file titles and born-digital metadata become part of the record. A strong rule is: rename only the working copy, not the original. Use a consistent naming pattern such as [RFQ code]_[discipline]_[sheet]_[revision].[ext] — for example RFQ1843_STR_GA_RevB.pdf.

A practical folder structure separates originals (00_Received_Originals), current PDF drawings (02_Drawings_PDF_Current), superseded PDFs (03_Drawings_PDF_Superseded), native CAD files (04_CAD_Native), exchange CAD formats such as STEP and DXF (05_CAD_Exchange), site images (06_Images_Photos), scope documents (07_Specs_Scope), clarifications (08_Clarifications), and the final estimator handoff pack (09_Estimator_Handoff). Records guidance recommends unique, human-readable names, dates in YYYYMMDD format, and filenames that still make sense when separated from their folder path.

For teams on Windows or SharePoint, there are practical platform limits. Microsoft documents reserved filename characters such as : * ? " < > / | and reserved names like CON and NUL. In Microsoft 365, the decoded path limit is 400 characters and each path segment is 255 characters. Keep names descriptive but short enough to survive cloud sync.

Before the estimator measures anything, do a revision pass. Compare title blocks, issue dates, revision letters, transmittal notes, and filename cues, then move superseded documents into a clearly labelled folder instead of deleting them. Bluebeam Sets lets users navigate multiple files as one collection including revisions and addenda, Batch Slip Sheet can insert revised pages across documents, and Compare Documents highlights differences through the Markups list.

Duplicates are not always harmless. Microsoft notes that duplicate filenames can cause sync conflicts. In RFQ work, two files with the same name may not be the same version. Compare filename, file size, modified timestamp, title-block revision, and visible content. If two PDFs claim to be the same sheet but differ, compare them and escalate rather than guessing.

For CAD files, ask one question: is this a complete package or just a piece of one? Autodesk eTransmit packages drawings with referenced support files such as CTB plot styles, PC3 configurations, and external references. SOLIDWORKS Pack and Go gathers related files, drawings, and dependencies into a folder or ZIP. If a native package is incomplete, ask for an eTransmit or Pack and Go export rather than reconstructing it manually.

For unsupported files, Autodesk Viewer supports DWG, DXF, DWF, STEP, IGES, SolidWorks, CATIA, and Parasolid formats. If your team cannot open the native file, request a neutral format plus a matching PDF drawing set and record that request in the clarification log. Scanned PDFs are image-only until OCR is applied — Adobe notes that OCR output should be reviewed for accuracy and completeness, making it useful for search but not a substitute for human drawing review.

When missing or ambiguous files surface during intake, a simple escalation matrix keeps decisions clear. A PDF drawing with no clear revision but an email saying latest is high risk — hold as provisional and request confirmation. Conflicting documents where the GA says 10 mm plate and the detail says 12 mm is also high risk — log the conflict and request clarification before take-off.

Duplicate filenames with different content are high risk — compare visually and move one to superseded. Missing CAD references where a DWG opens with unresolved Xrefs is medium to high risk — request an eTransmit. Unsupported proprietary CAD that cannot be opened is medium risk — request STEP or IGES plus PDF drawings. Low-legibility scans are medium risk — OCR for search but estimate with a stated assumption. Minor admin mismatches such as a filename reading RevA while the title block says RevB are low risk — proceed with a note after verifying content alignment.

A proper handoff is not here is the folder. It is a compact, decision-ready package. The estimator should receive the current drawing set, current scope documents, usable CAD or exchange files, a short scope summary, a list of open clarifications, and a clear note on what is excluded, provisional, or superseded. ISO 19650 guidance frames information delivery around giving the right information to the right destination at the right time for a defined purpose.

AI and workflow tools can help with the early mechanical part of intake: extracting text from scanned PDFs, grouping files by type, identifying likely duplicates, spotting missing metadata fields, or drafting a scope summary for review. The line to hold is the commercial one. ACSC guidance says organisations should understand the limits of AI systems including hallucinations and false positives, and should have qualified staff overseeing secure setup and use. NIST's Generative AI Profile recommends defined oversight roles and review of sources and citations. AI can draft and sort, but it should not be the final judge of scope, revision precedence, or pricing basis.

There is also a data trade-off. ACSC notes that third-party AI systems may process data across regions and may use inputs to retrain the model unless controls are in place. If RFQ packages contain client IP, defence-related drawings, pricing logic, or sensitive site details, that risk assessment should happen before a public AI workflow is switched on.

Should I rename client files or leave them as received? Leave originals untouched in an Originals folder, then rename only your working copies. That preserves provenance while letting your team apply a consistent system.

What is the best way to organise PDF drawings for estimating? Keep current PDFs separate from superseded ones, and review revisions before measurement starts. Bluebeam Sets, Slip Sheet, and Compare Documents are built to manage revisions without flattening history.

What should I do if the RFQ includes CAD files I cannot open? Ask for a neutral exchange format such as STEP or IGES plus matching PDF drawings, and record that request in the clarification log.

How do I deal with duplicate drawings with the same filename? Treat them as a revision risk until proven otherwise. Compare content, title-block revision, source timestamp, and file properties, then move the non-current version into superseded.

Are scanned PDF drawings good enough for quoting? Only with care. Adobe says scanned PDFs are image-only until OCR is applied, and OCR output should be reviewed for accuracy. Bluebeam Compare Documents suits scanned files better than vector overlay methods.

Can AI organise RFQ files automatically? It can help with classification and drafting, but should not be the final authority on scope, revision, or assumptions. ACSC and NIST guidance both recommend human oversight, validated outputs, and privacy controls.