SpecTrace Guide: Compare Document Versions and Review Every Change

Somewhere between version 3 and version 4 of a specification, something changed. Maybe a tolerance moved from 0.5 mm to 0.8 mm on page 41. Maybe a payment term in a contract appendix went from 30 days to 45. Maybe a procedure step was quietly deleted. Whoever sent the new file probably summarised the "main changes" in an email — and the summary is almost never complete.

Reading both versions side by side is the honest fix, and nobody has time for it. A 60-page datasheet against its predecessor is an afternoon of eye-strain, and the changes you miss are, by definition, the ones you did not know to look for.

SpecTrace is the Motifuse product built for exactly this job. You give it two versions of a document, it reads both, and it returns a reviewable list of findings — what was added, what was removed, what was modified — with each finding linked back to where it sits in the document. This article explains what SpecTrace actually does today, how the free and Pro plans differ, how uploaded files are stored and deleted, and where the product's current limits are. Everything below describes shipped behaviour; features that are planned but not yet available are labelled that way, exactly as they are on the product's own pages.

What is SpecTrace?

SpecTrace is a document and specification change-review tool that runs at motifuse.com/spectrace. Its unit of work is the comparison: two versions of the same document in, a structured set of findings out.

It reads seven file types — PDF, DOCX, XLSX, CSV, TXT, PNG, and JPEG — so the same workflow covers a revised requirements document, an updated spreadsheet of parameters, or a photographed page. Image files are stored and validated on any plan, but producing findings from them requires text recognition, which is a Pro capability.

Three properties define how it is built:

• Findings, not a wall of red and green. The output of a comparison is a list of discrete changes you can filter, sort, and work through one at a time, rather than a single marked-up blob.
• Evidence attached. Each finding carries the old text, the new text, and a precise location — the page and region in a PDF, the cell reference like D14 in a spreadsheet, the paragraph in a Word document — so you verify the change against the source instead of trusting the tool.
• Deterministic rules, not a model. The comparison is performed by parsers and explicit rules. No language model is involved, and running the same comparison twice produces the same findings.

SpecTrace sits alongside Reconova (data cleaning) and DocForge (document generation) in the Motifuse product family: three separate tools, each with a free tier and an optional paid tier. Its siblings have their own deep-dive guides — the Reconova data-quality guide and the DocForge document-generation guide — and this article does the same job for SpecTrace.

The problem it solves

Version-to-version review is a real task in a lot of jobs, and the standard tooling for it is weak:

• Word's compare view works only for Word documents, and struggles with complex layouts.
• Generic PDF diff tools show raw differences without structure — good for spotting that something changed, bad for producing a review you can act on.
• Manual review does not scale, and it fails silently: the reviewer who missed a changed tolerance rarely finds out until the consequence arrives.

The cost of a missed change is rarely the document itself. It is the purchase order placed against superseded requirements, the test plan that still checks the old limit, the subcontractor working from last month's revision. A change-review tool exists to make "what changed?" a question with a checkable answer instead of an estimate.

Who is SpecTrace designed for?

SpecTrace fits wherever documents are versioned and the differences carry consequences:

• Engineers and technical teams reviewing revised specifications and supplier datasheets.
• QA and compliance roles who must confirm that a procedure or quality document changed only where the change note says it did.
• Business analysts and product teams tracking requirement documents across iterations.
• Procurement and contract roles comparing a returned contract or pricing schedule against the version that was sent.
• Technical writers and documentation teams verifying what actually changed between releases of a manual.

The ten built-in project templates mirror this list — supplier specification changes, product datasheet revisions, SOP revisions, contract appendix updates, tender requirement revisions, equipment manual updates, pricing schedule changes, quality document revisions, safety instruction revisions, and laboratory report comparisons. A template pre-fills a project's priority, confidentiality, and tags; it never locks you in.

One boundary to know up front: SpecTrace is a single-user tool. Projects belong to one account, and there is no shared review, invitation, or approval workflow today (more on that under limitations).

Key features

Everything in this list is live in the product now.

Just as important is what is not on this list. AI-assisted change explanations, impact assessment, PDF, DOCX, and XLSX report formats, saved templates and filter views, and custom report branding are all marked planned on the pricing page — visible on the roadmap precisely so that nobody buys them by accident.

How SpecTrace works

A project holds one document's life: its versions, its comparisons, and its findings. You upload a baseline version and a revised version, and every upload is validated three ways — declared type, file signature, and the actual bytes after upload — then scanned for malware before it can be used. A file that fails scanning stays quarantined; there is deliberately no code path that lets an unscanned file through.

When you run a comparison, the engine extracts both versions into structured blocks — paragraphs, headings, list items, table cells — and aligns them in passes: identical unique blocks anchor the two versions, spreadsheet cells match by their position rather than their text, and remaining content is matched by similarity, including content that moved. Each aligned pair is then classified: is this a formatting-only edit, a changed number, a changed unit, a strengthened requirement, a removed safety warning? Running headers and footers are excluded by default, so a changed page-number style does not bury the real changes.

The result is a findings list with a summary on top — for example, that of 340 differences, 280 are formatting-only and 60 are meaningful — and a review queue you can filter by importance, change type, page, status, or confidence. Each finding shows the old text, the new text, and its location in both versions. You review, set a status, add notes where needed, and export.

Two details are easy to miss and worth knowing. Comparisons are idempotent: re-running the same pair with the same settings returns the existing result rather than consuming another comparison from your monthly quota, and failed runs do not count against it. And review decisions survive re-runs, because finding identities are derived from the compared content rather than assigned randomly.

Processing happens on Motifuse servers: documents are uploaded over an encrypted connection to private object storage, parsed and compared server-side by a dedicated worker, and results are returned to your session. This is different from most Motifuse free tools, which run entirely in the browser — comparing large documents is server work, and the product says so rather than pretending otherwise.

Step-by-step: your first comparison

1. 1Open SpecTrace and choose Continue without an account (or sign in). Guest work is kept for 7 days in that one browser; signing up later moves it into your account automatically.
2. 2Create a project. Pick one of the ten templates if it matches your document — a supplier spec change, a contract appendix update — or start blank. Set priority and confidentiality if they matter to your workflow.
3. 3Upload the baseline version — the document as it was. Give the version a label you will recognise, like "Rev B".
4. 4Upload the revised version — the document as it is now.
5. 5Run the comparison. It is queued, processed on the server, and returns a findings list with a summary of how many differences are meaningful versus formatting-only.
6. 6Work the review queue. Sort by importance, confirm the findings that matter, reject the noise, mark items that need action, and attach notes where a decision needs context.
7. 7Export the findings as CSV or JSON — reviewer decisions included — for your tracker, your change log, or your records.

For your first run, use a document pair where you already know some of the changes. Watching the findings list catch the ones you knew about — and usually a few you did not — is the fastest way to calibrate how much to trust it.

Real-world use cases

A revised supplier datasheet. A component supplier issues revision C of a datasheet. Upload revision B and revision C, run the comparison, and the changed electrical limits become a checklist to review against your design — including a tolerance change flagged as a tolerance change, not just "text differs".

A returned contract. Legal sends a contract; the counterparty returns "the same document with minor changes." Comparing the two versions turns "minor" from a claim into a list you can verify clause by clause, with changed payment terms and deadlines surfaced as date and price changes.

A procedure update audit. Quality management requires that procedure changes match their change notes. Comparing the released versions produces the evidence: every finding either maps to the change note or needs an explanation — and a requirement quietly downgraded from "shall" to "should" is flagged as a weakened requirement, which is precisely the kind of edit a manual read misses.

A parameter spreadsheet. Two versions of an XLSX of configuration values or test parameters. Cells are matched by position, so every changed value is reported against its cell reference — the kind of difference human review is worst at across hundreds of rows.

Requirements drift between iterations. Product teams comparing requirement-document iterations get an explicit record of scope change — additions and deletions included, which are the changes summaries most often omit.

Free and Pro access

SpecTrace's free tier is genuinely usable, not a teaser: it runs the same comparison engine as Pro, and you can start as a guest without creating an account. The numbers below come from the product's own plan catalogue.

Pro is a monthly subscription billed in Indian rupees; the current price is published on the pricing page, and the platform pricing overview shows how it sits alongside the other Motifuse products. Upgrading requires a free account first — a guest identity lives in one browser and cannot be recovered, so a purchase made as a guest would be lost with the cookie. Cancelling stops renewal at the end of the billing period, and nothing is deleted early.

Business and Enterprise tiers are listed on the pricing page for transparency, but they are request-based conversations, not checkout buttons — there is no self-service path to buy them, and the page says so.

Two things the pricing deliberately avoids: no feature exists only to force an upgrade (the free tier runs the full engine), and no capability is sold before it ships — planned features are labelled planned on every plan.

Security, privacy, and file handling

Documents you compare are often the sensitive kind — contracts, unreleased specifications, internal procedures — so the handling model is worth stating precisely.

• Uploads are quarantined and scanned before use. Every file lands in a quarantine area, is malware-scanned, and only a clean file is moved into working storage. A file that cannot be scanned stays quarantined and unusable — there is no "give up and let it through" path.
• Files are encrypted in transit and at rest. Uploads travel over an encrypted connection to private object storage with server-side encryption applied to every object. Access is through short-lived signed URLs — minutes for an upload, about a minute for a download.
• Retention is time-limited by plan: 14 days on Free, 365 days on Pro, stamped at upload time, after which a retention process deletes the files — bytes first, records second, so nothing orphaned is left behind. Deleting a document version yourself removes its file from storage immediately; deleting a project marks it deleted, and its files are removed by that same retention process.
• Guest sessions are temporary by design. Guest work is tied to a browser cookie (stored only as a cryptographic hash server-side) and is deleted after 7 days, files included, as documented in the cookie policy. Signing up before then moves the work into your account.
• No AI service touches your documents. The comparison is rule-based; no customer-facing AI feature is enabled, and nothing you upload is sent to any AI provider. The one external service involved is Google Cloud Vision, used only for text recognition of PNG/JPEG images on Pro.
• Logs are deliberately content-free. Audit records exclude even filenames — a filename regularly carries a supplier name or contract number — and worker logs contain no extracted text or block contents.
• Exports are guarded. CSV exports neutralise spreadsheet formula injection, and every export carries its own limitations statement, including that low-confidence and OCR-derived findings should be checked against the source before action is taken.
• Workspace surfaces are private and ad-free. Signed-in SpecTrace pages are excluded from indexing and never carry advertising, like every Motifuse workspace.

The security page describes the platform-wide model — authentication, workspace isolation, and abuse limits — and the privacy policy covers data handling in full. Neither page claims certifications the platform does not have, and neither will this article.

Limitations and responsible use

SpecTrace publishes its own boundaries on the product page, under the heading "The boundaries, before you rely on it." They matter for deciding whether it fits your workflow:

• It is a single-user tool. There are no shared projects, invitations, reviewer assignment, or approval chains. If your process needs multi-person sign-off, SpecTrace produces the findings; your process tracks the sign-off.
• It has no public API, no API keys, and no webhooks.
• It does not sign documents, and finalising a review is not a legally binding signature.
• It does not certify compliance with any standard, and it does not replace professional review.
• Scanned PDFs are not read today. Text recognition covers PNG and JPEG images on Pro; a PDF that is only a scan extracts no text, and SpecTrace tells you so rather than producing empty results. If you have a scan, upload the pages as images on Pro.
• Extraction from poor-quality scans is unreliable — and SpecTrace flags the findings it is unsure about rather than hiding the uncertainty. Findings below its confidence threshold are marked, filterable, and called out in exports.
• Reports are data, not documents. Findings export as CSV or structured JSON; formatted PDF, DOCX, and XLSX reports are planned, not available.
• No accuracy percentage is claimed. The product page deliberately publishes no accuracy figure, customer count, or unattributable testimonial — a comparison tool should be verified against its evidence, not trusted on a statistic.

Frequently asked questions

The bottom line

Most tools in this space either handle one format well or produce diffs that only a patient reader can use. SpecTrace's approach is narrower and more honest: seven formats, one workflow, findings with evidence attached, a review process with an audit trail, limits published in the open, and a roadmap that is labelled as a roadmap.

If version-to-version review is part of your work, the fastest way to evaluate it is the product itself: open SpecTrace, start a project as a guest with two versions of a real document, and see what the findings list catches that your last manual review did not.