The Spreadsheet Marathon
Downloading massive raw company dumps and wrestling with Excel crashing while you filter out thousands of irrelevant industry-classification codes.
Screening engine
From a raw dump to a screened set
Thousands of candidate companies enter quantitative screening. Persistent loss-makers, negative margins and extreme profit indicators drop out first; the semantic and FAR passes then decide what remains — and record why.
- Accept312
- Flag88
- Reject2,140
Study · FY 2024-25Arm’s-length range
- Lower quartile
- 4.81%
- Median
- 6.42%
- Upper quartile
- 8.17%
Operating margin (OP/OR) on the accepted pool · illustrative output
Curated jurisdiction content
226
Report-engine modules with jurisdiction-specific rules, filing thresholds and documentation conventions — of 250 ISO jurisdictions in the platform registry.
Jurisdictions
Written for the jurisdiction you file in
Every study carries its own jurisdiction module — the range convention, the local-file expectations, the fiscal year-end and the chapter structure a reviewer there will look for. The report engine resolves it from the study, never from a free-text field.
The Problem
Still doing your Accept-Reject matrix manually?
Every benchmarking cycle drains days from your team — and every manual judgment adds risk to the file.
Subjective Trade Descriptions
Spending hours reading vague company annual reports to determine if a company passes the functional profile test.
Audit Scrutiny
Worrying that your comparable selection process lacks the systematic, documented consistency that tax authorities around the world demand.
The Product
An engine, a workflow and the file — one study
Quartyl is not a report generator. The benchmarking engine, the governed review workflow and the deliverables all read the same stored study, so nothing has to be re-keyed — and nothing can drift.
Benchmarking
Comparables that survive scrutiny
- Quantitative rejects first — persistent loss-makers, negative margins, extreme profit indicators, insolvency flags
- A local embedding model scores every trade description against your tested party; an LLM pass adds functions, assets and risks with a verbatim quote as evidence
- Arm’s-length statistics computed on the accepted pool: median, quartiles, interquartile and 35–65 ranges, multi-year averaging
Governance
Decisions, not spreadsheet cells
- Accept, flag, reject and override each capture a reason and the person who gave it
- Analyst → manager review → partner sign-off, on a ten-state study lifecycle
- Every action in the study’s audit trail, with a hash-sealed evidence ledger and audit-packet export
Documentation
An audit file, not a spreadsheet
- Six-sheet Excel workbook: strategy, search result, Accept-Reject matrix, financials, margin analysis, enrichment
- Branded client PDF — cover, contents with live page numbers, abbreviations, chapters and annexures, with generated charts
- Word report for your own template work, all three rendered from the same stored study context
How It Works
Set up a study in 3 steps.
These are the three screens of the study wizard. Running it takes the dump through eight named pipeline steps; the dispositions, the manager review and the deliverables come after.
Study Info & UploadWizard step 1
Name the study, set the tested entity and financial year, and upload the dump — Excel (.xls, .xlsx, .xlsm) up to 50 MB. Quartyl detects the company, revenue, cost and operating-profit columns and shows you the mapping, so you can correct or ignore a column before anything runs.
Study ContextTested party & FAR profile
Jurisdiction, product or service, the function types the tested party performs, and the nature-of-business description. Together they drive the FAR affinity scoring and comparability screening. Optional scope fields — geographies, channel, IP ownership, RPT profile, materiality and your own exclusion criteria — narrow what you will accept later.
Parameters & RunEight named steps
Pick the profit level indicator (OP/OC, OP/Sales, TP/Sales, EBIT/Sales, EBITDA/Sales, Berry Ratio, EBIT/Total Assets), the percentile range, revenue band, related-party cap and single-year or multi-year averaging — then run. Progress is reported step by step, and a step your plan does not include is shown as skipped, not hidden.
Built for modern Tax Practices and Enterprise TP Teams.
- Named pipeline steps
- 8
- Named pipeline steps
- Profit level indicators
- 7
- Profit level indicators
- Jurisdictions with curated report content
- 226
- Jurisdictions with curated report content
- Excel dump ingestion
- Accept–Reject with recorded rationale
- 35–65% safe-harbour range rules
Security & Governance
Uncompromising data security and an audit trail.
Tax data is highly sensitive, and the file has to show who touched it, when, and on what basis.
Your Tenancy, Isolated
Every study, comparable and log row belongs to your firm’s tenant and is scoped at the query layer, not filtered in the UI. Roles decide what an analyst, manager, partner and firm admin can see and do.
Access You Can Prove
Short-lived sessions with server-side token revocation, TOTP multi-factor with single-use recovery codes, per-firm SSO, rate limiting on every endpoint, and an access log you can search and export.
AI With Receipts
The semantic screen runs on a local embedding model. The FAR analysis calls the model gateway your deployment configures, and its output carries the verbatim quote it relied on — accepting, flagging or rejecting stays a human decision, written to the evidence ledger.
Hosting region, encryption at rest, backup cadence and third-party certifications are deployment terms agreed per engagement — not product features claimed on a web page.
Free TP Tool
Try our free TP tools.
Type in a business description and our NIC Code Finder searches all 1,297 NIC-2008 sub-classes right in your browser — and surfaces the code and keywords to pull from your database.
NIC / NACE aware