Benchmarking Study in 30 MinutesBenchmarking Study in 30 Minutes
Learn how AI technology transforms transfer pricing benchmarking studies, enabling professionals to deliver high-quality benchmarking results in record time.
Borys UlanenkoCEO of ArmsLength AI
Sven van der HeideCPO of ArmsLength AI
60min webinar
Contents↓
Contents
TL;DR - Key Takeaways
AI technology reduces benchmarking time by up to 80%, with TPA Global now spending 4-5 hours instead of 15-20 hours per benchmark
Traditional benchmarking faces consistency issues, time constraints, and documentation quality problems that AI can solve
The AI-powered solution analyzes company websites for detailed descriptions, applies consistent acceptance criteria, and provides comprehensive documentation
Complete benchmarking studies can be done in approximately 30 minutes with the ArmsLength AI platform
Benchmarking Study in 30 Minutes with AI: Webinar Summary
Meet the Experts
Borys and Sven from ArmsLength AI demonstrated how their AI technology transforms transfer pricing benchmarking studies.
Borys is the founder and CEO with a background in transfer pricing, having started the company about 10 months ago with the ambition to bring AI technology to the transfer pricing profession.
Sven joined as a partner, bringing extensive experience in finance, tech, and data analytics from his time at a Big Four firm. He explained that his passion for tech and data led him to find ways to answer client questions more effectively through data-enabled and visualized solutions.
About ArmsLength AI
ArmsLength AI combines transfer pricing expertise with artificial intelligence technology.
Their platform focuses on enabling transfer pricing professionals to deliver high-quality benchmarking results in record time while ensuring data security, data quality, and integrity.
The core mission is to transform the way transfer pricing professionals work by leveraging AI technology.
The Traditional Benchmarking Process and Its Problems
The webinar outlined the five steps in traditional benchmarking:
The presenters demonstrated a live benchmarking study for a European multinational providing marketing support services to global subsidiaries, focusing on finding comparable marketing service providers in Europe.
Step 1: Data Loading (2-3 minutes)
Uploaded source data from a database (150 companies)
Mapped database fields to the platform (company name, ID, country, website)
Added financial data fields (operating profit, revenue, assets, employees)
Used AI to automatically search for missing websites (found websites for most of the 25 companies missing this information)
Step 2: AI Web Search (6 minutes)
AI analyzed company websites for 141 companies
Created detailed descriptions covering:
Business descriptions
Products and services offered
Functions and activities
Structure and affiliations information
Each company analysis took only a few seconds, with the entire batch completed in about 6 minutes
Step 3: Accept/Reject Analysis (4 minutes)
Set up the acceptance/rejection criteria using:
Library strategies (pre-defined templates for different industries)
AI assistance to generate strategies from high-level descriptions
Custom strategies typed in natural language
Selected a marketing services strategy including:
Products/services to accept (marketing services)
Products/services to reject (products)
Ideal functional profile (service provider)
Functions to reject (manufacturing, retail, distribution)
AI evaluated 133 companies against these criteria
Generated decisions on products/services compatibility, functional compatibility, and independence
Automatically tested for independence in addition to user-defined criteria
Step 4: Financial Analysis (ongoing)
Displayed real-time updated ranges as companies were analyzed
Calculated ranges based on accepted companies (39 accepted from the set)
Showed weighted average margins and individual yearly values
Allowed switching between different PLIs (profit level indicators)
Enabled comparison of ranges at different stages of the analysis process
Step 5: Human Review
Categorized companies by review priority (high, medium, low)
Enabled review of AI decisions with detailed evidence
Allowed override of AI decisions with manual motivations
Provided comprehensive information for each company:
AI-generated descriptions
Product/service and functional assessments
Screenshots of analyzed web pages
Specific quotes used by AI to make decisions
Translation options for foreign language sources
Changed status from "AI suggested" to "Human reviewed" after confirmation
Step 6: Export Results
Generated Excel report containing:
Detailed company descriptions
AI accept/reject results
Human review metrics and decisions
Financial data and analysis
Provided screenshots and supporting documentation:
Individual company screenshots
Complete archive of all screenshots for audit support
Quotes and evidence used for decision-making
Note
The entire process demonstrated how a complete benchmarking study that traditionally takes 15-20 hours can be completed in approximately 30 minutes with AI assistance.
Key Benefits Demonstrated
Time savings:
Reduces benchmarking time by up to 80%
TPA Global (customer example) now spends 4-5 hours per benchmark instead of 15-20 hours
Manual review portion, which typically takes 90% of time, is now largely automated
Quality improvements:
Consistent application of criteria across all companies
AI doesn't suffer from fatigue or time pressure
More accurate decisions than manual analysis
Standardized approach eliminates human inconsistencies
Better documentation:
Comprehensive screenshots for all companies captured at the time of analysis
Robust decision substantiation with specific website quotes
Audit-ready documentation that's difficult to challenge
Complete record of decision-making process
Business impact:
Improved margins for advisory firms through reduced time commitment
Better client experience through faster delivery
Enhanced quality making benchmarks easier to sell
Increased employee availability for other value-adding tasks
Improved job satisfaction by reducing tedious manual work
Real-World Results
Key point
The presenters shared feedback from TPA Global, an international network of transfer pricing professionals using ArmsLength AI since January:
Significant reduction in benchmark duration (from 15-20 hours to 4-5 hours including human review)
Reduced costs associated with benchmarking
Increased employee availability for other tasks
Improved job satisfaction
Reduction in error rates over time
AI decisions proving more accurate than human decisions on average
The technology has already analyzed 93,000 companies across hundreds of benchmarks, demonstrating widespread adoption.
Next Steps for Transfer Pricing Professionals
The presenters recommended:
Evaluating whether your firm does benchmarks or plans to add them
Considering how many benchmarks your firm performs annually
Booking a demo to see the technology in action
Getting a free trial with credits to test on existing benchmarks
Selecting a subscription package based on benchmarking volume needs
Tip
Different subscription packages are available based primarily on how many benchmarks or comparables a firm processes annually.
Related Resources
Want to learn more about transfer pricing and AI? Explore these related articles and resources:
Borys UlanenkoCEO of ArmsLength AI
Sven van der HeideCPO of ArmsLength AI