How AI can rescue product classification for indirect tax teams

It’s Monday morning. Your company is launching an aggressive number of new SKUs next week and the ERP cutover is in 10 days. Your indirect tax team is staring at a large spreadsheet of product descriptions that need tax codes assigned before go-live.

You know the drill: reading product descriptions line by line, researching commodity codes, cross-referencing state taxability rules, and hoping your classifications are consistent across reviewers. Your senior tax manager is tied up on the project, your newest team member is nervous about getting codes wrong, and the business is asking why tax is always the bottleneck.

This is the reality of manual product classification for indirect tax teams at multinational companies. Hours add up and hopefully no errors creep in, because when classification is wrong, thousands of invoices could be impacted.

Jump to ↓

The data and control problem with product classification

AI-enabled product classification for indirect tax: A practical evolution, not a silver bullet

The strategic shift: From data gathering to value creation

Who benefits most from AI-enabled product classification

Build strategic capacity now for your indirect tax team

The data and control problem with product classification

The pain of manual classification runs deeper than long hours. At its core, product classification is a data and control problem that creates hidden compliance risk.

First, it’s a data integrity issue. Product mapping connects your SKUs to the tax-relevant attributes that drive real-time determination in your ERP or tax engine. Product classification, selecting the right commodity code, is one critical step in that mapping process. When classification is done manually across thousands of SKUs, inconsistency is inevitable. Different reviewers interpret similar products differently. The same “industrial adhesive” might get three different codes depending on who reviewed it and when. Wrong codes at the mapping stage mean wrong tax at transaction time, which multiplies across every invoice until someone catches it.

Second, it’s a control and governance problem. Spreadsheet-driven classification workflows lack audit trails. You can’t easily document who classified what, when, or why, making it nearly impossible to defend your positions during audits or meet SOX requirements. When products change or tax rules update, there’s no systematic way to identify which SKUs need reclassification. You’re constantly playing catch-up, reacting to issues after incorrect tax has already been charged.

Third, it’s a time-sensitivity problem. Classification must be correct before SKUs hit live orders, not after. “We’ll fix it later” means incorrect tax on hundreds or thousands of invoices, customer credits, refund requests, and audit exposure. This pressure is especially acute during ERP implementations, new market entries, product launches, and regulatory changes.

The uncomfortable question you need to ask yourself is when tax authorities increasingly use AI to examine your transaction data, can you afford to keep mapping products manually?

AI-enabled product classification for indirect tax: A practical evolution, not a silver bullet

The good news is that AI-powered product classification offers a practical, low-risk path forward when it’s built on tax-specific intelligence, not generic automation.

How it works

CoCounsel is the AI engine embedded within Thomson Reuters ONESOURCE Determination. When you input product descriptions during your mapping workflow, CoCounsel analyzes them against the Thomson Reuters proprietary tax content database and the U.S. Product Tree (UNSPSC standards), returning up to three commodity code suggestions with confidence scores. Your team reviews and approves without requiring any separate login or additional tools.

How tax-specific AI classification works within your mapping workflow

Traditional product mapping follows a familiar sequence: extract product data, review and group similar items, manually classify each SKU with commodity codes, configure those codes in your system, then test. The classification step of matching product descriptions to the right tax codes is consistently the most time-intensive bottleneck.

AI-enabled classification with ONESOURCE Determination and CoCounsel automates this step. Input a product description like “industrial adhesive for automotive assembly,” and the system returns up to three commodity code suggestions, each with a confidence score:

Option 1: Industrial adhesives – automotive grade (94% confidence)
Option 2: Chemical compounds – industrial use (81% confidence)
Option 3: Manufacturing supplies – general (67% confidence)

Why tax-specific AI matters and generic tools fall short

Not all AI classification tools are created equal. Generic AI models may express high confidence in incorrect tax classifications because they lack specialized tax knowledge. The most effective solutions ground their recommendations in authoritative tax content.

This foundation provides critical advantages: superior accuracy grounded in tax authority guidance, reduced “hallucination” risk through specialized databases, and transparent decision-making that supports audit defense.

Practical benefits within your mapping workflow

Speed: Bulk classify 100 SKUs in 5 minutes versus days of manual review
Consistency: Same logic applied across all products, reviewers, and regions
Control: Confidence scores create audit trails; you maintain final approval
Risk-based workflow: Auto-process high-confidence classifications (90%+); route uncertain cases to specialists

How you maintain control while AI handles the heavy lifting

The AI provides suggestions; your indirect tax team makes final decisions. Instead of creating classifications from scratch, you review and approve. High-confidence classifications process quickly with minimal oversight. Your senior professionals focus expertise on complex, low-confidence cases that truly require human judgment, like nuanced taxability questions or conflicting jurisdictional interpretations.

The strategic shift: From data gathering to value creation

AI-enabled product classification fundamentally transforms how indirect tax teams allocate their time and expertise.

Traditional tax functions spend most of their time gathering and validating data, leaving only a fraction for strategic analysis. AI-powered classification with confidence scoring enables organizations to invert this dynamic.

When routine classifications process automatically at high confidence levels, tax professionals can redirect their focus to work that drives real value, like tackling complex product mapping challenges that require specialized expertise, conducting strategic tax planning and optimization across jurisdictions, providing proactive advisory to business stakeholders on new products and market entries, and managing cross-jurisdictional risk with sophisticated analysis.

Junior team members can confidently handle high-confidence classifications with automated workflows. Mid-level staff manage medium-confidence items with structured review processes. Senior tax professionals concentrate on the complex, uncertain classifications where their judgment is irreplaceable.

Who benefits most from AI-enabled product classification

ERP implementation teams racing against go-live deadlines with thousands of SKUs to map
Growing companies launching new products, entering new markets, or managing M&A integrations
Lean tax teams stretched thin by expanding product catalogs and compliance demands
Organizations facing audits that need defensible, consistent classification logic with clear audit trails

Build strategic capacity now for your indirect tax team

The stakes for accurate, efficient product classification are rising fast. E-invoicing mandates, real-time reporting requirements, and continuous transaction controls are becoming the norm globally. Tax authorities are already using AI analytics to examine your data with unprecedented depth.

In this environment, accurate and efficient product classification within your broader mapping process is a competitive advantage. Indirect tax teams that adopt AI-enabled classification now gain faster implementation cycles, more precise tax treatment, reduced audit risk, and strategic capacity to support business growth.

Discover how ONESOURCE can put your team ahead of the next classification challenge in our latest e-book, How tax determination tools are redefining career paths.