Global – October 27, 2026 -- Unstructured text makes up an estimated 70% to 90% of organizational data, according to Gartner analysts speaking at the Data & Analytics Summit in London this year, and Alteryx is pitching workflow-embedded text mining as the way to put it to use.
Three paths compete for the same text-analysis budget
Enterprises can write custom code, license a dedicated text-analytics tool, or run text mining inside the workflow platform that already handles blending, joins, and reporting. Alteryx argues the deciding factor is which approach keeps running unsupervised after its creator moves on.
The custom-code route depends on scarce talent. In the World Economic Forum's 2025 Future of Jobs Report, 63% of employers named skills gaps as the biggest barrier to getting more value from data and AI investments. A widely reported MIT-affiliated study found that only a small share of custom-built enterprise AI tools reach production.
Point solutions perform well on their intended use case but sit outside the analytic workflow. Their categorized output must be exported, re-joined, and re-explained to downstream report owners every cycle.
Five criteria frame the evaluation
Alteryx lists five tests for any approach: accuracy on a team's own jargon and typos, repeatability without a standing data-team request, an audit trail for how labels were assigned, integration with the analytic workflow, and a cost and tier fit with existing software buying.
Alteryx maps four tools onto a support-ticket backlog
Within Alteryx One Designer, the text mining tools run as part of Alteryx Intelligence Suite. Each step adds a column next to existing ticket data, with no export or manual rejoin.
- Text Pre-processing tokenizes and cleans raw text. Product documentation warns against running it immediately before sentiment scoring, because it strips words the sentiment model needs.
- Sentiment Analysis runs on raw ticket text to flag frustrated tickets.
- Zero-shot Text Classification sorts tickets into analyst-defined categories such as billing confusion or shipping delay, with no labeled training set.
- Topic Modeling surfaces language clusters that no predefined category captures.
Sentiment scoring is limited to English
Sentiment Analysis currently covers English text only. Text Pre-processing, Text Classification, and Topic Modeling extend to French, German, Italian, Portuguese, and Spanish.
In the source's example, a third of a support inbox arrives through a German site. The team could cover categorization and topics everywhere and sentiment only in English, hold the German queue for a later phase, or route German sentiment scoring through a separate process.
Alteryx also concedes the limits of workflow tooling. A model fine-tuned on millions of internal documents for one narrow, high-stakes decision calls for a dedicated machine learning effort.
Alteryx is offering a free 30-day trial of Alteryx One, and advises teams to run last week's ticket export through the steps above.