An honest comparison of three CV parsers Dutch and European staffing agencies consider. What does what, who it's built for, and the differences the sales demo doesn't mention.

In short

Textkernel and Daxtra are enterprise CV parsers that have been the market standard for two decades, used primarily by large international recruitment tech companies. Both ship a parser API that ATS vendors embed in their products, focused on multilingual extraction and high volumes. Spadework offers two related products: a CV-Parser API for ATS integration, and a CV-Transformer that goes a step further by converting CVs to the staffing agency's house style, matching them to a job description, and syncing back to the ATS.

Short answer: if you need a pure parser component for your own software, compare Spadework CV-Parser with Textkernel and Daxtra. If you run a staffing or recruitment agency and want a complete workflow (parser plus house style plus ATS sync), look at the Spadework Suite. Textkernel and Daxtra don't deliver that second part.

TL;DR

Recruitment teams in the Netherlands and across Europe ask this question more often, and the answer depends on what you're actually buying. A parser is a piece of technology that extracts text from a CV and structures it into fields. A CV transformation workflow is the full chain: parsing, formatting in house style, matching to a job, syncing back to the ATS. Textkernel and Daxtra cover the first. Spadework covers both.

In this article: what a parser does, how the three vendors differ on accuracy, languages, pricing and target audience, and how to choose based on who you are (ATS vendor or recruitment agency).

What is a CV parser, exactly?

A CV parser reads a document (PDF, Word, sometimes a scan) and pulls out structured data: first name, last name, work history, education, skills, languages. The output is JSON or XML, ready to load into a database or ATS. What a human does in ten to fifteen minutes, a parser does in seconds.

Two things to keep in mind:

A parser is a building block, not a finished product. Recruiters never interact with it directly. The parser sits hidden inside an ATS, CRM or recruitment tool.

Accuracy figures are vendor-reported. Independent benchmarks are scarce. Always test on your own CV dataset before signing.

The three vendors

Textkernel

Part of Bullhorn since 2024. Twenty years as the market leader in CV parsing, used by more than 60% of the global HR tech industry. Processes 2 billion CVs and job postings per year by their own count. Supports 29 languages for CVs and 9 for job postings. Two parser variants: a classic parser (fast, around 0.5 seconds per CV) and an LLM Parser (higher accuracy, higher cost, longer processing time). Vendor-reported accuracy: 95%+ on the most critical fields, with LLM Parser reducing remaining errors by up to 30%.

Built for: ATS vendors, large enterprise recruitment companies, international staffing groups. Pricing on request, enterprise contracts, typically several thousand euros per month with annual commitments.

Daxtra

UK-based, also more than twenty years in the market. Specializes in multilingual parsing with dedicated language engineers per language. Supports 40+ languages. Reports 90% accuracy out of the box, without training on customer data. Used by 2,500+ recruitment organizations globally, integrates with 400+ ATS systems. Processes 100 million CVs per month. Strong in APAC and at large international agencies serving niches in Chinese, Japanese or Russian.

Built for: international staffing agencies, ATS vendors with multilingual customers, niches where Chinese or Japanese CVs are standard. Pricing on request.

Spadework

Amsterdam-based, founded specifically for the recruitment automation layer on top of ATS systems. Ships two related products:

CV-Parser API: parser component that turns CVs into structured data, built for ATS vendors and internal tooling.

CV-Transformer: end-to-end workflow where a raw CV (any language or format) gets parsed, converted to the agency's house style, matched against a job description, optionally enriched with LinkedIn data, and pushed back to the ATS. Average processing time: 2 minutes per CV, versus roughly 60 minutes manually. GDPR-compliant, with anonymization options.

Built for: Dutch and European staffing agencies, contractors and recruitment firms that deliver CVs to clients in their own house style. Clients include Maandag, RGF, Bender, Kayak and Talentcare. Transparent pricing, no enterprise contract required.

Comparison table

Which one do you pick?

You're building your own ATS or recruitment tool

Then you're comparing three parser APIs: Textkernel, Daxtra or Spadework CV-Parser. The choice depends on languages (Daxtra wins on breadth, Textkernel on depth in Western Europe), volume (Textkernel and Daxtra are built for billions of documents) and budget (Spadework tends to be more transparent and lower priced for smaller volumes). Request a test on your own CV dataset from all three before signing.

You run a staffing or contracting agency

A pure parser doesn't solve your problem. You need something that receives the CV, parses it, converts it to your house style, matches it to the job and delivers it back into your ATS. The first part you can do with Textkernel or Daxtra, but the rest you have to buy elsewhere or build yourself. Spadework CV-Transformer covers the full chain in one product.

You do international hiring in niche languages

Daxtra is stronger on languages like Chinese, Japanese and Russian. Textkernel has deeper coverage in European languages. Spadework is multilingual but focused on the Dutch and European market.

FAQ

Is Spadework a direct competitor to Textkernel?

Only on the CV-Parser API. The CV-Transformer is a different product category. An agency using Textkernel still typically uses the output to manually or semi-manually build a styled CV. The Transformer replaces that last step.

What's the difference between parsing and transforming?

Parsing extracts data from a CV and places it into fields. Transforming takes that data and builds a new, styled document in the agency's house style, often tailored to a specific job. Parsing is a technical operation. Transforming is a commercial one.

How do you measure accuracy fairly?

Not by vendor-reported percentages. Ask each vendor to parse a batch of 100 CVs from your own dataset. Compare per field (name, experience, education, skills) against manual extraction. Count how many are correct, blank or wrong. That's the only honest number.

Do all three handle Dutch CVs?

Yes. Textkernel and Daxtra both parse Dutch with high accuracy. Spadework is built in the Netherlands and trained on Dutch CV conventions, including specific work history formats common in Dutch recruitment.

Is an LLM parser better than a classic parser?

Depends on the CV. For standard, well-structured CVs, a classic parser is faster and similarly accurate. For messy, niche or international CVs, LLM parsers usually win. Textkernel offers both; Spadework uses LLM architecture as the basis for both Parser and Transformer.

Sources

Textkernel platform page and developer documentation (textkernel.com, developer.textkernel.com), accessed May 2026

Textkernel x Firemind case study on Amazon Bedrock, 40% cost reduction and 2x speed (firemind.com, 2024)

Daxtra Technologies product page and blog archive (daxtra.com, info.daxtra.com), accessed May 2026

Pin.com independent comparison of resume parsing tools 2026

Gitnux best CV parsing software 2026

Spadework client data May 2026: average transformation time, client list, product specifications

GDPR compliance: ec.europa.eu/info/law/law-topic/data-protection

Want to see how Spadework CV-Transformer performs on your own CVs? Book a 30-minute demo with our co-founder Lucas. We'll show you what it saves on your workflow, in hours per week.