A data engineering consultant sold through a consultancy in London, Amsterdam or Munich bills roughly €800 to €1,400 a day. The same seniority bought as an independent contractor sits closer to €500 to €800, and staffed from Poland, Romania or the Baltics it sits materially below that again. The spread is wide because the title covers two different jobs, and most buyers pay the price of the expensive one while needing the cheap one.
Advisory and build are priced on different logic
Advisory work is judgement sold in days: reviewing a warehouse model before migration, arbitrating batch against streaming, auditing why a platform bill tripled in a quarter. Ten to twenty days, and the rate reflects how few people have made that call before at your scale.
Build work is throughput sold in months: ingestion, transformation models, orchestration, change data capture, tests, backfills, the unglamorous reconciliation that follows a cutover. Six to eighteen months of it, usually.
Consultancies quote both at the advisory rate. A buyer who signs a twelve-month pipeline programme at the price of architectural judgement overpays by 40% to 60% against what the same output costs when sourced as rented data engineers on time and materials. That one mismatch is the biggest avoidable line in most data budgets.
What sits inside the number
A day rate quoted to you is not a salary divided by working days. It carries:
- Non-billable time. Bench, holiday, sick leave and internal work mean a consultant bills 190 to 215 days a year, not 250.
- Delivery overhead. Account management, QA, recruitment and replacement cover.
- Risk. Who absorbs a bad hire, a notice period, a currency move.
- Scarcity in the specific stack. A warehouse generalist and someone who has run a petabyte-scale lakehouse under cost pressure are not the same purchase.
We publish live figures by role and seniority on our daily rates page, and the mechanics of how an hourly number is built up are set out in our piece on what an IT consultant hourly rate actually pays for.
Four routes to the same skill
| Route | Rate posture | You control | Where it breaks |
|---|---|---|---|
| Consultancy practice | Highest | Outcome, not people | Named senior sells, juniors deliver; knowledge leaves with the team |
| Independent contractor | Mid to high | The person | No cover for illness or exit; one opinion, unreviewed |
| Nearshore rented engineer | Mid to low | The person and the backlog | Needs your product owner and your code review |
| Permanent hire | Lowest per day, highest commitment | Everything | Three to six months to fill, and the role must still exist in year three |
The honest comparison is not rate against rate. It is cost per merged, tested, documented change. A cheaper engineer who waits two weeks for access to a source system is not cheap.
What moves a rate by 20% or more
- Stack depth. Streaming, CDC and lakehouse cost control sit above standard SQL and orchestration work.
- Contract length. A six-month commitment prices below a rolling monthly one.
- Regulatory exposure. Personal or payment data under audit adds process, and process adds days.
- On-call. Production ownership overnight is a different product from weekday build.
- Working language and overlap hours. Full overlap with a Western European team is standard from Eastern Europe; it is not from Asia or Latin America.
When renting an engineer is the wrong answer
If what you need is ten days of architectural arbitration before a board decision, hire an independent specialist and pay their rate. Renting build capacity to answer a design question wastes both parties’ time.
Skip us too when the work must be done on your premises under clearance, when your legal position requires a fixed-price deliverable with a contractor carrying the estimation risk, or when nobody internally owns the backlog. Rented engineers need someone who can decide what gets built this fortnight. Without that, you buy velocity in the wrong direction. Our note on what you buy under staff augmentation sets out the division of responsibility in full.
Scoping the first engagement
Start with six to eight weeks, two engineers, one measurable target: a migrated domain, a pipeline retired, a bill reduced. Ask for the CVs that will actually be on the work, not the practice’s reference team. Then extend on evidence. If you want a shape for the wider programme first, our data engineering service page describes how the engagements are structured.