Data platform modernization and migration consultants

Buyers use several terms for related work. Data migration consulting is moving data and code to a new system. Data warehouse consulting covers designing and running a warehouse, new or existing. Cloud data migration means the target is a cloud platform. Firms usually sell all three together, and a proposal should say which parts are in scope.

Code conversion, validation and report repointing often take more effort than copying data. Our legacy data warehouse modernization guide sets out the sequence step by step.

How firms package migration work

These examples come from the firms' own pages, checked in October 2026, and are not ranked.

Automation claims are the firm's own. Ask what the tool converts, and what rate it achieved on code like yours.

AWS Prescriptive Guidance lists seven migration strategies: retire, retain, rehost, relocate, repurchase, replatform, and refactor or re-architect. For a data platform, the first one matters most: tables, jobs and reports that nobody uses should be retired before anything is converted.

Our cloud data platform migration guide compares AWS, Azure and Google Cloud as destinations.

Our data platform TCO analysis converts one budget into each unit and models three years of cost. The data lakehouse guide covers open table formats and unified batch and streaming design. For partner choices on the two most-asked platforms, see Snowflake consultants and Databricks partners, and our guide to Snowflake versus Databricks consulting.

The rates are role medians from GSA CALC+: cloud architect $185 an hour (887 records), data engineer $149 (838 records) and data architect $158 (913 records). They are fully burdened ceiling prices on U.S. federal contracts, not commercial quotes, and commercial rates for the same roles are often higher. The hours are our assumptions: 40 hours a week for each full-time person. The last column reprices the same hours at the 25th and 75th percentile GSA rates for each role.

The estimates leave out the platform bill for the new system and the old one while both run, conversion and validation tools, BI rework by your own team, business-user testing and training. To size a program from your own inventory instead of a team size, multiply code objects by the share needing manual work and the hours each takes; the legacy warehouse guide shows the arithmetic. Add the overlap period explicitly, because both platforms bill while it lasts.

Platform vendors also publish conversion tools: SnowConvert AI, BigQuery Migration Service, Databricks Lakebridge and the Fabric Migration Assistant. Each lists the sources it supports, so check yours before accepting an automation claim.

Then require, in writing:

1. An assessment of your own code before a fixed price.
2. A conversion rate measured on a sample of your code, with a definition of what counts as converted.
3. Validation evidence for each wave: row counts, aggregates and report comparisons.
4. Parallel-run exit criteria tied to your business cycle, and a rollback plan.
5. A named team with continuity terms.
6. Fees tied to acceptance, with change control.

Our RFP template turns these into questions.

• A fixed price before an assessment of your own code.
• An automation figure with no definition and no measurement on your sample.
• A single cutover with no parallel period, exit criteria or rollback.
• BI and report rework excluded, or left to your team without being scoped.
• Dual-running platform costs, training and business testing missing from the budget.
• A target platform chosen before workloads were measured.
• Accelerators that only the firm can operate, with no code or license transfer.
• Case studies that report speed without the scope behind it: objects, volume, source dialect and what the timeline included.