Data strategy assessment and consulting

Whether your data can support the AI project you have in mind, and what to fix first if it can't. AI plans usually assume the data is in better shape than it is, and the gaps show up months into the project. The assessment finds them in a week, before you commit a budget.

In the sample report below, a fictional distributor wanted a sales assistant and a demand forecast. Its sales history was ready for forecasting, but 31% of its products lacked the attributes a sales assistant would need, and revenue had three definitions that disagreed by 3 to 6% a month.

Monday is the kickoff with your sponsor, and I get read-only access to your warehouse, BI tool and source systems. On Tuesday I interview the people who use and maintain the data, usually eight to ten of them across the business, IT and the data team. On Wednesday I do my own analysis of your technology and data: architecture diagrams, system setup, key tables, pipeline history and the most-used dashboards. Thursday goes to ranking the findings and sizing the options. On Friday I present the readout to your leadership team.

I don't change anything in your production systems during the diagnostic.

A written readout of about a dozen pages that your leadership team can act on:

• What's ready and what isn't for each thing your AI project needs, with the evidence behind every finding
• What each gap costs you today, in numbers
• The options for moving forward, each with time to first result, effort and main risk, plus my recommendation
• A roadmap for the recommended option and a fixed quote for its first phase
• The decisions you need to make next, and what happens in the first two weeks

The sample report for Arvelle Industrial Supply, a fictional distributor, shows the full format (PDF, 13 pages).

You decide. The roadmap is written so your own team can carry it out. If you want my help, I quote the first phase at a fixed price, and the engagement picks up from there: I help name data owners, get agreement on metric definitions, set up monitoring on the pipelines that matter and run the first pilot.

Some of what the roadmap calls for may sit in the other two services, such as analytics for a semantic layer or data engineering for pipelines that keep failing. I combine them in one plan and one quote.

The one-week diagnostic costs a fixed $8,000. Work after that is quoted at a fixed price once the diagnostic shows what it involves, so you know the cost before you commit to it.