Skip to content

Database consultant

Most of the database work I get is a data-model problem underneath, and the data model is the one part of a system you can’t cheaply undo once a live business is running on it. The framework and the front end can be swapped out with enough effort, the schema can’t.

The problems I solve here

The first is the data model that was rushed and is now load-bearing. It got designed in an afternoon at the start of a project, before anyone understood the domain properly, and now the whole business runs on it. The wrong shape forces awkward code everywhere, makes simple questions hard to answer, and quietly corrupts data through edge cases it can’t represent. By the time I see it, fixing it means a careful migration of live data, which is delicate work — and exactly why getting it right early matters so much.

The second is the database that’s become slow as the business grew. Queries that were instant on small data now crawl, the application is sluggish, and the instinct is to throw bigger hardware at it. Usually the real cause is a handful of missing indexes, a query pattern fighting the schema, or a data model that no longer fits how the data is actually used. The fix is diagnosis, not money — finding the few real bottlenecks rather than scaling around them.

The third is the business whose data is scattered and untrusted, locked across separate SaaS tools that each own a piece of it. They can’t get a straight answer about their own operation because no single place holds the truth. The work there is to bring the data they own into one model they control, so it can finally be trusted and built upon — which is where databases meet owning your data outright.

How I think about databases

The first principle is that the data model deserves more design time than almost anything else, because it’s the least reversible decision you’ll make. I’d rather spend real care getting the schema right up front than save a day now and pay for it with a painful live migration later. Most serious data problems trace back to a model designed before the domain was understood. Slowing down at that one point pays back for the life of the system.

The second is that most performance problems are design problems, not scale problems. The reflex when a database is slow is to buy bigger infrastructure, and it’s almost always the wrong first move. The usual culprits are a missing index, a query doing far more work than it needs to, or a schema shape that’s wrong for the access pattern — none of which more hardware fixes, it just hides them for a while at a cost. I find the actual bottleneck first. Throwing money at slowness is treating a symptom and renting away a problem you could have solved.

The third is that your data is the most important thing you own, and the database is where that ownership is real or it isn’t. When your data lives in a model you control, you can answer any question, build any feature, connect anything, and move on your own terms. When it’s scattered across rented tools that each hold a slice, you don’t really own your operation — you’re renting access to pieces of it. Consolidating your data into something you own is one of the highest-leverage things a business can do, because every later capability — reporting, automation, AI, new products — is built on top of it. Get the foundation right and owned, and everything above it gets cheaper.

How I’d actually do it

I’d start by understanding the domain properly before drawing a single table — what the real entities are, how they relate, how the data will actually be used and queried. The schema follows the business, not a generic template. I keep it as simple as the domain genuinely allows, because needless complexity in a data model is the most expensive complexity there is, but no simpler than the domain requires, because a model that can’t represent reality forces corruption later. That balance is the craft.

For a business with scattered data, I’d consolidate the one most valuable slice first — the data that, unified, answers the most pressing questions — into an owned model, prove the value, and bring the rest in from there. The MVP instinct again: not a grand data-warehouse programme, but the most useful unification first, owned, with value banked immediately and the foundation laid for everything that comes next.

Where this connects

This is one of the specialisms I’m hired for. The data model is the foundation that system architecture is built on — get it wrong and no architecture above it can fully recover — and it’s the bedrock under any dashboards you trust, since a dashboard is only as honest as the data beneath it. The engagement this most often powers is an architecture review when the question is whether your data foundation will hold, or an MVP development when it’s time to build the owned model. For what owning your data really means as a principle, see owning your data.

When you need a team instead

A large data platform — many sources, real scale, continuous operation and governance — is ongoing team work. That’s my agency, Digital Royalty, with the crew to build and run it. My lane is the senior call on the model, the diagnosis of why it’s slow or wrong, and the lean owned foundation. I’ll tell you when it’s genuinely outgrown one person.

Who this isn’t for

If you want to scale your way out of a database problem with bigger infrastructure rather than fix the design, I’m not your fit — I’ll find the real cause first. I’m worth hiring when you want the foundation right, because everything you build later sits on it.

Talk to me

If your database is slow, your data model is fighting you, or your data is scattered across tools you don’t control, tell me about your problem and I’ll tell you what’s actually wrong — and what it’s worth owning.