The readiness gap: why migrations fail before they start

We turn down about 30% of migration engagements. Here's why.

The five signals

1. No executive sponsor. Migrations are 9-18 month commitments with quiet quarters. Without a senior champion, the first re-org kills it.
2. No documented data model. Not a stale wiki — a current dictionary. If everything lives in three people's heads, the assessment phase doubles and key-person risk becomes the project's biggest risk.
3. Vague ROI. "We need to modernize" isn't a business case. "We need 4-hour SLA queries instead of 4-day batch" is.
4. Change-resistant team. If the team that built the current stack is scared of the new one, training won't fix it. We've never seen that go well.
5. Cloud uncertainty at IT/InfoSec. If cloud adoption itself is contested, fix that first. The data team can't carry that fight.

What "readiness" actually means

It doesn't mean perfect. It means three things are true:

• Someone senior owns the outcome.
• The team has the bandwidth and curiosity to learn.
• Leadership knows what success looks like in dollars.

If those three are true, almost every other gap is fixable inside the engagement. If any are missing, no amount of engineering rescues it.

The same logic applies to AI

We use a parallel checklist for AI Enablement engagements — same five-signal idea, different signals. Document corpora and access models matter more than the model choice; if you're scoping a RAG implementation or document automation build, the readiness work is what decides whether you ship in week eight or quietly stall in week six. For predictive ML, the equivalent gating happens around lineage, feature stores, and serving — the ML productionization deep-dive walks through what we look for.

The 15-minute readiness assessment we built scores you against all five. It's the same conversation we'd have over coffee — just faster and you don't have to share your calendar.