Most engagements that fail don't fail at the model. They fail at the data hand-off — and they fail quietly, weeks after the kickoff, when someone finally admits the export was never going to happen on the timeline anyone agreed to.
The blocker we see most often
A team commissions a project that depends on internal records: a CRM slice, a transactions table, a folder of contracts, a year of ticket history. Everyone nods. Then the actual access request hits the data team, the security team, the vendor with the export license, or the one engineer who knows the schema — and it sits. The clock keeps running, the invoices keep coming, and the work starts inventing around the missing data rather than with it.
We treat that pattern as a scoping failure on the seller's side, not a procurement failure on the buyer's side (Class E — this is consistent with what experienced data leads have written publicly for years about why analytics projects stall). If we accepted the engagement knowing the access path wasn't tested, we co-signed the stall.
What we do differently before the contract
The workshop exists, in part, to break this pattern. Before any statement of work, we run a small, concrete access check against the actual systems the engagement would touch (Class C). Not a questionnaire. An actual end-to-end pull of a representative slice, by the person who would have to do it under contract, through whatever approval chain governs production data.
What the check produces:
- A named owner for each data source — not a team, a person.
- The real approval path, with the people who actually sign, and how long their last comparable approval took.
- A working extract of a small but representative slice, in a format the receiving system can read without bespoke glue.
- A documented refusal, redaction, or scope limit where one exists — written down, not assumed.
If we can't get the slice, we say so in the readiness note and we don't sign. That sentence is doing a lot of work, so it's worth repeating: if the access path isn't demonstrably open, the engagement doesn't start.
Why "we'll figure it out in week one" is the trap
"We'll get you access on day one" sounds like momentum. In practice it transfers all the political cost of opening a data path from the pre-sale phase (where the buyer still has leverage and we still have time) into the delivery phase (where the meter is running and the easiest thing for everyone is to fake progress). The retainer treadmill starts there. The opaque pipeline starts there. The story that ends in "the model just needs more tuning" almost always starts with an access gap nobody made visible early enough.
We'd rather lose the engagement at the readiness step than win it and run the clock against a closed door.
What this looks like from your side
If you're scoping a project with us, expect three things in the readiness pass:
- A short, written inventory of every data source the work needs, with the owner and the approval path beside each one.
- A live access attempt on at least one source per category — production-shaped, not a sample CSV someone emailed around.
- A written go / no-go on each source before pricing is finalized. The no-gos get a remediation plan or a reduced scope. They don't get hidden.
It's slower at the front. It's dramatically faster — and cheaper — across the life of the work.
Related reading on this site
- Are you ready for the workshop? — the readiness self-check, including the data-access section.
- The workshop-readiness buyer's guide — what a readiness pass actually covers and what it costs.
- Book the workshop — the engagement starts here, and the access check is part of it.