Skip to content

Rung 2 · Fixed price · 6 weeks

One named deadlock, fixed end to end, with the hole closed behind it.

Dedupe at scale, a lifecycle rebuild, attribution that reconciles to finance, a migration that never finished, a broken sync. Gated rollout — three rows that cannot merge, then one, then ten, then a stratified 250 — with every batch authorised and every before-state captured.

A fit when

  • The audit (mine or yours) has named the deadlock and the route the platform gives you to fix it.
  • The fix is irreversible — merges, deletes, mass property writes — and needs guards, gates and an audit trail.
  • Your team will run it afterwards and needs recordings and a runbook, not a black box.
  • One deadlock. A second one is a second sprint, scoped after the first lands.

Not a fit when

  • The scope is "clean everything up". That is an audit first, then one sprint at a time.
  • You want a new platform implemented from scratch. That is an implementation partner, not a rescue.
  • Nobody on your side can authorise batches within a business day. The gates stall without a decision-maker.

What you get

Deliverables

Guard set and gates

The safety rules enforced in code, not in which rows are in the spreadsheet. Gates 0 → 1 → 10 → 250 before anything runs at scale; the workflow off between every gate.

Before-state capture

Every irreversible action has its before-state exported and stored in your systems before it runs. If a rule is wrong, the evidence exists to reverse what can be reversed.

Reconciliation after every batch

Executions compared to enrolments after each batch — the tripwire that catches silent partial runs the platform will not flag.

Review list with reasons

Every refused record routed to a static list with the reason written out, so the refusals are a work queue, not a mystery.

Runbook and recordings

The weekly working sessions recorded, and a runbook your team can operate the fix from without me.

The upstream rule

The import standard, sync rule or required property that stops the deadlock recurring — written, agreed, and enforced in the platform.

How it runs

Timeline

Weeks 1–2

Decide offline against a full export; write the guard set; build the executor or workflow; run the three rows that cannot merge.

Weeks 3–4

Gate 1 (one real record, checked field by field), gate 10 (stratified across stages and countries), gate 250. Each gate authorised in writing; before-states captured.

Weeks 5–6

Full run in authorised batches, reconciliation after each, review-list triage with your team, the upstream rule installed, runbook and recordings handed over.

What you have on day 30. Gates 0 through 250 have run clean, the review list is being worked by your team, and the guard set is documented. The full run is scheduled and the upstream rule is drafted.

Two rules on every engagement

Nothing irreversible runs without a captured before-state and your explicit authorisation.

Every engagement leaves recordings and a runbook your team can operate from.

Questions

What counts as one deadlock?

One object, one mechanism, one outcome — for example "duplicate companies in HubSpot created by contact imports" or "Marketo–Salesforce lead sync failing on ownership". If the audit found three, the scope memo ranks them and the sprint takes the first.

What if the platform has no native route?

That is usually the case at scale; the case study is exactly that. The sprint includes the build of a guarded executor where one is needed. If the build alone is the engagement, it is priced as a Custom Build.

Who authorises the batches?

You or a named delegate, in writing, within a business day. The gates are designed so the person authorising can see the before-state and the plan for each batch.

What happens to the records that were refused?

They stay untouched on a review list with the refusal reason. Most are resolved by your team from the runbook; a deliberate retry batch can be scheduled inside the sprint.

Is the audit required first?

Not if you have equivalent evidence. A 30-minute look at your findings decides it. Without evidence, the audit is the sprint's first two weeks and is credited.

Scope the sprint.

A 30-minute scoping call tells you which rung you need — or that you don't need one. No deck, no discovery theatre.