Find the Break in Your Workflow in One Sitting

This is the intake I run at the start of every repair, packaged as something you can run yourself. What follows is the operational work. Work the sections in order. The order is the diagnosis.

Broken workflow automation fails quietly. The form fires, the CRM receives the contact, sales sees the name. Everything looks alive. Then the contact has no source, the deal never opens, and the owner field stays blank. The failure queue catches it before the follow-up window closes.

The failure is rarely in the model. It is in the wiring, the retries and the missing audit trail. This kit finds the wiring.

Five worksheets, one checklist and one triage table. Print the workbook or work it in a doc. The point is the order: inventory before signals, signals before scores, scores before any fix list. Skip a layer and the ranking lies to you.

What is in this workbook

Stakeholder snapshot (fill this in)

Map every source that can create or update records, then every point that mutates them. The map takes 30 minutes. It saves 30 hours of guessing.

Table 1A · Workflow inventory (fill this in)

Table 1B · Entry points to sweep

Tick every source that can create a record. If it is not on the list, add it.

Table 1C · Mutation points (every one needs five things)

Every mutation point needs an owner, a purpose, a field map, a failure path, and a last-tested date. Missing owner means risk and missing failure path means hidden cleanup. Without a last-tested date the workflow is already stale until proven otherwise.

Twelve patterns I find over and over. Tick what you can prove is happening, then carry the ticks into Section 4. Each line tells you where to look: the signal is rarely where the pain shows up.

• 1. Silent webhook failures: the event never arrives and nothing pages anyone. Where to look: endpoint logs, provider delivery history, any step with no failure handler.
• 2. Duplicate writes: the same contact exists three times with different owners, and every report downstream is wrong. Where to look: creation logic with no search-before-create, form resubmits, retry storms.
• 3. Missing idempotency: a replayed event creates a second record instead of updating the first. Where to look: no idempotency_key, no event_id stored, no raw payload archive.
• 4. Stale state: lifecycle stage sits in New for three days while the deal never opens. Where to look: stage transitions with no logger, filters that stop a path silently.
• 5. No audit trail: nobody can say what ran, what changed, or who approved it. Where to look: missing correlation_id, no before and after values, run history past retention.
• 6. Permission drift: a token expired, a scope narrowed, or an API key was rotated while the workflow kept calling. Where to look: 401 and 403 responses buried in module logs, service accounts with no owner.
• 7. Blank owner assignment: the record exists, the owner field stays empty, nothing routes. Where to look: routing rules keyed on a field the form never sends.
• 8. Dead owner IDs: a user left, the CRM kept the ID alive, leads land in an inbox nobody reads. Where to look: owner status checks in the routing layer.
• 9. Field map drift: a property was renamed six months ago and the chain still writes the old name, the old pipeline ID, the old trigger condition. Where to look: any scenario built months ago with no audit since.
• 10. Retries that swallow the error: the chain stops at step four, retries twice, then gives up quietly. Where to look: ignore-error flags, modules with no error handler, filters doing the job of error handling.
• 11. Filters as lead graveyards: one missing field blocks a good record and no queue ever holds it. Where to look: every filter sitting between a lead and the CRM.
• 12. Unowned automation chains: nobody has audited or tested them, and they run every day. Where to look: anything with no owner, no purpose, no last-tested date.

Signal log

Score every candidate fix on two axes. Impact first, effort second. The result places the fix in one of four tiers: critical, high, medium, low. Same tiering I hand back from an intake.

Table 3A · How to score

Table 3B · Impact by effort matrix

Table 3C · Ranked fix list (fill this in)

You now have a ranked list. The last decision is which lane each fix belongs in. A quick win ships this week. A rebuild deserves scope, a quote and an owner. This section is the triage, the procedure I run when I pick up the work, and the evidence pack you hand to whoever builds it.

Repair triage: quick fix or full rebuild

Work down the ranked list. Each candidate gets exactly one lane. If you cannot decide, it is a rebuild. Guessing cheap on a structural problem is how the same break comes back.

The five-step procedure I run on a scoped rebuild

Every intake runs the same priority order. The order is the diagnosis. Fix the layer that silently breaks everything above it before you touch anything visible.

Procedure worksheet

Evidence pack for the stakeholder review

Three things make an audit fast. Two of them are optional, none of them are difficult. Assemble the pack before you open a single scenario. Half the value of an audit is not re-reading the same log three times.

• Required: what breaks, which systems are involved, and what the expected behaviour was.
• Helpful: error logs, webhook payloads, schema samples. These speed the audit up and are optional.
• Helpful: the spreadsheet someone keeps because they stopped trusting the CRM. It is a confession and a map at the same time.

Evidence pack checklist

• One-paragraph problem statement: what breaks, for whom, and what it costs in rework or missed follow-up.
• Completed inventory: Table 1A and Table 1C filled, owners named.
• Ticked failure signals: the signal log under Section 3 with evidence references.
• Ranked fix list: Table 3C with tiers and the reasoning behind each score.
• Triage lanes: quick fix or full rebuild, decided per candidate.
• Raw samples: one payload per source, one error log excerpt, one bad record screenshot.
• Event contract target: the shape you want every source to send (below).

Event contract to validate against

Use this shape as the target when you are checking whether a payload is complete. Tally can feed it, Typeform can feed it, a manual import can feed it. The system receives the same clean structure every time.

What happens next. Run the free workflow diagnostic, or read the Meridian Legal Group case study to see what a rebuild looks like in production, then decide. When the template is not enough, the same systems ship as client engagements. The work covers workflow repair, lead qualification, document pipelines and agentic implementation.

Everything above is yours to read. The workbook adds the empty worksheets, so you fill them in on your own stack with your own owners, fields and failures. It opens in any browser, on any device, and it saves as you type.

When you want my read on your own stack, the workflow automation questionnaire takes ten minutes, and I reply with the first fix I would make and what it costs.