Back to all articlesCustomer onboarding

Diagnose delayed onboarding before choosing a recovery action

Alactic Editorial9 min read

Prepared with AI assistance by Alactic Editorial. Examples identified as hypothetical are illustrative, not customer results.

A customer has received another reminder to finish onboarding, but the implementation is still stuck. The task says “customer action required.” Nobody can explain whether the customer lacks a file, cannot approve the proposed setup, or is waiting for a defect to be fixed.

Before sending another reminder, identify the controlling cause: the condition currently preventing the next agreed onboarding outcome from being accepted. Establish that outcome, inspect evidence at the blocked step, and ask what would happen if the suspected blocker disappeared.

This approach diagnoses active onboarding delays. It does not replace technical troubleshooting or detailed recovery scheduling. Its purpose is to identify the intervention justified by the evidence and define how to tell whether it worked.

1. Start with the stalled outcome, not the overdue task

An overdue task is a useful signal, but a poor diagnosis. “Upload data,” “approve configuration,” and “complete training” describe activities. They do not necessarily explain what prevents onboarding from advancing.

Write the blocked outcome in a form both parties can inspect:

  • Outcome: What must be usable, demonstrated, or accepted?
  • Acceptance evidence: What observable result would show completion?
  • Current gap: What specifically differs from that result?
  • Next dependency: What cannot proceed because of this gap?

In a hypothetical implementation, “customer has not uploaded the account file” might become: “The onboarding team cannot validate account ownership because the supplied file lacks the agreed owner identifiers.” That narrows the investigation without yet proving whether the cause is missing input, an unclear requirement, or a failed export.

If nobody can agree on what a usable file contains, stop treating the file as late. The immediate problem is an unresolved acceptance decision. Asking for faster delivery of an undefined artifact will not resolve it.

2. Separate symptoms, blockers, and underlying causes

  • Symptom: An observed sign of delay, such as an unanswered message or failed acceptance check.
  • Blocker: The immediate condition preventing the next outcome, such as absent owner identifiers.
  • Underlying cause: The reason that condition persists, such as no agreement about which ownership rule applies.

The useful stopping point is not necessarily the deepest explanation. It is the cause you can support with evidence and act on within the onboarding effort.

Ask: “If we removed this condition now, could the next agreed outcome proceed?” If not, identify the remaining dependency before declaring a single cause.

For example, customer silence may reflect an unavailable owner. Finding a replacement will not unblock work if that person lacks authority to choose between conflicting requirements. Conversely, a decision may already be recorded, leaving only an absent person who can supply the input.

Where several blockers coexist, record them separately. Identify what must be resolved first, what can proceed independently, and which suspected blockers remain unverified. A useful diagnosis explains the dependency, not the customer's motivation.

3. Follow a decision tree without guessing at unknowns

All cases and artifacts below are fictional, hypothetical examples, not customer results. This filled tree illustrates four possible branches. Its order checks definition and ownership before classifying an absent artifact or failed implementation.

Shared uncertainty rule: At every question, an unanswered or unverified condition is UNKNOWN, not NO. Mark the diagnosis provisional and gather the smallest piece of evidence that distinguishes the plausible branches. This applies equally to acceptance rules, owner authority, input availability, and implementation checks.

START: Name the blocked outcome and acceptance evidence.

1. Is the outcome or acceptance rule unresolved?
   YES → B: Unresolved decision.
   Evidence: Two ownership rules; no authorised choice recorded.
   Action: Establish the decision owner if unassigned, then
   obtain and record that owner's decision.
   Release: A selected rule is usable by the implementation team.
   NO → Question 2.

2. Is an accountable, available owner able to perform
   the specific authorised next action?
   NO → D: Unavailable or unable customer owner.
   Evidence: Named owner absent; substitute cannot release data.
   Action: Secure a capable, authorised delegate or agree a pause.
   Release: Delegate has demonstrated authority, required access,
   and capacity to perform the specific next action.
   Pause: Acknowledged unresolved dependency, not a release.
   YES → Question 3.

3. Is a required, defined input absent or unusable?
   YES → A: Missing input.
   Evidence: File omits required owner IDs; supplier confirms
   the IDs are available and can be supplied.
   Action: Request those fields, not another generic upload.
   Release: Replacement file passes agreed input checks.
   NO → Question 4.

4. Does the implementation fail an agreed check?
   YES → C: Implementation defect.
   Evidence: Valid test input produces duplicate assignments.
   Action: Give implementation owner a bounded defect record.
   Release: Failed check passes; affected work is revalidated.
   NO → Recheck the dependency map and stated blocker.

AT ANY QUESTION: UNKNOWN → Provisional diagnosis.
Gather discriminating evidence before selecting recovery.

This is a reasoning aid, not a mandatory investigation sequence. A confirmed defect may be obvious immediately. The discipline is to avoid classifying an issue from assumptions about requirements, ownership, or available inputs.

4. Match the response to the diagnosed condition

Case A: A defined input is genuinely missing

In this hypothetical case, the account file lacks required owner identifiers. The specification is agreed, the identifiers exist, and an available person can supply them.

Response: Request the missing fields, explain why validation needs them, and specify the checks the replacement must pass. Another broad request to “finish the upload” discards information already learned.

Failure case: The customer discovers the identifiers do not exist in the source system. Reclassify the issue: an alternative identifier or changed acceptance rule now needs a decision. Repeating the file request cannot resolve that choice.

Case B: No authorised decision has been recorded

In this hypothetical case, sales and operations prefer different account ownership rules. The configuration team could implement either, but no authorised choice has been recorded. That does not necessarily mean nobody has authority.

Response: Check whether an accountable approver is already assigned. If decision authority is unassigned, establish the decision owner through the project's agreed governance first. Then present the options, their consequences for the agreed outcome, and the choice required. Obtain a recorded decision.

Failure case: A responsive contact says “go ahead” but cannot approve the choice. Responsiveness is not authority. Confirm who can decide rather than interpreting informal feedback as acceptance.

Case C: The implementation fails an agreed check

In this hypothetical case, the inputs and ownership rule are accepted, but a reproducible test creates duplicate assignments.

Response: Give the implementation owner a bounded defect record: expected result, observed result, relevant setup, and reproducible evidence. Include only the customer information needed to investigate.

Failure case: The requested behaviour falls outside the agreed scope. Return to a scope decision instead of circulating the issue as a technical defect. Diagnosis identifies the workstream; it does not prescribe an integration fix.

Case D: The responsible owner cannot act

In this hypothetical case, the named owner is absent and the proposed substitute cannot authorise release of the required data. The input requirement itself is clear.

Response: Ask the customer sponsor to establish a delegate who can perform the specific next action. Confirm the delegate's authority, required access, and capacity. Evidence might include the sponsor's delegation, confirmation that necessary access works, and an accepted commitment to carry out the action.

A delegate accepting responsibility is insufficient if they still cannot obtain the file or perform the permitted release. Do not request broader access than the action requires.

Alternative disposition: An agreed pause makes the unresolved dependency explicit; it does not remove it. Record who will revisit it and what condition would allow work to restart.

5. Gather evidence that changes the next action

A diagnostic conversation should reduce uncertainty, not retell the project history. Ask questions whose answers would change the recovery choice:

  • “Show the last attempted step and its result.” Distinguishes observed failure from work not yet attempted.
  • “Which agreed acceptance rule does this fail?” Separates a defect from an undefined expectation.
  • “Does the required input exist, and who can supply it?” Distinguishes missing material from missing ownership.
  • “Who can authorise this choice?” Separates an undecided question from unassigned authority.
  • “If this arrived today, what would still prevent progress?” Exposes stacked dependencies.

Prefer a relevant artifact over a confident label: an incomplete file, a failed check, a decision record, or confirmation of delegation. If two explanations remain plausible, choose evidence that would separate them. For example, checking whether the source contains the required identifiers distinguishes an omitted export field from an unavailable input.

When evidence cannot yet be obtained, name the gap: “Possible defect; valid-input test not yet observed.” Assign someone to obtain that evidence and someone to evaluate it. This is more actionable than “technical blocker” without supporting detail.

6. Keep a diagnostic record another owner can challenge

The following filled artifact continues the hypothetical ownership-rule example. Reuse its field structure in an existing project or customer record. It is a recommended operating design, not a claim about a software feature.

Blocked outcome
Validate account assignments against the agreed ownership rule.
Observed symptom
The submitted account file has not been accepted.
Acceptance evidence
The implementation owner can evaluate assignments using one authorised ownership rule.
Evidence inspected
The file contains the required fields. The decision record still lists two competing rules without a selected option.
Current diagnosis and confidence
Case B: unresolved decision. Supported by the inspected file and decision record; no implementation defect established.
Decision authority
The customer sponsor is the designated approver under the hypothetical project's governance.
Counterfactual
Replacing the file would not resolve which rule validation should use.
Recovery action and owner
The onboarding lead will present both options and their acceptance consequences to the sponsor for a recorded choice.
Release evidence and evaluator
The implementation owner will confirm that the sponsor's selected rule is specific enough to configure and validate.
Next dependent action
Run assignment validation using the selected rule and the existing file.
Remaining uncertainty
Validation may reveal a separate implementation defect once the rule is settled.
Current disposition
Blocked pending decision; not waiting for a replacement upload.

The worked reasoning matters: the file's presence alone does not prove it is usable, so its fields were checked. With that input uncertainty reduced, the unresolved rule explains why validation cannot proceed. The next action therefore changes from requesting another file to obtaining a decision.

Keep this record factual and revisable. “Awaiting customer” hides the missing condition; “customer is disengaged” assumes motivation. Neither tells the next owner what to change.

7. Verify the action without declaring the whole onboarding recovered

At the next review, use three observable checks:

  1. Was uncertainty reduced? Evidence supports the diagnosis, or the missing evidence is explicitly identified.
  2. Did the blocking condition change? The input became usable, the decision was authorised, the failed check passed, or the delegate demonstrated the ability to act. An agreed pause remains an unresolved dependency.
  3. Could dependent work resume? The next agreed step was attempted, not merely relabelled as unblocked.

Clearing one blocker does not recover the whole onboarding. If an authorised rule enables validation and validation then reveals a defect, that is a newly exposed blocker. If the replacement file still lacks identifiers that never existed, the original missing-input diagnosis was incomplete or wrong. Record the distinction and choose the next action from the new evidence rather than repeating an ineffective intervention.

Customer onboardingImplementation managementRoot cause analysisCustomer operations

Put the next conversation to work.

Explore Alactic Communications for WhatsApp conversations, campaigns and AI-assisted workflows. Choose the product and plan that fit your team.

Explore Communications