Troubleshooting — Symptom to Cause

Every way a campaign visibly goes wrong, what it actually means, and the shortest path to a fix — in the order you should check them.

Jump to module

Lesson 1: The Thirty-Second Diagnosis

Before anything else, read three things on the campaign page, in this order:

  1. The phase badge — what it is doing right now.
  2. The last error, if one is shown — why it is not doing anything.
  3. The outcome line — what it has achieved so far.

Those three answer most questions without further investigation. The phase badge in particular is doing more work than it looks like:

PhaseIt meansYour move
ScheduledNext generation is on the books, with a dateNothing. Wait.
GeneratingA run is in flight right nowNothing. Wait.
Needs accountsArmed with nowhere to publishConnect or fix an account
Awaiting approvalContent exists, waiting on a personOpen the Approve queue
PostingContent exists, window openNothing
DoneWindow closedRead the outcome
No badgeDraft or pausedActivate or resume it

A campaign with no phase badge at all is a draft or is paused. That is the single most common "it isn't doing anything" — it was never armed.

Lesson 2: Nothing Is Being Generated

Work down this list. Each step rules out the ones below it.

1. Is it active? Only active campaigns generate. A scheduled campaign whose window has not opened will not produce content — it must not produce content for a window you have not reached. A draft never dispatches at all.

2. Does it have a start date? A campaign with no start date owes no generation. It has nothing to count weeks from. This is legal — a campaign can be named before it is dated — but it will sit there indefinitely.

3. Does it have usable accounts? The phase says needs accounts when the campaign names no account, or none of the ones it names still holds a valid connection. Nothing can be generated or published in this state.

Fix the connection and the campaign proceeds. Adding a valid account to an active campaign re-arms it — you do not need to deactivate and reactivate.

4. Is your content-creator agent paused? A campaign runs through an agent. If your only content-creator agent is paused, the run refuses cleanly and names the agent. Resume it. Bolta deliberately will not hire a second agent around your pause.

5. Has the week already been generated? Bolta records which week it last produced. A second attempt at the same week finds the work done and stops. That is not a failure — check the timeline for the content it already made.

6. Has the window closed? Once the window closes, nothing more attaches. If you want more, duplicate.

7. Read the last error. If none of the above applies, the campaign will have recorded why. It carries when it happened, a plain message, which producer was involved and which week it was trying to produce. That message is the answer.

Lesson 3: Content Exists but Nothing Publishes

This is the most common shape of a disappointing campaign, and the reason is almost never technical.

Check the empty reason first. The campaign derives why it published nothing and reports the earliest provable blocker (Module 12). The top three, in the order they actually occur:

"Awaiting approval" — the modal answer. The drafts are fine. Nobody decided on them. Campaign content always waits for a person, on every autonomy setting, so this is not a misconfiguration and there is no switch that changes it. Open the queue.

"Nothing approved" — drafts were seen and none passed. Different from the above, and more interesting: somebody read them and said no. If that is one week, edit the drafts. If it is two weeks running, the angle or the context is wrong — see Lesson 4.

"Publishing failed" — the attempt was made and rejected. Usually a dead token. Reconnect the account. Platforms expire connections quietly and this is the failure that looks most like a Bolta bug and is least often one.

Then the less common ones:

  • "Quality blocked" — drafts exist and the quality gate held them. Review and edit them by hand.
  • "Not sent yet" — approved and queued; the slot has not arrived. Check the scheduled time.
  • "No accounts connected" — nothing could ever have published.
  • "Runs failed" — every run attributed to the campaign failed, either to start or partway through. Read the last error.
  • "Nothing generated" — runs dispatched and produced nothing. Check the plan and the last error.

One number that confuses people

A campaign can look like it published nothing when it published partially. That is guarded: publishing is counted from per-account records, so a post that went out on one account and failed on another counts as published on the one that worked. If you have a dead token on one platform, your campaign numbers are still honest about the others.

Lesson 4: The Content Is Wrong

Not broken — wrong. Different problem, different fixes.

"It's generic. It could be any business."

Almost always a context problem, not a prompt problem. Bolta writes more generally when it knows less, and it will not invent facts to fill the gap.

In order of leverage:

  1. Fill in Business DNA. Single biggest improvement available.
  2. Write real context notes. The specific thing, in a sentence. They reach every week, not just the first.
  3. Attach a source. The landing page, the menu, the event listing.
  4. Check your voice profile. A thin voice produces a thin register.

"It doesn't sound like me."

That is voice, not campaigns. Check which voice profile the campaign names — it may be inheriting a workspace default you have not looked at in months. Then work on the voice itself; the Voice DNA course covers it.

"The angle is wrong."

Read week one's drafts as a set, not individually. Signals you picked wrong:

  • Problem/Solution producing vague pain nobody recognises → you have no material about actual customer problems. Try Product Education.
  • Thought Leadership producing confident generalities → it has nothing you have genuinely done to ground a position in. Give it material or switch.
  • Offer & Promotion producing posts with no actual offer → there is no offer in the context. Put one there or change angle.
  • Quick Tips producing listicles → the instruction says one tip per post; if you are seeing "7 ways", check your context notes are not asking for roundups.

The angle is a per-campaign choice, so switching is cheap. What is not cheap is running four weeks on the wrong one.

"It planned a format it can't make."

Check the picker. Formats Bolta cannot produce are shown greyed with the reason, and the mix drops keys it does not recognise. Stories and avatar videos are not produced at all; blog posts need a connected site; emails are written but never sent.

"It's making too much."

The mix is per week. Halve the mix, not the duration.

What to take away

  • Phase badge, last error, outcome line — in that order, thirty seconds.
  • Not generating? Active → dated → accounts → agent → week done → window closed → last error.
  • Not publishing? It is nearly always the Approve queue.
  • Wrong content is usually thin context, not a bad model. Business DNA first.