Skip to content

Deciding what to do next

How the ranked "what to do next" list is built from your own records, what each field means, which of them are deliberately left blank, and why the work always happens somewhere else.

Updated

"What to do next" ranks the suggestions the Social Operator recorded for the most recent week it planned. It adds the things you need in order to decide: which objective each one affects, how urgent it is where that can be established, how much work it is, what it would cost, whether an external provider is involved, whether an approval is needed, and how much weight the evidence behind it can bear.

It suggests. It does not act. Every "Open" control on the page is a link into the workflow that already owns that work, and that workflow keeps its own permissions, its own approvals and its own record of what happened.

BEFORE YOU START

  • Access to the workspace, and active membership of the organisation. Reading the list needs nothing else.
  • The "administer" or "moderate" permission to accept, edit, dismiss, postpone or ask for something else. Deciding what your organisation is not going to do is a governance act even though it performs no work.
  • A planned week. The list reads what the Social Operator recorded; with no run there is nothing to rank, and the page says so.

STEPS

  1. Open "What to do next" for the workspace, or go to /app/workspaces/{workspace}/cmo.
  2. Read the note at the top. It states that the work happens where it always did.
  3. Read each recommendation in rank order. Rank 1 is what is blocking most.
  4. Read the fields underneath: objective, urgency, effort, confidence, approval, credits, providers, budget, the evidence, and what that evidence cannot tell you.
  5. Choose "Open" to go to the workflow that does the work, and do it there.
  6. Record what you decided: Accept, Postpone, Dismiss, or "Suggest something else".

WHAT YOU SHOULD SEE

Each recommendation carries a rank, a title, the reason it appears, and a field list. Some fields will say "Not known" followed by a sentence explaining why. That is the correct answer for most recommendations most of the time, and it is written out in words rather than shown as a dash or a zero.

WHAT THIS WILL NOT DO

  • It will not publish, schedule, send, reply, start an experiment or spend money. Accepting a recommendation records that you intend to do it; it does not do it.
  • It will not predict an uplift, a reach, a conversion, a revenue figure or a performance outcome. None of those is derivable from what this product records, so none is printed — not even as a zero.
  • It will not estimate a budget. There is no advertising integration in this build, so no budget floor, platform minimum or cost figure is readable from anywhere.
  • It will not generate a different suggestion when you ask for one. "Suggest something else" records that this one did not suit you. The Operator proposes only what your own records support, so there is no second suggestion waiting behind the first, and inventing one would be the opposite of what this page is for.
  • It will not overwrite what the Operator said when you rename a recommendation. Your wording is shown in front; the original stays readable underneath.
  • It will not require Autopilot. Every recommendation is readable and actionable by hand.

WORTH KNOWING

Urgency is only claimed where a recorded fact makes something time-critical: a connection in an error state stops publishing outright, and a planned week has an end date. Everything else has no derivable deadline, and the page says that rather than inventing a priority.

Confidence means one of three things and the page always says which. "Recorded fact" means the rows exist and the recommendation says only that they do — there is no sample and nothing to be confident about. "Established band" and "Provisional band" describe how large a measured sample was. Neither band is a statistical significance test, and none is implied.

Effort is a band over the number of records you have to work through, not a duration. Nothing in this product records how long any of this work takes, so no minute figure is shown.

Where a recommendation involves promoting a post, it states the comparison, your own baseline, the sample it was built from, and the fact that a single post cannot establish that anything about it caused the difference. It is kept even though this product cannot promote anything, because you can still promote it yourself — and you can only decide to if somebody tells you.

IF IT DOES NOT WORK

  • "No week has been planned for this workspace yet." means the Social Operator has not run. Plan a week on the Social Operator page first.
  • "Nothing to suggest for the week of ..." is a real result, not a failure. Nothing recorded supported a recommendation.
  • "administer or moderate permission required." means your role can read the list but not record decisions on it.
  • A postponement is refused without a date. "Later" with no date is a quiet dismissal, so a date is required and shown back to you.
  • Where a recommendation says the workflow it points at is not registered in this build, no link is offered rather than a broken one.

COMMON QUESTIONS

Is this different from the Social Operator page?

Yes. The Social Operator page plans a week and retries a week that failed. This page ranks what that week produced and records what you decided about each item. They read the same proposals.

Does accepting something make it happen?

No. It records your intention. The work is done in the workflow the recommendation points at, under that workflow's own approvals and its own audit trail.

Why is urgency blank on most of them?

Because nothing recorded says when they stop being worth doing. A priority the records cannot support would be a number with a costume on.

  • Planning the week with the Social Operator

    How the Social Operator gathers evidence from your own recorded activity, proposes at most six things to do this week, and why it never acts on its own.

  • Choosing how much the product does for you

    The three modes - manage it yourself, be told what to do, or have the plan built and run - what each one actually changes, and how to set a limit, pause everything or take over by hand.

  • What your recorded results support

    Reading the performance recommendations, the sample size every claim must clear, and the list of questions the recorded data cannot answer.

  • Reading your reporting

    What the workspace report totals, why a figure can say "Not reported" instead of zero, and the two ways a figure reaches it.