What is switched on and broken, or reaching nobody?
The Automations analysis lists every workflow, smart campaign, automation, journey or flow your tool reports, and sorts each one into a verdict you can act on: broken, reaching nobody, holding people, or running fine. Every verdict rests on a count, a date or an error your tool itself reports, so you can check each one in the tool before you touch anything. If you want the whole Analysis tab rather than this one card, read Read your first analysis first.
Before you start
Plan required
The card itself runs no plan check. You reach it through Insights, which Starter does not include, so a Starter workspace cannot open it. Free Trial, Pro, Business, Enterprise and Partner all can.
- HubSpot, Marketo, Mailchimp, Salesforce Marketing Cloud or Salesforce, connected.
- On HubSpot, the private app needs the automation scope. Without it the card says HubSpot refused the read and names the scope.
- On Marketing Cloud, the installed package needs Journeys: Read.
What each tool calls them
HubSpot has workflows, Marketo has smart campaigns, Mailchimp has automations, Marketing Cloud has journeys and Salesforce has flows. The card uses your tool’s word throughout, including in the counts and the headline.
Open the analysis
Open Insights in the left-hand menu, then Analysis.
Pick the tool along the top.
Choose Automations.
The card reads the list as your tool reported it. The line under the heading gives the date of that reading.
Click Read the latest from <Tool> when you have changed something in the tool.
A reading is kept for an hour, so within that hour the card answers from the last one rather than calling the tool again.
This card ignores the date range at the top of the page
Every other analysis on the page is scoped by the created date of your records. This one is not, and when a range is set the card says so in its own words: “Automations are not records, so the created-date range does not apply here: every one is shown.”
The reason is that an automation is judged on what it is doing now, not on when it was made. A workflow created four years ago and enrolling people this morning is a live workflow, and cutting it out of a quarterly view would hide the one most worth reading.
The four numbers at the top
The headline sentence gives the total and how many are switched on. Under it sit four tiles. Each counts automations, never people.
| Tile | What it counts |
|---|---|
| Switched on | Automations your tool reports as enabled. The line underneath reads “acting on people right now” |
| Need a person | Automations that landed on one of the five fault verdicts below. These are listed in full further down, most urgent first |
| Switched off | Everything not switched on, with the number of those that were never used |
| Left with a default name | Names that still read “New…”, “Untitled…”, “Unnamed…”, “Copy of…”, “Test…”, “Clone…” or “Workflow 2”. Not a fault; a reason two people switch on the wrong one |
Note
Nothing on this card changes anything in your tool. It reads, judges and lists. Every fix is a change you make in the tool yourself.
The nine verdicts
Every automation gets exactly one verdict. The first five are faults and are listed under The <automations> that need a person, most urgent first in this order.
| Verdict | What it means | The test |
|---|---|---|
| On, with errors | Switched on, and your tool itself reports something wrong inside it | Your tool returned at least one validation error. The card names the step number, the step and the error in the tool’s own words |
| Off, with people still inside | Switched off while people were part-way through it | Switched off, and the tool reports one or more people still active inside it |
| On, reaching nobody | Switched on and taking nobody in | Either your tool reports nobody has ever entered it and it is at least 14 days old (or carries no creation date), or people entered it in the past and none in the last 25 days or more |
| On, goal never reached | It has a goal and not one person has met it | It has a goal, 30 or more people have entered it, and the goal count is zero |
| On, with no goal to judge it by | Running, with nothing to say whether it works | It has no goal and 30 or more people have entered it |
| On, worth a re-check | Running, still taking people in, last edited a long time ago | Last edited 18 months ago or more. Listed apart from the faults, never counted as one |
| On and running | Nothing to report | None of the above applies |
| Off, never used | Switched off and never ran | Switched off, never run or nobody ever entered it, and more than 90 days old |
| Off | Switched off, and it did run | Anything else that is switched off |
One verdict each, and which test wins
The tests run in the order of the table, with one exception: On, worth a re-check is decided before On, with no goal to judge it by. An automation last edited two years ago that also has no goal reads as worth a re-check, so it sits in the re-check card rather than among the faults.
Age is not a fault
On, worth a re-check sits in its own card headed Worth a re-check, not a fault. Nothing about an 18-month-old workflow says it is wrong. It is there because it is still acting on people using decisions taken then, which is worth a read when your offer, pricing or team has changed since.
How “reaching nobody” sharpens over time
Tools report a lifetime count: how many people have ever entered. A lifetime count on its own cannot tell you whether anyone entered last month. So MarqIQ saves a dated reading each time you open this card, and compares the newest reading against the newest saved one that is at least 25 days old.
- On your first visit there is no earlier reading, so only the “nobody has ever entered it” test can fire.
- After 25 days there are two points, and a switched-on automation that has stopped taking anyone in becomes visible.
- A lifetime count that has fallen gives no reading at all. People get deleted and merged, and a fall says nothing about who entered since.
- Up to 120 readings are kept per automation.
Switched on, and starting from the same thing
Two switched-on automations whose starting conditions are identical are grouped together in their own card. Identical, not similar: the comparison is on the exact starting conditions your tool reports, so a near-miss is never grouped.
That matters because the same person enters both at the same moment. That is how someone receives two emails within a minute, and how one field gets written twice with two different values in the same second.
The card names the shared trigger in your tool’s own terms. On HubSpot a single form or a single list is named — “the “Request a demo” form”, “the same list” — and anything else reads “exactly the same starting conditions”. On Mailchimp it names the trigger and the audience, such as “someone joining the audience in “Newsletter””. On Marketing Cloud it names the entry source.
Note
This card exists only where your tool reports starting conditions: HubSpot, Mailchimp and Marketing Cloud. Marketo and Salesforce do not return them in the form this needs, so on those two the card is absent rather than empty.
Do people reach the goal?
A separate card lists the switched-on automations that have a goal, with Entered, Reached the goal and Share. The share is goal-reachers out of everyone who has ever entered, which is the denominator your tool keeps.
Two automations are never ranked against each other here, and the card says so. A welcome sequence whose goal is a reply and a renewal sequence whose goal is a payment are not comparable, and a league table of the two would be a number that means nothing.
What each tool reports, and what it cannot
A verdict can only rest on what your tool returns. Where a tool reports no counts, the count-based verdicts cannot fire on it, and the card says which is which.
| Tool | What is read | What is not there |
|---|---|---|
| HubSpot | Every contact workflow with its enrolled and active counts, plus workflows on companies, deals and tickets by name and state. Steps, goals, starting conditions and HubSpot’s own validation errors are read for the 150 busiest — switched on first | Steps, goals and errors for anything past the 150 busiest. The card says so when it applies |
| Marketo | Every smart campaign except Marketo’s own system campaigns: state, type, dates and a link, and the invalid status as a reported problem | How many people a campaign has processed. Marketo does not report it here, so no count is shown and no count-based verdict fires. The first 4,000 are read |
| Mailchimp | Classic automations, up to 1,000, with the audience each sends to and whether anything has ever been sent | How many people entered. Mailchimp reports emails sent instead, so On, reaching nobody fires only where nothing has ever been sent. Journeys built in the newer journey builder are not in Mailchimp’s API and are not listed |
| Marketing Cloud | Journey Builder journeys, with the cumulative and current populations, in pages of 50 up to 500 | Automation Studio — the scheduled data and file jobs. These are not read; the journeys are what act on people |
| Salesforce | Flows: whether each is switched on, whether the running version is the newest, how it starts and who last changed it | How many records a flow has run on. Salesforce does not report it, so a flow is judged on the rest |
How far this has been proved
The HubSpot and Marketo readers have run against live accounts. Mailchimp, Marketing Cloud and Salesforce follow those vendors’ documented APIs and are tested against payloads of that shape, but have not yet run on a customer account. Read their output against the tool before you act on it.
What this analysis does not check
Knowing the edges saves you looking for something that is not here.
- It names the kind of each step — Send an email, Wait, Set a contact property — but never the field a step writes or the value it writes there. A workflow overwriting your first-touch source is found by the Field discipline analysis, which reads the change history on the record and names the writer.
- It does not judge whether an automation has an exit, because no tool in this list reports that in a form that could be tested.
- It never compares two automations’ performance. Their goals differ, so the comparison would have no meaning.
- Automations the tool created for itself are left out of the clutter counts and marked as not yours to tidy.
- Nothing is sampled. Every automation the tool lists is read, within the per-tool limits in the table above.
What to do with it
The What to do next card holds up to six actions, in order of how much each is likely to matter, and each one names the automation and the number behind it. Work down that list rather than down the full table.
- Anything On, with errors comes first: it is switched on, so every person entering it meets a step that does not work.
- Then decide what happens to the people inside an automation that was switched off part-way. Either switch it back on until they finish, or take them out so something else can pick them up.
- Then check what starts anything reaching nobody. A form that was replaced, a list that was emptied or a field that was renamed are the usual three, and each one means leads are being missed in silence.
- Then the two that start from the same thing, so you can check they do not send one person two emails or write two values to one field.
- Then goals: the busiest automation with no goal, or the goal nobody has ever reached.
- Archiving the never-used ones is last, and is tidiness rather than repair. The action appears once five or more are switched off and never used.
Note
The re-check prompt is added only when fewer than four of the above applied, so a card already full of faults does not push it off the list.
If it did not work
The Automations card says MarqIQ cannot read this tool’s automations yet
Why: The analysis reads five tools: HubSpot, Marketo, Mailchimp, Marketing Cloud and Salesforce. Every other connected tool gets that sentence.
Fix: Read the automations in the tool itself. Pipedrive, Zoho CRM, Dynamics 365 and Pardot all keep theirs behind an API MarqIQ does not read, and the Integration Audit Report names the screen to open in each.
“HubSpot refused: the private app needs the “automation” scope”
Why: The connection was made without that scope, so the workflow list cannot be requested at all.
Fix: Add the automation scope to the private app in HubSpot, then reconnect HubSpot from Data sources.
The card shows numbers but says they are the last successful reading
Why: The newest read did not finish — usually an expired token or a tool that answered with an error. Rather than show nothing, MarqIQ shows the last reading that did complete and dates it.
Fix: Read the reason in that same line, fix the connection, then click Read the latest from <Tool>.
Nothing is marked On, reaching nobody although you know a workflow has gone quiet
Why: That test needs either an automation with nobody ever in it and at least 14 days old, or two readings at least 25 days apart. On a workspace opening this card for the first time, the second test has no earlier reading to compare against.
Fix: Come back after 25 days, or check the enrolment count in the tool in the meantime. The card sharpens by itself as readings accumulate.
An automation shows no Entered count at all
Why: Marketo and Salesforce do not report how many people or records an automation has taken in, and Mailchimp reports emails sent instead.
Fix: Read the verdict rather than the blank. On those tools the card is built on state, errors and dates, and the table above says exactly what each one gives.
A workflow you know has errors is listed as On and running
Why: Errors, steps and goals are read for the 150 busiest workflows only, switched on first. Past that, HubSpot returns the name and the counts and nothing else.
Fix: Check the note under the table — it says when the detail read was capped — and open that workflow in HubSpot.
Next steps
Related articles
Was this article helpful?