Act, Park, and Ignore: what each button does per row kind

What this covers

HQ Decisions and Action Plans both read the same five kinds of network-level rows: decision plans, interventions, consent requests, program audits, and program violations. Every row kind has its own lifecycle in its own core table, and the three buttons on a tile, Act, Park, and Ignore, mean something different depending on which kind you're looking at and what state it's currently in. This article is the row-by-row reference: what each button does, what it writes, and why a button is sometimes greyed out and unclickable even though it's still sitting there on the card.

For the click-to-database mechanics (the admin gate, what happens on a failed write, which table each mutation lands in), see the companion piece, Act, Park, and Ignore per source kind. This article stays at the surface: what you see on the tile, in which state, and what happens when you click.

Nothing here ever reads a franchisee's private business data. These five row kinds are all records HQ itself created or received: a plan HQ drafted, an intervention HQ opened, a consent ask HQ sent, or a program check HQ ran. Franchisees own their operating numbers; what you're clicking through here is HQ's own decision trail.

Where to find it

Both Decisions and Action Plans sit in the HQ sidebar, at hq.verinode.ai/decisions and hq.verinode.ai/actions. They read the identical merged row set, but only one of them actually shows the buttons this article is about:

• Decisions opens as a gallery of tiles (the same finding-tile cards used everywhere in Verinode). This is the only place the Act / Park / Ignore button row lives.
• Action Plans opens straight into a Gantt timeline (or Calendar, one click over), and there's no toggle back to a card gallery on this page, it's plan-execution mode, not triage. Clicking a row or a step block here opens the same detail slider Decisions uses, but that slider is read-only for HQ rows: it shows the record and lets you open the agent chat panel to ask about it, it never carries an Act, Park, or Ignore button of its own.

Practically: do your triaging, mutating clicks on Decisions. Action Plans is where you watch what's already moving, not where you change its state.

The three buttons, and the one that isn't a button

Every open (non-terminal) tile can show up to three buttons along its bottom edge, always in the same left-to-right order:

• Act, filled copper. The affirmative move: start something, approve something, resolve something, depending on the kind.
• Not now. This is the on-screen label for what the underlying system calls Park everywhere else, in the code, in server logs, in the companion article. The button reads "Not now," not "Park." Once a row is actually parked, this button disappears and is replaced by a single "Resume" button instead (which reactivates the row, it's wired to the same handler as Act).
• Ignore, bordered, secondary. Dismisses the row: archives a plan, closes an intervention with no action taken, withdraws a consent request, disputes an audit, or waives a violation, depending on the kind.

A fourth thing, Discuss, is not a button on the tile. Opening a row (clicking anywhere on its body other than one of the three buttons) opens the glass detail slider and, at the same time, hands that row's context to the AI agent panel so you can ask Verinode questions about it in chat. Nothing about the row's state changes from opening it. Think of Discuss as "read and ask," always available, and Act / Park / Ignore as the only three moves that actually write anything.

Buttons render even when they don't apply. When a kind doesn't support Park (interventions, consent requests, program audits) or Act (consent requests), the button still shows up in its slot, greyed out, disabled, and non-clickable, rather than disappearing from the row. If your cursor doesn't do anything on a button, that's not a bug, it means that action was never wired for that row's current kind or status. The per-kind sections below spell out exactly when each button is live and when it's dead weight.

The rule that governs everything below: once a row is Acted, the button row disappears

Before the per-kind detail, one mechanic explains a lot of what looks like inconsistent behavior at first glance. The tile's button row is driven by the row's mapped status, not by which of the five kinds it is:

• If the row is currently Acted (green accent, "Acted" stamp) or Resolved ("Done" stamp), the three-button row is replaced entirely by a single primary button: Edit plan on an Acted row, Review outcome on a Resolved one. Clicking that button, or clicking anywhere else on the tile, takes you off the Decisions/Action Plans board and opens the record at /decisions/<id>, the same URL IQ uses for an individual operator's own decision workspace.
• This swap happens regardless of what Park or Ignore were wired to do for that kind. If a row's mapped status is Acted, Park and Ignore are not reachable from the tile at all, whatever the per-kind table below says they'd otherwise do.

This matters most for interventions and program violations, because both of them reach the Acted state after a single click:

• An intervention moves from queued straight to Acted the moment you click Act once (queued → contacted collapses into the same "Acted" bucket the shell tracks). From that point, the tile shows one "Edit plan" button; there's no further Act/Ignore click available on the card itself to advance it to in-progress or resolved.
• A violation moves to Acted the moment you Park it (escalate it). Escalating a violation is meant to be a middle step before resolving or waiving it, but the tile collapses to "Edit plan" as soon as it's escalated, so Act (resolve) and Ignore (waive) stop being reachable from the card at that point too.

Decision plans and consent requests behave the same way once they reach Acted (an active plan, an approved consent request); program audits are the one kind that never reaches this state in the log at all, more on that below.

Decision plans: activate, complete, pause, archive

A decision plan is a network-wide playbook, directive, recommendation, or experiment HQ has drafted. Its lifecycle runs draft → active → completed, with paused and archived as side branches, and it stays visible in the log for roughly six months after it's created regardless of where it ends up.

• Act is the one button that means two different things depending on where the plan sits: on a draft or a paused plan, Act activates it. On an already-active plan, Act completes it. Once the plan is completed or archived, the row has already left the three-button state (see the collapse rule above), so Act isn't offered there.
• Not now (Park) only works on an active plan, it pauses it. On a draft plan, the button is present but disabled, there's nothing yet to pause. Once paused, the three-button row is replaced by a single "Resume" button, which reactivates the plan (it calls the same handler as Act).
• Ignore archives the plan from any status, draft, active, or paused. This is the one button on this kind that's always live no matter what state the plan is in.

Interventions: one click of Act, then it's out of your hands on the tile

An intervention is a specific franchisee location HQ has flagged for follow-up. Its lifecycle runs queued → contacted → in_progress → resolved, with no_action as a side branch.

• Act on a queued intervention moves it to contacted, and that single click is also enough to flip the row's mapped status to Acted, collapsing the button row (see above). There's no Not-now button on interventions at any point, buildup for Park was never wired for this kind, so that slot always renders greyed out.
• Ignore closes the intervention as no action taken, from any status, and is the only button besides Act that's ever live here.
• Once you've clicked Act once, continuing the intervention to in_progress or resolved happens by opening the card (Edit plan), not from a second click on the tile.

The franchisee's location name on an intervention tile is anonymized (something like a masked location label) unless your network is configured as a single legal entity. That's the same privacy boundary the rest of HQ holds to: you see that a location was flagged and what for, not a window into that location's own books.

Consent requests: HQ can only pull back its own ask

A consent request is HQ's own ask of a franchisee, most often permission to unlock a name-attached benchmark view. Its lifecycle runs pending → approved or declined (the franchisee's call, made on their side of the platform), with withdrawn and expired as side branches on HQ's side.

• Act never does anything on this kind, at any status. The button always renders disabled. HQ cannot approve or decline its own request; only the franchisee can, from their own account.
• Not now (Park) is also never wired for this kind. Always disabled, in every status, including once the request has expired, where the three-button row would otherwise collapse to a single "Resume" button, disabled, since there's no Act handler behind it to resume it.
• Ignore is the only live button, and it withdraws HQ's own request. It only works while the request is pending or already approved; once it's declined, withdrawn, or expired, the row has moved past the states this button can act on.

Ignore is genuinely the only lever HQ has on this kind. There is no Act, and no Park, ever, only a single working button whose whole job is "take back the ask."

Program audits: the one kind that never leaves the open state

A program audit is a check run against a party or franchisee on one of HQ's programs. The board only ever loads audits that are currently submitted or disputed, an approved audit drops off the Decisions/Action Plans row set entirely on the next load, since the underlying query stops matching it.

• Act approves the audit. This is live whether the audit arrived freshly submitted or after you've already disputed it once.
• Not now (Park) is never wired for audits. Always disabled.
• Ignore disputes the audit. Also live in both submitted and disputed state, since (unlike every other kind on this page) the log doesn't actually distinguish "submitted" from "disputed" in how the tile looks, both render with the exact same Pending look and the exact same two live buttons.

That last point is worth calling out on its own: a disputed audit's tile looks identical to a freshly-submitted one. There's no visible marker on the card telling you "I already disputed this once." Open the record if you need to check whether a dispute is already on file before clicking Ignore a second time. Program audits are also the one row kind that never reaches the Acted collapse described above, because they never carry a mapped status other than Pending in this log; the only way an audit leaves the board is by being approved (which removes it) or by aging out of the window HQ's programs feature tracks separately.

Program violations: resolve, escalate, or waive, but only before you escalate

A program violation is an unresolved problem a check turned up against one of HQ's program rules, tracked separately from an audit and from its own severity rating (low, medium, high, or critical). Its escalation state runs open → escalated → resolved or waived.

• Act resolves the violation. Live from open and, in principle, from escalated too, though once a violation is escalated the tile has already collapsed to a single "Edit plan" button (see the collapse rule above), so in practice you're resolving it from that card, not from a live Act button.
• Not now (Park) escalates the violation, and only works while it's still open. The instant you click it, the row's mapped status becomes Acted and the three-button row disappears in favor of "Edit plan," so there's no clicking Park twice on the same violation.
• Ignore waives the violation: same close-out as Act, but the record keeps "waived" distinct from "resolved" so you can tell later whether a violation was fixed or let go. Also live from open, and in principle from escalated, with the same caveat as Act above.

Practically: on an open violation, all three buttons are live, resolve it outright with Act, waive it outright with Ignore, or escalate it first with Not now if it needs more visibility before you close it out. Once you escalate, the card's next move happens through Edit plan, not through a second round of these three buttons.

Quick reference

| Row kind | Act | Not now (Park) | Ignore | |---|---|---|---| | Decision plan | Activates a draft/paused plan; completes an active one | Pauses an active plan only | Archives from any status | | Intervention | Advances queued → contacted (one click, then the tile collapses) | Never wired, always disabled | Closes as no action taken, any status | | Consent request | Never wired, always disabled | Never wired, always disabled | Withdraws HQ's own ask, pending or approved only | | Program audit | Approves, submitted or disputed | Never wired, always disabled | Disputes, submitted or disputed | | Program violation | Resolves, open (or escalated, via Edit plan) | Escalates, open only, then the tile collapses | Waives, open (or escalated, via Edit plan) |

Empty states, verbatim

On Decisions, with no filter narrowing the board and nothing on it yet: "No decisions for you yet. They'll land here as Verinode spots cost savings, risk, and growth opportunities in your data." With a status filter chosen and nothing matching, the message narrows to it, for example "Nothing parked for you right now."

On Action Plans, the Gantt and Calendar empty states are written per filter bucket rather than as one generic line. A few examples: on Overdue, "Nothing overdue. Steps past their due date land here so they don't slip." On Ready to start, "Nothing ready to start. Plans whose first step has come due will land here." On Completed, "No completed plans yet. Plans you resolve land here so you can run a retrospective." If nothing at all matches the current filter, it reads simply "Nothing matches this filter." If the filter has real rows in it but none of them carry a synthesized timeline (the program-audit case from the callout above), it reads "Nothing in this filter right now."

Related reading

• HQ Decisions: the franchisor inbox: search, status counts, the business-area filter, and how the whole board is laid out.
• Action Plans vs Decisions: how the two HQ surfaces differ: the full Gantt / Calendar / Select-and-cancel walkthrough for the Action Plans page.
• Act, Park, and Ignore per source kind: the click-to-database mechanics, the admin gate, and what gets written on a successful mutation.
• The four HQ decision sources and how they map into one log: how each source's native status becomes the shared Pending/Acted/Parked/Ignored/Resolved set.
• The interventions queue for at-risk locations: how a franchisee location gets flagged in the first place.
• Audits and violations: enforcing a program: how a program audit or violation is created and what it means for the program.
• Benchmark consent requests: what a consent request unlocks once a franchisee approves it.