At-risk franchisee interventions in progress

What an intervention is

An intervention is HQ's tracked record of one franchisee that a human decided needed follow-up: margin sliding, a certification about to lapse, engagement gone quiet, a brand or safety concern. It is one of the four network-level sources that feed both Decisions and Action Plans, the other three being decision plans (playbooks), consent requests, and program audits and violations. Interventions are the ones tagged with the Risk business area.

Interventions are never auto-generated from a scoring model. An HQ admin opens one deliberately, from that franchisee's own record on the Network page, with a reason attached. This article covers what happens after that: the four-stage lifecycle an open intervention moves through, how to work it from Action Plans, and exactly how its notes are protected so only a signed-in HQ admin can read them. For the flagging step itself (the dialog, the reason dropdown, the duplicate-flag guard) see The interventions queue for at-risk locations.

Where to find it

A freshly flagged intervention (status Pending) shows up first on Decisions (hq.verinode.ai/decisions), the full franchisor inbox, alongside every other item HQ hasn't acted on yet. The moment you click Act on it, it becomes Acted and moves to Action Plans (hq.verinode.ai/actions), in the same top group of the HQ sidebar, right below Decisions. Action Plans is where an intervention "in progress" actually lives: its subtitle reads "Network initiatives in flight, plans you've activated, interventions in progress, consent requests approved. The work running across the network."

Action Plans opens straight into the Gantt view (a Calendar view is one click away), which is the natural read for an intervention once it's moving, since it's really a small dated checklist, not a flat card. Once an intervention reaches Resolved, or is Ignored, it's terminal: it stops appearing on Action Plans, and on Decisions it settles into the Resolved or Ignored filter instead of Pending or Acted.

The four-stage lifecycle

Every intervention runs through the same four checkpoints, drawn as a single segmented bar on the Gantt: Flagged, Contacted, In progress, Resolved. These are Verinode's own tracked checkpoints for HQ's outreach, not a claim about what's happening inside the franchisee's business.

• Flagged is checkpoint zero, stamped complete the moment the intervention is opened, anchored to the real flagged date.
• Contacted is a fixed marker at day 3 after flagging. It turns green (completed) the first time you click Act, whichever of contacted or in progress the record has actually reached.
• In progress is a fixed marker at day 7 after flagging. It turns green once the intervention has been advanced a second time.
• Resolved turns green only when the intervention is actually closed as fixed, stamped with the real resolved date. Until then it sits as a placeholder about 21 days out; once resolved, Verinode moves it to reflect how long the intervention actually took (at least 7 days after flagging), rather than leaving it at the placeholder.

Behind these four labels, the raw record only ever holds one of five states: queued (Flagged, not yet touched), contacted, in progress, resolved, or no action (closed via Ignore without being worked). On Decisions, these map into the platform's shared status filter as Pending (queued), Acted (contacted or in progress), Resolved, and Ignored. There's no "Parked" state for an intervention: it either sits open, is being worked, or is closed one way or the other.

Working an intervention: Act and Ignore

Open an intervention's tile, on Decisions or from its row on Action Plans, and two buttons are live:

• Act moves the intervention one step forward: queued advances to contacted, contacted advances to in progress, in progress advances to resolved (and stamps the resolved date). The first click tints the tile green with a small Acted stamp and the button row narrows to a single next action, Verinode's visual signal that this one is off the "needs a decision" pile and now being worked.
• Ignore closes the intervention immediately as no action taken, from any open status. Use it when a flag turns out to be a false alarm or something the franchisee already resolved without HQ's help.

There's no Not now for an intervention: the button still renders in the row (greyed out, non-clickable) since interventions don't have a "come back later" state to park into. Once an intervention is resolved or ignored, both buttons disappear and only Discuss remains, so you can still ask the agent about a closed-out item, you just can't reopen it from here.

Reviewing progress from Action Plans

Once an intervention is Acted, Action Plans is where you track it going forward. Its urgency bucket dropdown (All, Planning, Ready to start, Overdue, Due today, Due tomorrow, Later, Completed) sorts every in-flight initiative by its next unfinished checkpoint, not by title, so a stalled intervention whose Contacted or In progress marker has gone red surfaces near the top alongside any other network initiative that's falling behind. See The Gantt timeline for exactly how the bar and its colors work, and Urgency buckets for how the bucketing itself is computed.

Clicking an intervention's row (its label or its bar) opens the same workspace overlay used everywhere in HQ: the flag reason as the title, the franchisee's name, the Risk domain, the flagged and resolved dates, and the notes body described next. It also hands the intervention to the AI agent panel, so you can ask for a recommended next step or a check-in script grounded in what HQ actually knows about that franchisee, without leaving the queue.

How intervention notes are encrypted

The notes field on an intervention is HQ's own running record: what a broker said, what came out of a call, what the next step is. It is written and edited from the franchisee's own record on the Network page (see The interventions queue for at-risk locations for the Add notes / Edit notes flow), and it's exactly what you're reading when you open an intervention's workspace from Decisions or Action Plans. This is the detail this article promised: precisely how that note is protected, and who can actually read it.

• Every HQ group has its own admin-only data encryption key (DEK), entirely separate from anything tied to a franchisee's own operator-side vault. It is wrapped once per admin, under that individual admin's personal HQ Vault Key, the same personal key each admin sets up under Settings.
• There is no master copy of this key anywhere in Verinode's infrastructure. Unwrapping it requires an authorized admin's active, signed-in session; Verinode's own servers cannot decrypt an intervention note on their own.
• When a note is saved, Verinode writes it under this key so that decrypting it back requires that same active admin session. If your session doesn't currently have the group's key in scope (a session that just started, or a group that hasn't set this protection up yet), Verinode falls back to showing an unencrypted copy of the same note rather than showing nothing at all.
• This key protects only what HQ itself writes about a franchisee inside HQ's own workspace. It has nothing to do with, and never touches, a franchisee's private business records, which stay under keys HQ never holds.

Setting this up is a one-time step under Settings, in the Group vault section (only visible once you've set up your own personal HQ Vault Key). Its description reads: "Encrypts HQ-authored broadcasts, intervention notes, prospect pipeline, and shared SOPs under admin Vault Keys, Verinode cannot decrypt these without an active admin session." Until an admin clicks Protect group broadcasts + prospects + notes there, the group has no admin-only key provisioned yet; once it's set up, the section shows Group vault active.

Franchisee identity and the privacy boundary

Whether an intervention shows a franchisee's real name or a de-identified stand-in depends on your network's entity model, a setting on the group itself:

• Most networks (the default, independent operators): the franchisee's real name never appears on this surface. You'll see a stable label like Franchisee #A1B2 instead, built from the last four characters of that franchisee's own operator ID, so the same franchisee always carries the same label across every intervention without ever exposing the real business name.
• Same-entity networks (one legal entity operating multiple company-owned locations): the real location name passes through unchanged, since there's no separate business to anonymize from.

This applies everywhere an intervention's franchisee identity could surface: the tile, the Gantt row label, the workspace, and the text handed to the AI agent, so a network running in the default anonymized mode never has an identity leak through a chat answer either.

Empty states

• Decisions, viewing All, nothing on the board at all: "No decisions for you yet. They'll land here as Verinode spots cost savings, risk, and growth opportunities in your data." Since interventions are opened by an admin rather than generated automatically, this is the honest state for a network where nobody has flagged anyone yet.
• Decisions, a specific status filter with nothing in it (say, Resolved with nothing resolved): "Nothing resolved for you right now" (the same pattern for whichever filter is active).
• Action Plans, nothing in the current urgency bucket: each bucket has its own wording, for example Overdue reads "Nothing overdue. Steps past their due date land here so they don't slip," and Completed reads "No completed plans yet. Plans you resolve land here so you can run a retrospective." See Urgency buckets for the full list.
• On the franchisee's own Network-page record, the read-only interventions list there reads: "No franchisees flagged for intervention. Use the directory to flag at-risk operators."

None of these mean anything is broken. An empty Action Plans bucket means nothing in your network has reached that particular stage yet, not that tracking has failed.

Best-practice example

A network benchmark report shows a franchisee's net margin down from 47% to 38% over two straight quarters. From that franchisee's record on the Network page, an admin clicks Flag, picks "Margin trending below network," and adds "coaching call" as the intervention type. The intervention lands on Decisions, Pending, Risk. The admin opens it, asks the agent for a check-in script grounded in the actual margin gap, and makes the call. Afterward, they log what happened in the notes, "labor mix and pricing reviewed, follow-up scheduled," and click Act to move the record from queued to contacted. From here it tracks on Action Plans: if the Contacted or In progress marker turns red before the next check-in happens, that's the cue to follow up again. Once margin recovers or a credible plan is in motion, the admin clicks Act through to resolved, closing the loop with a record other admins can read.

Related reading

• The interventions queue for at-risk locations, the flagging dialog, the franchisee detail drill-in, and the full notes history behind an intervention.
• Action Plans: the network work in flight, how this surface fits alongside Decisions across all four HQ sources.
• Action Plans vs Decisions, the full comparison of the two surfaces this article draws from.
• The Gantt timeline, exactly how the bar, its colors, and its ordering work.
• Urgency buckets, how Planning, Ready, Overdue, Today, Tomorrow, and Later are computed.
• Network health, the franchisee directory and quartile views an intervention often starts from.
• HQ overview, how Decisions and Action Plans fit into the rest of the HQ shell.

Data sources