Verinode
Open Verinode

Playbook Kinds: Playbook, Directive, Recommendation, Experiment

Every decision plan your organization ships lives in one table, the network data, and every row in it carries a `plan_kind` classification, a required field, never blank. Kind is not a status (draf…

6 min read·Updated July 14, 2026
On this page

What a plan's "kind" is

Every decision plan your organization ships lives in one table, the network data, and every row in it carries a plan_kind classification, a required field, never blank. Kind is not a status (draft, active, paused, completed, archived is a separate column) and it is not the adoption percentage. It is a category tag that says what shape of thing the plan is: a standing process, a one-shot instruction, a soft nudge, or a network-level test. On the Playbook Scoreboard, this tag is the first thing you read on a tile after the plan's title, it is how you tell at a glance whether you are looking at a canonical playbook or something else entirely.

This article is the reference for that one tag: the four kinds Verinode recognizes, what each one means, and exactly how the kind combines with the affected-of-total franchisee count that sits next to it on every tile. For the page as a whole, see Playbook Scoreboard: the network-wide adoption view. For the hero row's own "Distinct kinds" tile, see Playbook Hero Metrics.

Where to find it

Open Playbooks from the HQ sidebar, at hq.verinode.ai/playbooks. The kind tag appears on every tile in the three ranked rows beneath the hero band:

  • Top performers, highest adoption
  • Need diagnosis, adoption below 40%
  • All active playbooks

It sits in the tile's sub-line, directly under the plan's title, paired with the franchisee count. It does not appear on the four hero tiles themselves (Active playbooks, Completed (90d), In draft, Distinct kinds); those are pure counts, not per-plan tags. Clicking any tile with a kind tag opens that plan's full record on the Decisions page, where the same kind is shown again, spelled out, at the top of the detail view.

The four plan kinds

Every plan carries exactly one of four plan_kind values, and the choice is made once when you author the plan in Broadcast, it cannot be changed afterward. The Scoreboard shows the matching label on the tile:

| Database value | Label on the tile | |---|---| | playbook | Playbook | | directive | Directive | | recommendation | Recommendation | | experiment | Experiment |

Playbook is the network's default, canonical kind: a standing, multi-step process you want every membership following on an ongoing basis (a safety stand-down cadence, an estimate-review protocol, a post-job survey program). It is the kind you will see most often on the Scoreboard.

Directive is a one-shot franchisor instruction with a date attached: do this, by then. Its language and intent are firmer than a recommendation, but Verinode never enforces it, nothing locks a membership out for missing a directive. A directive is also the one kind that carries warning-level severity when it lands in the Decisions log; the other three carry info-level severity. See Network playbooks and directives as decision rows for how that severity plays out.

Recommendation is a soft nudge with no enforcement behind it, guidance your organization wants to surface rather than an ask you expect memberships to complete.

Experiment is a network-level test, the kind you use when you want to try an approach with a subset of franchisees before deciding whether to roll it out more widely.

Where a kind is chosen: Broadcast

Open Broadcast, select the Initiatives tab, and start a new plan. The kind picker offers exactly these four choices, matching the plan_kind values the database accepts on write:

  • Playbook: a canonical multi-step plan for the network.
  • Directive: a one-shot franchisor instruction, do this, by this date.
  • Recommendation: a soft nudge with no enforcement behind it, surfaced as guidance rather than an ask.
  • Experiment: a test run at the network level, typically opt-in for a subset of franchisees before any wider rollout decision.

The kind is set at creation time and is fixed for the life of the plan; if you decide a plan should have been a different kind, you author a new one. See Broadcasting to your network for the full authoring and push flow, and Scoreboard vs Decisions Inbox for how the Scoreboard differs from the inbox where the kind is first chosen.

How kind combines with the franchisee count in the sub-line

The sub-line format is not identical across the three ranked rows. It is built from the same two pieces, the kind label and the affected-of-total franchisee count, but the punctuation differs by row:

  • Top performers and Need diagnosis: Kind · N of M franchisees, for example "Playbook · 18 of 24 franchisees." Written out in full because these two rows are wider, double-width tiles with room for the complete phrase.
  • All active playbooks: Kind · N/M, for example "Playbook · 18/24." Compressed because this row packs up to eighteen standard-width tiles into one horizontally scrolling line and needs the tightest possible sub-line.

In both formats, N is the plan's current_affected_count, how many franchisees the plan is currently reaching, and M is current_total_count, how many it was pushed to. Both numbers are cron-decorated, the same nightly job that calculates the adoption percentage shown as the tile's label also writes these two counts, and Verinode never breaks the total down by which specific memberships are or are not counted, that per-franchisee detail lives one click away on the plan's own page, not on the Scoreboard tile.

If a plan has been activated but the nightly cron has not decorated it yet, both counts read 0, so the sub-line shows "Kind · 0 of 0 franchisees" (or "Kind · 0/0" in All active playbooks) alongside an "Awaiting cron" label in place of a percentage. This is expected on a plan that just went active minutes ago, not a sign the push failed. Top performers and Need diagnosis exclude undecorated plans entirely, so this zero-state sub-line only shows up in practice on All active playbooks, where every currently active plan appears regardless of whether it has a real number yet.

  1. 1Read the kind first. It tells you whether you are looking at a standing Playbook, a one-shot Directive, a soft Recommendation, or a network-level Experiment.
  2. 2Read the count next to it. "N of M" or "N/M" is how many franchisees the plan is currently reaching out of how many it was pushed to, not a ranking and not a pass/fail grade on its own.
  3. 3If the count reads "0 of 0" (or "0/0") next to "Awaiting cron," give it until the next overnight cron run before drawing any conclusion.
  4. 4Click the tile to reach the plan's full record on Decisions if you need the per-membership breakdown behind the total.

The privacy boundary

A plan's kind and its affected-of-total count are both properties of your own authored plan and your own network's aggregate response to it. HQ never sees an individual franchisee's private business data (their jobs, their financials, their clients) anywhere on the Scoreboard or on the plan detail it opens. What the count shows is how many memberships have engaged with something your organization wrote and pushed, not what is happening inside any single membership's own business. Franchisees own their data; kind and count together measure your program, not theirs.

Related reading

Data sources

  1. 1.Decision plan records, by kind, status, and franchisee count. the network data.
Was this helpful?