How to Run Better City Strategy Sessions in 2026

The city strategy session is one of the better-run meetings in local government. Someone builds the deck. Departments send their numbers. A facilitator keeps time. People arrive prepared, because in a city, arriving unprepared is a public act.

Then the next session comes around, and the same three items are still red.

We watch this from an unusual seat. ClearPoint hosts the strategic plans of 79 city and town governments — 1,903 plans, 88,833 tracked items, and more than 1.7 million status updates posted against them. We never see the room. We see what people typed into the plan on the day the room met, and what happened to those items the period after.

One pattern keeps surfacing. Of the 340,079 status updates these cities scored between 2020 and 2025, 15.2% carry a written explanation and 4.1% carry a written next action. Ninety-six times out of a hundred, an item leaves the review with a color and nothing else.

That number is the session problem in miniature. The meeting happens. The discussion is genuine. And the plan records a color.

So this is a guide to the other 4%. How to size the agenda to the room you actually have, who to seat in it, the four lines to write on every item before anyone stands up, and how those lines become tracked initiatives instead of a memory. Facilitation craft matters less here than most guides assume. What decides whether a session mattered is what survives it in writing.

What a recurring city strategy session is for

A city strategy session is a standing leadership review — usually monthly or quarterly — where the management team scores progress on the strategic plan, works the items that are off track, decides what happens next on each one, and records those decisions against the plan itself. It is the operating cadence, and it is a different meeting from the annual retreat that sets direction in the first place.

That distinction does real work. The retreat produces priorities. The session produces movement, or fails to. If your plan flatlined weeks after the offsite rather than during it, the diagnosis lives in our companion guide on why local government strategy workshops stall. This guide picks up at the next meeting, and every one after it.

Build the invite list from the ownership report, not the org chart

The standard advice is to bring the leadership team plus one representative per department, and to cap the room at ten to fifteen people. Sensible enough. It also produces a room where most of the people present own nothing on the agenda.

Here is what ownership actually looks like inside a city plan. Across the cities we host, the median one has 19 distinct people who own at least one item — the first quartile sits at 6 or 7, the third at 36. And ownership concentrates hard: in the median city, the top three owners hold 47.3% of every owned item in the plan. Half the plan sits with three people.

That changes the invite question. The room you want holds the people who will be asked to move something, plus the person who can unblock them. Representation belongs in the report that goes out afterward.

A workable rule:

• Pull the ownership report first. Sort by who owns items that are currently red or yellow. That list is your core room.
• Seat the unblocker. Most stuck municipal items are stuck on money, staffing, or a decision that sits one level up. Somebody in the room needs the authority to resolve it that morning.
• Invite the rest to read, not to attend. Owners with everything green do not need 60 minutes. They need the report.

The council-and-staff dynamic — elected officials setting direction while staff carry operations — is real, and it is the reason so many city goals leave a room deliberately vague. We cover that mechanic in detail in the workshops guide. For the recurring session, the practical version is simpler: keep the session a staff working meeting, and send council a report from it. A working review and a public accountability moment are two different meetings, and merging them turns the first one into theater.

Size the agenda to the time you actually have

Cities put more on the review table than the table can hold. In our data, the median city scores 42 items in a single reporting period. The 75th percentile scores 126. The 90th percentile scores 250. One city posted 1,015 scored items against one period.

Run the arithmetic on 42 items in a 60-minute session and each item gets 85 seconds — assuming nobody says hello, nothing runs long, and no one asks a follow-up question. At the 90th percentile, you would need to clear an item every 14 seconds.

No facilitator fixes that. The agenda has to shrink before the meeting starts, which means the session runs on exceptions.

• Green items are read, not discussed. They go in the pre-read. The session acknowledges them in one line.
• Red and yellow items get the whole hour. Cap it at eight. If you have more than eight exceptions, you have a prioritization problem that a meeting cannot absorb.
• Each exception gets a time box and a required output. Not "discuss the permitting backlog." Rather: "permitting backlog — decide the next action and who owns it, 7 minutes."
• Anything needing more than its box goes to a named side meeting with a date, decided in the room.

The myth worth retiring: "review and discuss." Almost every strategy-meeting guide builds its agenda around reviewing progress and discussing what is behind. We can measure what that produces. Across 34,449 occasions when a city scored an item red and then scored it again the following period, the items that left the review with a written explanation and nothing else recovered to green 14.2% of the time. The items that left with no note at all recovered 14.6% of the time. Within the resolution of this data, explaining why an item is red performs the same as saying nothing about it. Explanation is how a room feels productive. Something else has to happen for the number to move.

The four lines to write on every item before anyone stands up

The thing that does move the number is small, boring, and almost never prescribed: before the room breaks, every exception carries a written next action, on the item, in the system of record.

Four lines per exception. No more.

1. The item. One element from the plan — a measure, an initiative, a milestone. Not a theme, not a department.
2. The owner. One name. A person, not a division.
3. The next action. The specific thing that happens next, written in a sentence someone could execute without having been in the room.
4. The date. When that action lands, and therefore when this item comes back.

Three of those four are conventional advice. The third is the one that gets skipped, and it is the one that carries the effect.

What most guides skip: write the decision on the item, in the room. Looking at the same 34,449 red-then-rescored observations across 65 cities: when the review left a written next action on the item, it recovered to green 21.7% of the time and climbed off red entirely 31.1% of the time. With a color alone, those figures were 14.6% and 19.9%. Roughly a 1.5× improvement in recovery, from one sentence typed before people stood up.

The obvious objection is that better-run cities write more decisions, so we tested it inside each city. Among the 23 cities with at least 20 red observations in both buckets, 20 recovered more often when a next action had been written than when it had not — a city-level average of 25.6% versus 18.8%. The effect holds within organizations, not only between them.

And the practice is rare. Only 6.0% of red items in our municipal data left their review with a next action recorded.

Where the writing happens matters as much as whether it happens. A decision captured in meeting minutes is a decision the plan cannot see. A decision typed onto the item is a decision that shows up next period, next to the number it was supposed to move.

When we stand this up with a city, the sticking point is almost never the discipline of the leadership team. It is that nobody has been given the job of typing during the meeting. Someone narrates the decision, everyone nods, and the note gets written later by a person reconstructing it from memory — or not at all. The cheapest fix we know of is to name a scribe with edit rights before the session starts and put their screen on the wall. The room then argues with the actual sentence, which is a much better argument than the one it was having.

There is a second reason to write the owner down at the item level. Municipal plans get less owned the more abstract the item is. Across the 88,833 items we host for cities:

The ladder is consistent: the closer an item sits to a concrete task, the more likely it carries a name. Objectives — the things a strategy session spends most of its air on — are unowned 80.8% of the time. That figure tracks the cross-sector pattern in our research set, where roughly 77% of strategic objectives have no owner and owned objectives are about twice as likely to reach a healthy status.

Which gives you a cheap in-session move: when an objective is the exception, do not assign the objective. Assign the next action underneath it. Action items are the best-owned things in a city plan for a reason — they are small enough that a name sticks to them.

How often to run it, and what the annual review cannot see

Cities run their plans on a declared reporting cadence, and the cadence decides what the session is even able to look at.

Read the annual row carefully, because it is the one that flatters. Items reviewed once a year are scored red 5.1% of the time — less than half the rate of items reviewed monthly. That gap measures visibility rather than health. An annual review catches the plan at the one moment in the year when everything has been reconciled, rewritten, or quietly closed, which is why it also produces the fewest recorded decisions of any cadence, at 2.0%. A city that reviews once a year has not found fewer problems. It has looked once.

The practical read for a city:

• Monthly for the operating measures that a department can actually move inside 30 days — service response times, permit throughput, case backlogs. See the local government KPI categories for what belongs at this tempo.
• Quarterly for objectives and capital initiatives, where a month of movement is noise.
• Annual for direction only — the retreat. Treat the annual number as a summary, never as the review.

Whatever you choose, put the next twelve sessions on the calendar in the current one. Cadence that depends on someone remembering to schedule it is not a cadence.

The five days after the session

A session's output has a short half-life. The window to convert it is roughly a working week.

• Day 0, before adjourning. Every exception has its four lines typed onto the item, by the named scribe, while the room is still seated. Nothing is left to be written up later.
• Day 1. The report goes out — the same view everyone just looked at, so nobody rebuilds it in a slide deck. Owners receive their items, not the whole plan.
• Days 2–3. Any exception that needed a side meeting gets that meeting held, and the outcome is written onto the item, not into an email thread.
• Day 5. Someone checks that every next action from the session actually exists in the system with a date. This takes about ten minutes and it is the single highest-yield ten minutes in the cycle.

Then the reminder does the rest of the work. Look at which cities in our data keep their plans current and one boring trait shows up: a scheduled nudge fires on an ordinary Tuesday, and an owner answers it. For the mechanics of standing that up — owners, escalation paths, and automated reporting — the workshops guide has the 90-day runbook, and our strategic planning software guide covers what the system of record needs to do.

For session-agenda structure and facilitation context specific to local government, the ICMA Knowledge Center's strategic planning collection is the useful primary source. And if you want models of what a mature city plan looks like before you design your own review, we broke down four of them in local government strategic planning examples.

Turn this session's decisions into something you can track

If your plan does not yet live somewhere that can hold an owner, a next action, and a date on each item, start with the sheet. Our free Gap Analysis Template is built around exactly the four lines above — objective, current state, gap, action, and a single named owner per row — with a health status that calculates itself from what you enter, plus a worked example so nobody starts at a blank grid. Bring it to the next session and fill it in the room.

The part worth remembering

Every city we host runs sessions. The plans that move and the plans that stall look identical on the calendar. The difference shows up in a field most agendas never mention: whether the item carries a sentence about what happens next, written while the people who decided it were still in the room.

One sentence, four lines, six percent of the time. That is the whole gap between a good meeting and a plan that moves.

Frequently asked questions

Why do city leadership teams struggle to run effective strategic planning sessions?

In our municipal data, the difficulty is rarely facilitation. It is that the session's output never gets recorded against the plan: across 340,079 scored status updates from 79 municipal governments, only 15.2% carry a written explanation and 4.1% carry a written next action. The agenda is usually too large to work as well — the median city scores 42 items in a single reporting period, which allows 85 seconds per item in a one-hour meeting.

How often should a city hold strategy sessions?

Monthly for operating measures a department can move inside 30 days, quarterly for objectives and capital initiatives, and annually for direction only. Cadence changes what the session can even see: items on an annual reporting cycle are scored red 5.1% of the time versus 11.9% for monthly items, because the annual review catches the plan after everything has been reconciled. Annual cycles also produce the fewest recorded decisions, at 2.0% of updates.

Who should attend a city strategy session?

Build the invite list from the ownership report rather than the org chart. In the median city we host, 19 distinct people own at least one item and the top three owners hold 47.3% of every owned item, so a department-representative room seats people who own nothing on the agenda. Invite the owners of items that are currently red or yellow, plus one person with the authority to unblock money, staffing, or a decision one level up.

What should be on a city strategy session agenda?

Exceptions only. Green items belong in the pre-read and get one line of acknowledgement; red and yellow items take the whole meeting, capped at roughly eight. Give each exception a time box and a required output — the next action and its owner, decided in the room — and send anything that needs more time to a named side meeting with a date.

How do you turn a strategy session into tracked initiatives?

Before the room breaks, write four lines onto each exception in the system of record: the item, one named owner, the specific next action, and the date it lands. Red items that left a review with a written next action recovered to green 21.7% of the time versus 14.6% with a color alone, and the effect holds within individual cities. Then check on day five that every next action actually exists in the system with a date.

What is the difference between a strategic planning retreat and a strategy session?

A retreat is the one-time annual event that sets direction — vision, priorities, and the shape of the plan. A strategy session is the recurring leadership review, usually monthly or quarterly, that scores progress on that plan and decides what happens next on the items that are off track. The retreat produces priorities; the session is what makes them move, or does not.

Data note: figures come from ClearPoint's anonymized research dataset, municipal-government segment, queried 2026-08-21 — 79 city and town governments, 1,903 plans, 88,833 tracked items, and 340,079 scored status updates for reporting periods between 2020-01-01 and 2025-12-31. Training, sandbox, demo, and archived plans are excluded. Recovery figures compare an item's status in one reporting period with its status in the next. These are observed patterns across real plans, not a controlled experiment.