How to plan a leadership offsite agenda around the decisions it must make

Disclosure: NOXI publishes this guide and is one of the tools it mentions; the method works in any presentation software.

Most leadership offsite agendas are lists of topics: strategy, culture, the market, the budget, team building. A topic list fills two days easily and decides very little, because a topic has no finish line. This guide builds the agenda the other way round. You start from the six decisions the leadership team must leave with, budget the minutes against them, push every status update into a pre-read, and finish with a decision log that the company hears about within a week. It gives you the two-day agenda with minutes that add up, a one-day and a remote version, the inside of a 90-minute decision session, a 10-slide pre-read deck, decision rights for each item, the prioritisation arithmetic, a readiness checklist and a complete fictional example for January 2027.

What a leadership offsite is for, and what it is not for

A leadership offsite exists to make the few decisions that need the whole leadership team in one room for a long stretch of time: where the company focuses next year, what it stops, who owns what, and how the team works together. It is not for status updates, which a document can carry at a fraction of the cost.

The test is simple. If an item could be settled by one executive with two colleagues in a 30-minute call, it does not belong at the offsite. If it needs every function's trade-offs on the table at once, such as moving budget from one priority to another or stopping a project that three teams depend on, it does. Most companies hold their main leadership offsite in January, or just before the fiscal year starts, because that is when next year's plan has to become a set of choices.

An offsite is also the most expensive meeting most companies hold. Nine leaders for two days of eight hours is 144 leader-hours before travel and venue. Every hour spent on slides that could have been read in advance costs nine leader-hours, which is the economic reason for the rule in this guide.

Pick the decisions first: six questions the offsite must answer

A decision-first agenda starts with a list of six decisions, each written as a question with its options and a deadline, agreed by the CEO about six weeks before the offsite. Topics such as "strategy" or "culture" become questions such as "Which three initiatives do we fund in 2027?" so that everyone knows when a session is finished.

Six is a working number for two days: enough to cover priorities, resources and people, few enough to give each one 60 to 90 minutes. Write each decision with the same five parts, so that the person preparing it knows what to bring.

The five parts of a decision brief are: the question; the options (two to four, never one); the criteria you will use to choose; the person who recommends and the person who decides; and the deadline, which is the end of the session. A useful sentence to test each brief: "By [time] on [day], [decider] will choose between [options] using [criteria], based on the recommendation of [name]." If you cannot fill every bracket, the item is not ready to be on the agenda.

Your six may differ. A company in a turnaround may replace D1 with "Which markets do we exit?"; a company after an acquisition may add "Which systems do we keep?". Keep the count near six and the form identical. Topics that are not decisions, such as a market briefing, still belong, but as context sessions that feed a named decision.

The two-day leadership offsite agenda, with minutes that add up

A two-day leadership offsite runs 480 minutes a day, 960 in total, and gives 480 of them (50%) to the six decision sessions. Context sessions take 180 minutes, team sessions 120 and breaks and meals 180. Day 1 decides what the company will and will not do; day 2 turns those choices into targets, owners, money and a message.

Each day: 30 + 60 + 15 + 60 + 90 + 60 + 90 + 15 + 60 = 480 minutes, 08:30 to 16:30. Across both days, decisions take 90 + 90 + 60 + 90 + 90 + 60 = 480 minutes (50%), context 60 + 60 + 60 = 180 (18.75%), team sessions 30 + 60 + 30 = 120 (12.5%) and breaks and meals 15 + 60 + 15 twice = 180 (18.75%). The dinner on the first evening is not on the agenda and should stay free of decisions; people who were outvoted in the afternoon need an evening that is not a second round.

Three choices in the order matter. Context comes before the decision it feeds, so the market session runs before D1, and the financial frame before D3 and D5. "What we stop" follows straight after the priorities, while the trade-off is still fresh. And day 2 opens with a 30-minute review of the decision log from day 1, which is when second thoughts surface; it is cheaper to reopen a decision at 08:30 on day 2 than in an email in February.

Inside a 90-minute decision session

A 90-minute decision session follows six fixed steps: restate the question and options (10 minutes), hear the input roles (30), score silently and individually (15), debate the gaps in the scores (20), the decider decides and says it aloud (10), and the minute-taker writes the log line (5). The silent scoring step prevents the first or loudest voice from setting the outcome.

The steps add up to 10 + 30 + 15 + 20 + 10 + 5 = 90. For a 60-minute session, cut the input round to 20 minutes, silent scoring to 10, the debate to 10 and the decision step to 5 (10 + 20 + 10 + 10 + 5 + 5 = 60). If the session ends without a decision, it ends with one of two other outcomes: a named owner who will bring a decision by a date, or a parked item with a review date. "We discussed it" is not an outcome.

The one-day version: four decisions in 480 minutes

A one-day leadership offsite can carry four decisions: priorities, what to stop, targets with the budget envelope, and owners with the company message. It gives 300 of its 480 minutes (62.5%) to decisions, so it only works when the pre-read has done the context work and the leadership team has read it.

The total is 20 + 40 + 90 + 15 + 75 + 45 + 60 + 75 + 15 + 45 = 480 minutes: decisions 90 + 75 + 60 + 75 = 300 (62.5%), context 40 (8.3%), team 20 + 45 = 65 (13.5%) and breaks 15 + 45 + 15 = 75 (15.6%). What the one-day version drops is the team session on how you work together. Hold that one separately, for example as a half-day in the following quarter, rather than squeezing it into the last hour.

A remote or hybrid leadership offsite

A remote leadership offsite works best as four half-day sessions of about three hours spread over one or two weeks, with one or two decisions in each. Video calls tire people faster than a room does, so a decision session is capped at 75 minutes with a break after it, and silent scoring happens in a shared form rather than by a show of hands.

The pre-read: a 10-slide deck and a short memo, sent five days ahead

The offsite pre-read is a 10-slide deck, optionally with a four- to six-page memo, sent to every participant at least five days before the offsite. It carries everything the leadership team needs to know before deciding: last year's numbers, the market, the candidate initiatives, the financial frame and the draft options. On stage, the pre-read is discussed, never presented.

Write every slide title as a full-sentence finding ("We grew 20% but net revenue retention is flat at 108%"), not a label ("Financials"), so that a leader who reads only the titles still gets the argument; our guide to consulting-style slides with action titles shows the method. On the financial frame, show the range as well as the base case, because the budget decision depends on how wide it is; the patterns in presenting forecast uncertainty apply directly.

Budget the reading time, because a pre-read nobody finishes moves the status update back onto the agenda. Adults read English non-fiction silently at about 238 words a minute on average, with most between 175 and 300, and more slowly in a second language, according to a meta-analysis of 190 studies (Brysbaert, Journal of Memory and Language, 2019). A five-page memo of about 2,500 words is therefore about 10.5 minutes (2,500 ÷ 238); ten data slides at about two minutes each add 20. Thirty minutes is a realistic ask. Ninety is not.

Some teams read in the room instead. Amazon's 2017 shareholder letter describes the company's narratively structured six-page memos, which are read silently at the start of each meeting "in a kind of 'study hall'" (Amazon, 2017 Letter to Shareholders). If your leaders often arrive unprepared, open the state-of-the-business session with 20 minutes of silent reading of the memo; it costs less than an hour of presenting it.

Collect questions on the pre-read in writing two days before the offsite. The 60-minute state-of-the-business session then answers the questions people actually have, grouped by theme, instead of walking through slides they have already read.

The opening: four slides that set up two days

The offsite opening takes 30 minutes and four slides: the purpose in one sentence, the six decisions with who decides each, the ground rules, and how disagreement will be handled. After the opening, nobody should be surprised by what is being decided or by who has the final say.

1. Purpose. One sentence: "In two days we will choose next year's three priorities, what we stop, the targets, the owners and the budget split, and agree what we tell the company."
2. The six decisions. The questions from the decision list, each with its session time and the person who decides.
3. Ground rules. Phones and laptops closed in decision sessions; one conversation at a time; the parking lot for anything off-question; the minute-taker writes decisions only, not discussion; what is discussed stays in the room, what is decided goes to the company on an agreed date.
4. How we will decide. Input is heard before scores are revealed, the named decider decides, and once a decision is made, everyone supports it in public, including those who argued against it.

Ask each leader to say in one sentence what they need the offsite to decide. If an answer is not on the list, it goes to the parking lot now, not at 15:00 on day 2.

Decision rights: who recommends, who agrees, who decides

Every offsite decision needs a named decider before the offsite starts, and one person who prepares the recommendation. Bain & Company's RAPID method names five roles: Recommend, Agree, Perform, Input and Decide. Writing the roles into the pre-read prevents the most common offsite failure, which is a good discussion that nobody has the authority to close.

In Bain's description, the recommend role drives the process and develops a recommendation for the decider, the input role provides expertise or information, the agree role must agree that the recommendation is feasible, the perform role carries out the decision, and the decide role makes the final call and commits the organisation to action (Bain & Company, RAPID Decision Making). Bain also notes that agree roles should be assigned sparingly. Here is a typical map for the six decisions in a nine-person leadership team.

Everyone in the room is an input role for every decision, so the table leaves that column out. The CEO holds the final say on five of the six, which is normal for a leadership team; the value of the map is that the recommender, the narrow agree roles and the performers are named before anyone walks in.

Disagree and commit

The second half of decision rights is what happens after the decision. In his 2016 letter to shareholders, Jeff Bezos recommends the phrase "disagree and commit" for decisions where there is conviction but no consensus, and applies it to leaders themselves (Amazon, 2016 Letter to Shareholders). The same letter argues that most decisions should be made with about 70% of the information you wish you had, and that many decisions are reversible "two-way doors" that can use a light-weight process.

At an offsite, two-way-door thinking saves time. Mark each decision in the log as one-way (hard to reverse, such as an org change) or two-way (easy to revisit, such as the split of an investment envelope in a quarter). Spend the debate time on the one-way doors and set a review date on the two-way ones.

The prioritisation session: from 14 candidate initiatives to three

To cut a long list of candidate initiatives to three priorities, give each leader as many votes as there are priorities to fill, collect the votes silently, check the leading candidates against an impact and effort grid, and let the decider confirm. With nine leaders, three votes each and 14 candidates, there are 27 votes; three initiatives that take 20 of them (74%) have a clear mandate.

The method has four steps. First, the recommender shows the shortlist on slide 6 of the pre-read: each candidate placed on a grid of expected impact against effort, with a one-line rationale. Second, each leader casts three votes in writing, at most one per initiative, so that no one can pile all their votes on a favourite. Third, the votes are revealed together. Fourth, the group debates only the gaps: a high-vote item in the high-effort corner, or a low-vote item someone thinks is essential.

The vote is input, not the decision. A dot vote measures preference, and preference can favour the easy option or the vote-caster's own function. That is why the grid stays on the screen during the reveal, and why the named decider, after hearing the debate, states the three priorities aloud. If the top three include two high-effort initiatives, the decider can swap one for the fourth-placed item and must say why.

The "what we stop" session: the decision most offsites skip

A leadership offsite should stop at least as many things as it starts, because new priorities compete for the same people and money as existing work. The "what we stop" session runs straight after the priorities, takes 90 minutes, and ends with at least three named things stopped, each with the money and people it frees and a date.

Stopping is harder than starting because every stop has an owner in the room. Make it easier with criteria agreed in advance and applied to a prepared list (pre-read slide 7), not to whatever comes up. A candidate qualifies if it meets two or more of these.

Stops include meetings, reports and events, not only projects. A weekly report that nobody reads frees analyst time; a conference that brings few new customers frees budget. The output is a line in the decision log, with the amount freed added to the investment envelope for D5.

The team session: how the leadership team works together

The team session agrees three to five operating norms for the year: how the leadership team makes decisions between offsites, which meetings it holds and how often, and how it handles disagreement. It takes 60 minutes on day 1 and produces norms specific enough to check in the next quarterly review.

Ask each leader beforehand to answer two questions in writing: "What does this team do that slows decisions down?" and "What should we keep?" Group the answers, then agree norms in the same decision format as the rest of the offsite. Good norms are testable: "Decisions taken in the weekly leadership meeting go into the decision log by Friday", "No decision is reopened without new information", "Each leader runs one skip-level meeting a month". Vague norms such as "communicate better" do not survive January.

Disagreements about the pre-read itself are common before the offsite: two leaders want the same slide to say different things. Settle them before the deck goes out, with the recommender's version and the dissent noted on the slide; our guide to resolving conflicting presentation feedback has a method that works for leadership teams.

Facilitation: an outside facilitator or the CEO

The CEO should not facilitate an offsite where the CEO also decides most items, because one person cannot both run a fair process and argue for an outcome. The usual choices are the chief of staff, an outside facilitator, or a rotating facilitator from the leadership team for the sessions they do not own.

Three roles sit beside the facilitator. The minute-taker writes only decisions, owners, dates and parked items, in the decision-log format, live on a shared screen. The timekeeper warns at ten minutes and two minutes before the end of each session. The parking-lot owner keeps the list and, at the close, gives each parked item an owner and a date.

The decision log slide and the 72-hour follow-up

A decision log is the table of every decision the offsite made: the decision in one sentence, who decided, who owns it, the next date and whether it is reversible. It is built live during the sessions, read back at the end of each day, and sent to the leadership team within 72 hours of the close.

The log has a fixed set of columns so it can be scanned in a minute: number, decision, decider, owner, next date, door (one-way or two-way). The worked example below shows a complete log for six decisions. Two rules keep it useful. A decision only enters the log when the decider has said it aloud in the room. And the log sent after 72 hours must match the log read back on the last afternoon; changes made afterwards go to the next leadership meeting as reopened decisions, in writing.

After the 72-hour log, the decisions move into the normal management rhythm. The targets become objectives and key results; our OKR presentation guide shows how to present them. The budget split feeds the annual plan and budget presentation, and decisions that need board approval go to the next board meeting presentation with the offsite's recommendation and its alternatives. At 30 and 90 days, the leadership team reviews the log: decisions done, decisions on track, decisions reopened.

What we tell the company: from the offsite to the all-hands

The company should hear what the leadership offsite decided within about a week, at an all-hands, in five slides: why the leaders met, the three priorities, what stops and why, the targets, and what changes for each team. Silence after an offsite invites rumours that cost more than the announcement.

D6, the last decision session, agrees the message, the date and the cascade: which leader tells which team, and what managers hear before the all-hands. Say what stops as clearly as what starts, and say who is affected; people working on a stopped project should hear it from their leader the day before, not from a slide. Our all-hands meeting presentation guide has a full run of show for the meeting itself, including how to take questions.

Leadership offsite agenda topics and ideas, mapped to decisions

Good leadership offsite topics are ones that end in a choice. A topic that does not feed one of the decisions is better handled in the pre-read, a recorded briefing or a separate meeting. The table maps common offsite agenda topics to the decision they serve and the session type that fits them.

Team-building activities have their place, usually in the evening or a short block after lunch. Keep them out of the decision sessions and keep them optional where they involve physical activity, alcohol or personal disclosure.

The offsite readiness checklist

An offsite is ready when six things are true a week before it starts: the decisions are written as questions, each has a named decider and recommender, the pre-read has been sent, the agenda adds up, the roles beside the facilitator are filled, and the date of the company announcement is fixed. Check each item with a clear pass or fail.

Worked example: Kestrel Health's January 2027 leadership offsite

Kestrel Health is a fictional company that sells a virtual physiotherapy programme to employers and health insurers. It has 380 employees and a nine-person leadership team: CEO, CFO, COO, Chief Commercial Officer, Chief Product Officer, CTO, Chief Medical Officer, Chief People Officer and CMO. Its offsite runs on Thursday 14 and Friday 15 January 2027, facilitated by the chief of staff. Every number below is fictional, and the arithmetic is shown.

The plan, working backwards from the offsite

The pre-read: this year in five numbers

Kestrel's 2026 in five numbers: revenue $48.0M, up 20% on $40.0M (48.0 ÷ 40.0 = 1.20); gross margin 61%; net revenue retention 108%; 212 employer clients and 6 insurer contracts; EBITDA −$1.9M, or −4% of revenue (0.04 × 48.0 = 1.92). The financial frame for 2027 sets revenue at $57.6M (48.0 × 1.20) and an investment envelope of $4.0M for new priorities: $2.9M of new money plus whatever the stops free.

D1: from 14 initiatives to three

Nine leaders, three votes each: 27 votes. The results: insurer channel, from 6 to 12 contracts, 8 votes; prevention module for existing employer clients, 7; automated clinician scheduling, 5; launch in Germany, 3; video-first physiotherapy for rural members, 2; pelvic-health programme, 1; data platform rebuild, 1; the other seven candidates, 0. Check: 8 + 7 + 5 + 3 + 2 + 1 + 1 = 27. The top three took 20 of 27 votes (74%), with a gap of 2 to fourth place. Germany scored high on impact but sat in the high-effort corner; it is parked with a review at the 90-day meeting. The CEO confirmed the three.

D2: three things stopped

Kestrel stops its consumer app subscription pilot ($0.5M in the 2027 draft plan), custom reports built for individual employer clients ($0.4M of analyst time, replaced by one standard dashboard) and its third annual client conference ($0.2M). Freed: 0.5 + 0.4 + 0.2 = $1.1M, so the envelope is 2.9 + 1.1 = $4.0M. The Chief Medical Officer, in the agree role, confirmed that no stop affects a clinical obligation to current members.

D3 to D6 on day 2

D3 set four targets for 2027: revenue $57.6M; net revenue retention 110% (from 108%); EBITDA break-even in the fourth quarter (from −$1.9M for 2026); programme completion 65% of members (from 58%). D4 named the owners: the Chief Commercial Officer for the insurer channel, the Chief Product Officer for the prevention module, the COO with the CTO for scheduling automation, and a new Head of Insurer Partnerships, reporting to the Chief Commercial Officer, to be hired by 31 March 2027. D5, decided by the CFO within the envelope, split the $4.0M: insurer channel $1.8M (45%), prevention module $1.4M (35%), scheduling automation $0.8M (20%); check 1.8 + 1.4 + 0.8 = 4.0. D6 fixed the all-hands for Thursday 21 January, with the teams on the three stopped items told by their leaders on Wednesday 20 January.

The decision log Kestrel sent on 18 January

Kestrel's pre-read was built as a 10-slide deck in NOXI from the brief above. The parked list, with Germany at the top, goes into the 90-day review on 15 April; that is what keeps "parked" from meaning "forgotten".

Here are two slides from this article's worked example, as NOXI built them, unedited.

Three AI prompts for the offsite pre-read and decision log

AI tools draft offsite material well when you give them the decisions, the numbers and the structure, and ask them to invent nothing. Leadership material is often confidential: use only what your company allows in the tool, and check our checklist for confidential material in AI presentation tools before you upload a board pack or salary data.

How to build the offsite pre-read deck in NOXI

NOXI turns your planning files into a structured deck and can match the style of a template you already use. It is free, with no card and no watermark; uploading files needs a free account, and a full deck uses about 25 of the 1,000 monthly credits.

Frequently asked questions

What should be on a leadership offsite agenda?

A leadership offsite agenda should list the decisions the team must make, each written as a question with its session time, plus the context sessions that feed them, a team session on how leaders work together, and breaks. A two-day agenda typically covers six decisions: priorities, what to stop, targets, owners and org changes, the budget split and the company message. Status updates go in the pre-read.

How long should a leadership offsite be?

Two days of about eight hours each is the common length for an annual leadership offsite, enough for six decisions with 60 to 90 minutes each plus context and team sessions. A one-day offsite can handle four decisions if the pre-read carries the context. A remote offsite works better as four half-day sessions of about three hours over one or two weeks.

What topics should a strategy offsite cover?

A strategy offsite should cover the topics that end in a choice: which initiatives to fund, what to stop, next year's targets, who owns each priority, and how to split the investment budget. Market reviews, customer insight, risks and engagement survey results belong too, but as context sessions or pre-read pages that feed one of those decisions, not as open-ended discussions.

What should you send before an offsite?

Send a pre-read at least five days ahead: a 10-slide deck, and optionally a four- to six-page memo, covering why you are meeting and who decides what, this year's key numbers, market insights, the candidate initiatives, the financial frame, options for each decision, draft targets and the agenda. Keep it to about 30 minutes of reading, and collect written questions two days before.

Who should facilitate a leadership offsite?

Someone other than the CEO, because the CEO usually decides most items and cannot also run a neutral process. A chief of staff works when the team trusts the process; an outside facilitator is worth the cost when there is conflict, a new CEO or a hard decision about what to stop. Add a minute-taker, a timekeeper and a parking-lot owner.

How do you make decisions at an offsite?

Name one decider and one recommender for each decision before the offsite, for example with Bain's RAPID roles. In the session, restate the options, hear input from everyone, collect scores or votes silently and reveal them together, debate only the gaps, then have the decider state the decision aloud. Record it in the decision log, and expect everyone to support it afterwards, as in "disagree and commit".

What is a good offsite agenda for a team?

A team offsite uses the same pattern on a smaller scale: one day, three or four decisions such as the team's priorities for the quarter, what to stop, who owns what and how the team works together, with a short pre-read sent a few days ahead. Keep a third of the day for the team session and social time, since team offsites often matter as much for working relationships as for decisions.

What happens after a leadership offsite?

Within 72 hours the decision log goes to the leadership team, matching what was read back in the room. Within about a week, the company hears the outcome at an all-hands, with affected teams told the day before. Targets become objectives, the budget split feeds the annual plan, and the log is reviewed at 30 and 90 days, including every parked item.

Should an offsite use slides?

Use slides for the pre-read, the four opening slides, the decision grid in each session and the decision log, but not for presenting context the team has already read. A good rule is no new slides in a decision session: refer to pre-read page numbers. Some companies use a written memo read silently at the start instead, as Amazon describes in its 2017 shareholder letter.

Sources & methodology

How this guide was made: the decision-first agendas, the minute budgets, the decision-session steps, the pre-read template, the prioritisation arithmetic, the stop criteria, the readiness checklist and the decision-log format are editorial recommendations based on common leadership-team and facilitation practice, not published standards. Kestrel Health and every name and figure in the worked example are fictional, and every calculation is shown. The sources below support the facts about the world; all were checked on 15 October 2026.

• [1] Narratively structured six-page memos instead of slide presentations, read silently at the start of meetings in a "study hall" — Amazon, 2017 Letter to Shareholders (Jeff Bezos).
• [2] "Disagree and commit"; decisions with about 70% of the information you wish you had; reversible "two-way door" decisions with a light-weight process — Amazon, 2016 Letter to Shareholders (Jeff Bezos).
• [3] The leadership principle "Have Backbone; Disagree and Commit" — Amazon, Leadership Principles.
• [4] The five RAPID roles (Recommend, Agree, Perform, Input, Decide) and the advice to assign agree roles sparingly — Bain & Company, RAPID Decision Making (13 October 2023).
• [5] Average adult silent reading rate in English of 238 words a minute for non-fiction, most adults between 175 and 300, lower in a second language; 190 studies, 18,573 participants — Marc Brysbaert, How many words do we read per minute? A review and meta-analysis of reading rate, Journal of Memory and Language 109 (2019) 104047.

Build your offsite pre-read in NOXI, free: the six decisions, the five numbers, the impact and effort grid and the decision log, from your planning files and in the style of your own template, exported as an editable PowerPoint or PDF.