ingridleiria Event and Offsite Planning
An offsite consumes a leadership team's entire attention for one to three days, plus travel, plus the fortnight of half-preparation before and the week of catching up after. Priced properly it is the most expensive meeting the organisation holds, and in most companies also the least designed. The agenda is assembled from whoever asked for a slot, the sessions are presentations, the conversation is genuinely good, and three weeks later nobody can name a decision that came out of it.
The cost goes beyond the money. The team learns that the offsite is where things get discussed rather than settled, so the next one attracts lower expectations and less preparation, and the questions that needed a room full of people go back to being handled in fragments. A team that has had three inconclusive offsites has lost its only forum for deciding anything hard.
This skill designs the event backwards from the artefacts it must produce, then handles logistics thoroughly enough that nothing but the content needs attention on the day.
When to use this, and when not to
Use it for any gathering that costs a day or more of several people's time: a leadership offsite, an annual planning session, a team away day, a customer workshop, a client event, a conference presence. Use it when an event is already scheduled and the agenda is a list of topics, the most common state and the one this fixes fastest, and when the last offsite produced nothing and you have been asked to make this one different, where the honest answer usually involves fewer sessions rather than better ones. It applies equally to a half-day, where pre-work matters more because there is no evening to recover ground.
Do not use it for the recurring weekly or monthly meetings that make up the operating rhythm; that is operating-cadence-design, and treating a regular meeting as an event produces ceremony. Do not use it in place of a decision that could be made in a memo: if a question can be settled by decision-memo and a fifteen minute conversation, flying eleven people to a hotel is an expensive way of avoiding writing. Do not use it for a genuinely social event; a celebration needs logistics rather than session design, and dressing it in outcomes makes it worse.
okr-planning and strategic-plan-and-action-plan are usually the substance of a planning offsite; this skill sequences their steps into a room rather than replacing them.
What you need before starting
The outcomes, as artefacts. Two to four things that will exist afterwards and do not now. Missing: this is the first work and cannot be skipped. Ask whoever called the event what will be true on the Monday after; if the answer is a feeling, ask what would have to be written down for it to be justified.
Who decides the agenda, and who owns the event. Usually the chief executive decides content and the Chief of Staff owns delivery; the roles must be distinct or the owner spends the event arbitrating. Missing: name it in writing before booking anything, because agenda authority is what gets contested at the last minute.
The attendee list, final. Numbers drive venue, format and cost, and every extra person changes what the room can do: above about twelve, unstructured discussion stops working. Missing: build the list from the outcomes and challenge any name not tied to one.
The budget, and who approves an overrun. Missing: work from a per-head figure used before, present it as a range, and get approval before contracting anything with a cancellation term. If the budget is cut after commitments are made, reduce venue standard, accommodation nights, external facilitation and food in that order, protecting the days and the attendee list last.
The dates and the blockers. Board meetings, month end, a customer go-live, school holidays, religious observances, notice for people arranging childcare. Missing: circulate three candidate windows before committing, and check the two most senior calendars yourself.
Accessibility and dietary needs. Missing: ask everyone directly and privately, well ahead, about the venue rather than about people: step-free access, hearing loop, quiet space, distance from transport. A requirement discovered on the day lands on one person publicly.
The state of the substance. What is decided, what is genuinely open, what the pre-work must establish. Missing: the highest-risk gap. An offsite that opens a question the chief executive has already closed loses the room in its first session. If an event was called to settle something already settled, say so rather than running it.
The method
Write the outcomes before anything else, and get them agreed. Two to four, each an artefact: annual priorities agreed with owners and numbers; a decision on the pricing change with its reasoning recorded; a leadership team that has met its three new members and worked on something real together. If an outcome cannot be written as something that exists afterwards it is a mood, and moods are legitimate secondary goals but not a reason to fly people in. Rank them: when the day runs late, and it will, the ranking decides which session is compressed, and that call made at 16:40 without one is usually wrong.
Test whether the event is the right instrument. For each outcome, ask what it needs that a memo and a call cannot provide. The legitimate answers are real-time disagreement between people who avoid it in writing, the whole group hearing the same thing at once, or relationships built through working rather than talking. An outcome with none of those comes off the agenda and is handled beforehand, which usually shortens the day and improves it. At very short notice, under two weeks, apply this hard: cut to one outcome and treat the rest as logistics.
Design each session from its output, working backwards. Before timing is set, write four things per session: the output it must produce, the inputs and who prepares them by when, the format, and the facilitator. A session without a named output is a discussion, and discussions expand to fill the time given.
Assign the pre-work with names and dates, and enforce it. Pre-work is the largest single determinant of whether a session decides or discusses. The rule that makes it real: any session whose pre-read is not circulated 48 hours before is cut, or converted into a working session that produces the missing input. State the rule when the pre-work is assigned and apply it once; after that it is never tested again. Send pre-reads as documents to be read, not decks to be presented; presenting material people could have read is the most expensive use of an offsite hour.
Choose formats deliberately and vary them. Three consecutive presentations lose the room, and the third is wasted. Presentation only where information must be transmitted and cannot be read; individual writing before discussion where the goal is independent views rather than the loudest one; small groups then plenary for anything needing everyone's contribution above eight people; a structured decision process for anything that must be settled; open discussion only to surface unknowns. For the highest-stakes item, silent reading at the start of the session beats pre-work, because it guarantees everyone has read it.
Build the day plan with the hard thing early and real buffers. Put the hardest decision before lunch on day one, when attention is highest and there is time to recover if it goes badly. Break every ninety minutes and leave a fifth of the schedule unallocated, since agendas that fill every minute produce a rushed close and the close is where value is captured. Place social time deliberately: dinner after the hardest session lets people repair, dinner before it makes the session harder to open.
Assign a facilitator and a separate capturer for every session. The facilitator runs the flow and holds the time; the capturer records decisions, actions and owners in real time using
meeting-to-decisions, visibly where the room can see them. One person cannot do both, and the usual result of trying is a day of good conversation with no record. Where a facilitator has a stake in a session's outcome, swap them out: someone arguing their own case cannot hold the room and the group knows it.Protect the close. The final session reads back every decision and action with owner and date, out loud, and confirms each. Forty-five minutes, scheduled first and defended, never the item that absorbs the overrun. This is where an offsite becomes a set of commitments rather than a memory.
Run the logistics from one tracker with an owner and a date per line, reviewed weekly from booking and daily in the final week, so nothing but the content needs attention on the day.
Follow up on a fixed schedule. Within 24 hours, the record circulated and the decision log updated. Within a week, actions loaded into whatever tracker the organisation already uses, plus a participant survey and the owner's retrospective. At 30 days, a check on the actions reported through
weekly-status-update. An offsite whose actions are not checked at 30 days was a retreat, and participants will treat the next one as optional.
Session design
| Field | What it means | | Output | The decision, document, list or shared understanding that exists afterwards | | Inputs | The pre-read or data required, its author, and the date it circulates | | Format | Presentation, writing then discussion, small groups then plenary, structured decision, open discussion | | Facilitator | Who runs it, with a written flow and timings | | Capturer | Who records decisions and actions | | If it overruns | What gets cut, decided now rather than in the room |
The last field is the one people skip and the one that saves the day: deciding in advance what a session gives up if it runs long turns a mid-afternoon panic into a decision already made. For any session that must decide something, write the decision as a closed question in advance using decision-memo and circulate that memo as the pre-read, since sessions that arrive without the question written spend thirty minutes agreeing what is being decided.
Logistics, budget and the run sheet
Venue. A layout matching the formats you chose, the check people forget: theatre seating cannot do small groups and a boardroom table cannot do a workshop. Natural light, breakout space, wall space for capture, connectivity tested rather than promised, a quiet room. Visit if you can; if not, ask for photographs of the room itself.
Travel, food and equipment. Arrival times that do not require a 05:00 flight before a day of decisions, and where the group is distributed, an explicit view of who bears the travel burden, rotated rather than always falling on one region. Dietary needs collected privately and confirmed with the venue in writing. Screens, adapters, whiteboards, pre-reads printed locally.
Budget and contracts. A per-head figure built by line, with ten to fifteen percent shown as contingency rather than hidden in the lines. Where it is presented visually, keep it monochrome, separate categories by texture rather than colour, and put the legend outside the plot area so it survives being printed or projected badly. Record the cancellation terms and the last date to reduce numbers: that is the difference between a postponement costing nothing and one costing the venue fee in full.
Communication and the run sheet. A joiner pack a week ahead carrying the outcomes, agenda, pre-work deadlines, travel, dress and the contact for the day, because participants who know what the event is for prepare differently. Then a run sheet, minute by minute: times, session, who speaks, what is set up, what moves and who moves it. It is not the agenda. It exists so the owner never has to think on the day, which is what lets them notice the room.
Client events and conference presences
The structure holds; the outcomes become commercial, as meetings booked and qualified conversations captured. Add four things: a target list of the people you want to attend, built before the invitation; an invitation and reminder sequence with named senders, since an invitation from a person outperforms one from an organisation by a wide margin; a staffing rota, so the stand is never unattended and never crowded with your own people talking to each other, the most common failure of a conference presence; and a capture process agreed in advance, recording each conversation on the day rather than from memory afterwards. Follow-up goes within 48 hours, personalised to the conversation, because a week's delay turns it cold.
Worked example
Situation. A healthcare technology company of about 250 people planned a two-day leadership offsite for eleven, six weeks out, at a venue ninety minutes from the office. The chief executive's brief was "align the leadership team on next year". The previous year's offsite had produced a well-received strategy discussion and no decisions; two of its three actions were never started, and the head of engineering had said privately he would rather ship for two days.
Task. End with the annual priorities agreed with owners and numbers, and a resolution on whether to build or buy a data integration layer, open for five months and already through three leadership meetings.
Action. The outcomes were written first and cut from six to three. The discarded ones included "a shared understanding of the market landscape", which failed the artefact test, and a session on the performance review framework, which needed no cross-functional debate and became a document circulated the week before plus fifteen minutes for questions, freeing ninety minutes. Build-or-buy was ranked first: it was the one thing that genuinely needed everyone in a room, two leaders having avoided a direct disagreement in writing for months.
The wrong turn came early. The first agenda gave that question a ninety minute session on day one, called "build vs buy: discussion and decision". Tested against the session design template it had no defined inputs: the buy option was uncosted, the build estimate was six months old, and neither leader had written down what they thought the decision hinged on. That session would have reproduced the previous three meetings at a much higher cost per hour.
It moved to day two and the fortnight before was redesigned around producing the inputs. The head of engineering and the chief technology officer co-wrote one decision memo with a closed question, three options including doing nothing, costs in ranges, and a statement of what each believed the decision turned on. It arrived 30 hours before the event rather than 48, so the rule was applied in modified form: the session stayed, its first thirty minutes became silent reading, and the facilitator was swapped from the chief technology officer, who had a stake, to the chief financial officer, who did not.
The rest of the agenda put the priorities work before lunch on day one, in small groups then plenary because eleven people is above the threshold where open discussion works. The main room was rebooked from a boardroom layout after a photograph showed a fixed table that would have made small groups impossible. The close was scheduled for fifty minutes at 16:00 on day two and protected when build-or-buy ran over; what gave way was a session on hiring sequencing, designated in advance as the thing to cut and handled in a memo the following week.
Result. Seven annual priorities agreed with named owners and numeric targets, down from a first draft of fourteen. Build-or-buy was decided in the room: buy, with a twelve week integration pilot and a stated reversal condition if latency targets were missed, recorded with its reasoning in the decision log.
The record went out 19 hours later. At 30 days, six of the seven priorities had their first milestone met or credibly in progress, and the seventh was formally deferred rather than quietly dropped, itself a change from the previous year. The survey returned one complaint, that day one ran too long and dinner at 19:30 was too late after an 08:30 start; the next event started at 10:00.
The buy decision was reversed nine months later when the vendor's roadmap changed, which the reversal condition had anticipated but not for that reason. The value of the offsite was not that the decision was right. It was that it was made, recorded with its reasoning, and could be revisited on evidence rather than by attrition.
A second scenario, where it goes differently
The same company, a one-day gathering for the whole 250-person organisation, held in a hired hall, with a plenary format and no possibility of small-group decision work.
Almost every design choice inverts. The outcomes cannot be decisions, because 250 people do not decide anything. They become communication and connection outcomes in testable form: every person can state the year's three priorities and how their team contributes, and every person has had a substantive conversation with someone from a team they do not work with. Both are checkable in the follow-up survey, which keeps them from being moods.
The agenda is built around a few well-prepared presentations, against the offsite rule about avoiding them, because transmission genuinely is the goal. What replaces the decision sessions is structured interaction: table groups mixed by function, each with a task and a written output, so the connection outcome has a mechanism rather than a coffee break. Capture inverts too: with no decisions to record, what is captured is the questions asked, the highest-value output of an all-hands and the one most reliably lost, grouped and answered in writing within a week.
What changed is the group size and therefore what the room can do. Above roughly twelve people a room cannot decide; above about forty it cannot discuss. Design for what the size permits rather than running an offsite format at a scale where it cannot work.
Output
The outcomes statement, one page, agreed and ranked:
| Rank | Outcome, as an artefact | How we know it exists | Owner after the event |
The agenda, one row per session:
| Time | Session | Output | Format | Facilitator | Capturer | Pre-read and author | If it overruns |
The pre-work and logistics trackers:
| Item | Author | Circulate by | Sent? | Session it feeds |
| Item | Owner | Due | Status | Cost | Cancellation terms |
The capture template, per session:
| Decision | Owner | Date | Reasoning in one line | | Action | Owner | Due | First checkpoint | | Parked item | Why | Who picks it up |
Plus the joiner pack, the run sheet, and the follow-up plan: record within 24 hours, actions and survey within a week, 30 day check in the weekly status.
Failure modes
Outcomes that are moods. Recognise them by the words align, energise and share, and by agendas whose sessions are named after functions rather than outputs. Fix by asking what will be written down afterwards that is not now, then rebuilding the agenda from the answer.
No pre-work, then a session that produces it. Recognise it when the first thirty minutes go on establishing facts. The commonest way an expensive hour is wasted; the fix is the 48 hour rule, applied once, in public.
Decisions taken and not recorded. Recognise it when the record takes more than 24 hours to appear, since reconstruction cost rises steeply after the first evening. Fix with a named capturer per session and a visible running list. Actions then die at day 10 unless the 30 day check goes into the weekly status the day the event ends.
The close absorbed by an overrun. Recognise it when the last session is the one shortened. Fix by designating the sacrificial session in advance and scheduling the close before anything else.
The venue that cannot do the format. Recognise it from a photograph showing a fixed table. Check the layout against the chosen formats before signing.
One person facilitating, capturing and arguing. Recognise it when the event owner is also making the case. Split the roles and swap the facilitator out of any session where they have a stake.
Edge cases
A distributed team where some cannot travel. Do not run a hybrid decision session as an afterthought; remote participants in a decision room are observers. Either bring everyone, or make the decision sessions fully remote and use in-person time for what only presence provides. Prepare a remote fallback anyway, because someone will be stuck at an airport.
The chief executive changes the agenda on the morning. This will happen. Hold the ranked outcomes and ask which outcome the change serves and which session it displaces; the ranking makes it a thirty second conversation rather than a negotiation. Agree in advance what interrupts in a crisis: normally the chief executive and one named person step out while the rest continues, with a stated point at which the event is suspended.
The offsite is really about a conflict between two people. Say so privately beforehand. A group session will not resolve it and attempting one publicly damages both. Handle it separately, and design the agenda so the event does not depend on the resolution.
The venue or the dates fall through late. Check the cancellation terms, then downgrade the venue rather than lose the dates: an ordinary room in the right week beats a good room after the quarter starts.
Quality bar
- Every outcome is an artefact that will exist afterwards, and the outcomes are ranked.
- Every session has a named output, facilitator and capturer, and a pre-read with an author and date.
- The pre-work rule is stated in advance and applied.
- The closing session is scheduled first, protected, and reads back every decision and action with owner and date.
- The sacrificial session is designated before the day, not chosen during it.
- The venue layout was checked against the formats and connectivity was tested rather than promised.
- The budget carries a stated contingency and the cancellation terms are recorded.
- The record is circulated within 24 hours, actions tracked within a week, and a 30 day check scheduled the day the event ends.
Adapting this to your context
The defaults come from leadership offsites of eight to twelve people at commercial companies, held over one or two days at a hired venue.
- The twelve-person threshold. Above roughly twelve a room stops deciding and above forty it stops discussing. Both hold well, but a group used to formal chairing, such as a trustee board or a works council, can decide at twenty, while an unchaired group of ten often cannot.
- The 48-hour pre-read rule. Written for people whose reading time is discretionary. Where attendees are clinical, shift-based or teaching, circulate a week ahead and expect the rule to be tested harder, because they cannot simply read it the evening before.
- The budget shape. A per-head figure with ten to fifteen percent contingency assumes commercial spending discretion. Public and charitable funders often cap venue and catering spend or require competing quotes; get that rule before you shortlist venues.
- The 30-day action check. Fits a monthly operating rhythm. Where the cycle is a term or a season, set the check at the next natural review point and put it in the calendar the day the event ends.
- What not to change. Every outcome is an artefact that will exist afterwards, and the closing session is scheduled first and reads every decision and action back aloud.
Related skills
meeting-to-decisions is the capture discipline used inside every session and produces the record circulated afterwards. decision-memo writes the closed question and options for any session that must decide, and is the right pre-read for it. okr-planning and strategic-plan-and-action-plan supply the substance of a planning offsite and receive its outputs. operating-cadence-design covers the recurring meetings this sits outside, where an offsite's follow-through actually lives. program-management takes actions large enough to need a plan. weekly-status-update carries the 30 day check. tailored-client-deck produces client-facing materials.