How to Use a Visitor Forecast for Museum Staffing and Planning

Most museums already plan around visitor numbers — usually from last year's figures, a sense of the season, and experience. A forecast does not replace that judgement. It gives it a number, a range, and a reason, early enough to act on.

This page covers what to actually do with the forecast once you have read it. If you are not yet sure what the chart, band and table are telling you, start with how to read a museum visitor forecast.

Match the forecast length to the decision

Different decisions have different lead times. Use the forecast length that matches the moment the decision is locked.

Short forecasts are normally the most accurate, so the closer the decision sits to the day it affects, the more precise the input you can give it.

Staffing and rosters

The forecast table gives you a predicted number and a range for each open day. Turning that into people takes one piece of local knowledge: roughly how many visitors one person can serve well in each role.

You do not need a formal model for this. Pick a handful of past days at different levels — a quiet Tuesday, a normal Saturday, a peak vacation day — note how many staff were on and whether it felt right. Those become your reference points. The forecast then tells you which reference day each coming day resembles.

How much buffer to add depends on what an error costs:

• Plan closer to the upper end of the range where being short-staffed damages the visitor experience or breaks a rule you cannot bend — ticket desk queues, gallery supervision, cloakroom on a peak day, safety-related cover.
• Plan to the predicted number where over-staffing is the more expensive mistake, and keep flexible cover available — an on-call shift, or a colleague who can move from a back-of-house task.

Also read the Day type column. A public holiday or the first weekend of a school vacation behaves differently from an ordinary day at the same visitor count — arrivals bunch up, group size changes, and the same total can feel much busier at the entrance.

Opening hours and capacity

The 90-day forecast is where opening-hour questions belong. Look for stretches where predicted numbers stay low across several weeks: those are candidates for reduced hours, maintenance, staff training or gallery work. Stretches that stay high are where extended hours, timed entry or an extra late night pay off.

If you are considering timed-entry slots or a capacity cap, use the upper end of the range rather than the predicted line. Capacity decisions are about the busy case, not the average one.

Café, shop and consumables

Ordering runs on lead time, so the 30-day forecast usually drives it. Two habits help:

• Order against period totals, not individual days. Totals are more reliable than single days, because days that land high and low cancel each other out.
• Watch the mix, not just the volume. School vacations, a family exhibition and a conference audience produce very different café and shop spend at the same visitor count. The driver cards tell you which kind of demand is coming.

Marketing and programming

The 90-day forecast is a campaign-planning tool. The quiet weeks it shows are the ones worth promoting — a campaign aimed at a week that was going to be busy anyway mostly moves visitors around rather than adding them.

The driver cards make this concrete. If they show that exhibitions are pulling the coming period below a typical week, that is a programming signal, not just a forecast. If school vacations are lifting a period sharply, that period may need capacity management more than marketing.

After a campaign, compare what happened with what was forecast before it launched. The gap between the two is the closest you will get to a clean read on what the campaign added.

Budgets and board reporting

For finance conversations, use the period total rather than the daily detail, and quote the range alongside it. "We expect about 18,000 visitors next month, with a realistic range of 16,000 to 20,000" is a more useful sentence than a single number, and it is the one you can defend.

The Forecast track record card is what makes that credible. It shows how close past forecasts came to the visitor numbers you later confirmed, using your own data — including a separate figure for how accurate whole-period totals have been. Bring that figure to the meeting with the forecast.

Budget conversations rarely stop at what is coming, though. The next question is usually why the last period turned out the way it did. That is what the Analyse trends page is for: pick two periods — this spring against last spring, this quarter against the one before — and VisitorCast shows the difference in visitors per open day and splits it into named drivers, with whatever it cannot attribute shown honestly as Unexplained.

It is the report to open before a board meeting or a budget review. Alongside the driver split it shows how the two periods compare day by day, whether the underlying level of attendance has shifted rather than just fluctuated, and how your weekly rhythm has changed. That turns "we were down 4% last quarter" into "we were down 4%, and most of it was a weaker exhibition programme and one fewer vacation week" — a sentence that leads somewhere.

Build a weekly rhythm

The teams who get the most out of forecasting do the same small loop every week:

1. Generate a fresh forecast once your latest visitor numbers are uploaded.
2. Read the trend card and the driver cards — has anything changed since last week, and does the explanation match what you know?
3. Confirm next week's roster from the 7-day forecast.
4. Adjust the coming month from the 30-day forecast.
5. Check last week — where was the forecast off, and was there a reason worth recording as an event?

Step 5 is the one most teams skip, and it is the one that compounds. Every surprise you record as an event or exhibition becomes a pattern the model can use next time. See how to get the best forecasts.

The honest limits

A forecast is a qualified guess, not a fact. This is the most important thing to understand before you plan on it.

VisitorCast produces the best estimate it can from the data it has: your visitor history, your opening days, your exhibitions and events, and the signals you have enabled. That is genuinely useful — it is a far better starting point than memory or last year's figures alone. But it remains an estimate about a future that has not happened yet, and it will sometimes be wrong. Not slightly wrong. Occasionally properly wrong.

Never treat a forecast as the truth. Treat it as the most informed opinion available, to be weighed against everything you know that the model does not.

And there is a great deal the model does not know. It has no idea about a rail strike, a burst pipe, a road closure outside your entrance, a heatwave, a viral post, a competing festival across town, a coach party that cancels, or a national news event that keeps everyone at home. Unless that kind of thing is recorded in your data as a repeating pattern, it simply does not exist as far as the forecast is concerned. Your team knows about most of these. The model never will.

So use it accordingly:

• Keep your own judgement in the loop. If you know something the forecast cannot, you are right and it is wrong. Override it without hesitation.
• Do not make an irreversible decision on a forecast alone. Anything expensive, contractual or hard to undo — seasonal hiring, a permanent hours change, a large non-returnable order — deserves the forecast as one input among several, not as the deciding vote.
• Keep a fallback for the days that matter. Where being wrong would genuinely hurt, plan so that a surprise is recoverable: flexible cover, an on-call shift, stock you can hold back.
• Expect the misses. Check the forecast track record so you know the realistic size of a typical error for your museum, and build your plans with that much room in them.
• Never present a forecast as a promise. In a board paper or a budget, quote the range and say plainly that it is a projection. A number presented as certain will be remembered as a commitment.

The right test is not whether the forecast was exactly right — it never will be, and any tool claiming otherwise is overselling. The test is whether planning with it consistently beat planning without it. For most museums it does, and that is the honest claim VisitorCast makes.