How to compare your event results with industry benchmarks

You've finished the event, pulled the numbers, and found an industry report that seems to tell you whether you did a good job. Your attendance rate is higher than the average. Your marketing cost is lower. That sounds reassuring, until someone asks what those averages actually include.

Were those events paid or free? Were they counting ticket holders, purchasers, or everyone who registered? Did the marketing cost include the people who made the ads? If you can't answer, you haven't yet established whether the comparison means anything for your business.

Use event performance benchmarks to identify questions worth investigating. Before turning one into a target, check who was measured, what was counted, when it happened, and how the result was calculated. Then compare your own similar events on the same basis. A credible external reference can help you see a weakness you've accepted for too long, but it shouldn't decide your budget without that work.

This guide walks through that comparison, including a worked example where an event's headline attendance rate improves while attendance falls within both of its ticket groups. You'll also get a practical record you can copy, questions to send a report publisher, and a way to turn a finding into a decision you can afford.

Start with the decision behind the comparison

The first thing you want to do is say what you might change. Are you deciding whether to spend more finding new buyers, revise your arrival information, increase the ticket price, or keep the same venue? Each decision needs different evidence. Collecting every average you can find makes the review longer without necessarily making the decision better.

Suppose you're worried that too many paid ticket holders don't attend. A comparison with similar paid events could help you judge whether to investigate. An average attendance rate for free online sessions tells you much less. Those participants made a different commitment, and attending may require nothing more than opening a link. The percentage alone doesn't make the experiences comparable.

Separate your reference from your requirement. A benchmark describes a result in a defined group. A target is the result you're trying to achieve. Your minimum acceptable financial result comes from your own revenue, costs, commitments, and appetite for risk. A published average can't make an event affordable when your budget says otherwise.

You can be above a relevant average and still need to improve. Perhaps your attendees are experiencing an entry problem that the comparison doesn't measure. You can also be below an average while making a sensible choice, such as paying more to reach a new audience whose purchases remain worthwhile. Name the decision before deciding which direction is better.

Our event KPI guide explains how to choose the measures that matter before, during, and after an event. Here, the next question is whether another result gives you a fair reference for one of those measures. Keeping those jobs separate stops a useful industry report from becoming a collection of targets nobody agreed to pursue.

Write a sentence before opening the report: "We need to decide whether to change our paid-attendee arrival process before the next edition." That is specific enough to guide the search. "We need to improve engagement" leaves almost any percentage looking relevant, even when it describes something you don't sell or control.

Find out who the benchmark actually describes

Read the methods before the headline. Look for the population the publisher wanted to study and the people or organizations that supplied the data. A survey of event professionals can include people running internal meetings, free community programmes, exhibitions, and paid conferences. Their answers may be valid while having limited use for your particular decision.

Ask whether the data describe events, companies, campaigns, or individual attendees. These units aren't interchangeable. An organizer who runs many editions can answer one survey question about their business. That answer doesn't become a separate observation for every event they delivered. Equally, thousands of ticket records from one large festival don't represent thousands of different event businesses.

Here's an example from our own research. The Loopyah Event Creators Survey 2026, fielded through Pollfish on 11 February 2026, includes 350 organizers. The screening responses identify 227 paid-ticket organizers and 123 free-ticket organizers. That is useful context for reading the report. It also tells you why the full sample shouldn't automatically become a benchmark for businesses selling paid admission.

If a finding uses all respondents, describe it that way. Don't silently rewrite it as a finding about paid festivals or commercial conferences. You would need the relevant subgroup result and its question wording to make that narrower statement. A report can inform your thinking without supplying a target for every event category.

Next, check the event format and buying situation. An established local music night, a destination conference, and a specialist workshop can all sell tickets, yet their attendance commitments, sales periods, and delivery costs differ. A comparable reference needs to match the factors that could reasonably change the measure you're assessing.

For attendance, paid versus free admission and travel requirements may be central. For acquisition cost, the distinction between previous buyers and people discovering the event may matter more. For catering costs, duration and what the ticket includes can change the comparison. Choose the relevant characteristics for the decision instead of requiring every event to be identical.

Don't reject a source just because it covers a broader population. Use it at the level it supports. It might identify an issue to investigate or help you ask a better question. The limit is claiming that its average establishes what your own event should achieve when the source doesn't provide that connection.

Match the calculation before comparing the percentage

Two reports can use the same label and calculate different things. "Attendance rate" might mean checked-in guests divided by valid tickets, attendance divided by venue capacity, or people who joined a session divided by everyone invited. Ask for the actual calculation. If it isn't available, keep the comparison provisional.

For a paid event, start by reconciling gross ticket sales, refunds, remaining valid paid tickets, complimentary tickets, and recorded entry. Decide which groups belong in the measure. Then check whether the external source treats those groups the same way. A rate based on paid ticket holders can't be compared directly with one that includes a large complimentary allocation.

The top and bottom of a calculation must describe the same group. If you count complimentary guests among arrivals, include the relevant complimentary tickets in the population you're comparing them with. If you report paid-ticket attendance, use paid-ticket arrivals and valid paid tickets. Don't choose whichever combination produces the most flattering percentage.

Orders create another easy mistake. A purchaser buying several tickets creates one order, several tickets, and potentially several attendees. Marketing cost per order, cost per ticket, and cost per new customer answer different questions. A report using one of those measures won't tell you whether you're above or below another, even if both headings say acquisition cost.

Use the same care with money. Establish whether revenue is before or after refunds, whether transaction fees are deducted already, and whether taxes collected for authorities are excluded. For costs, check the treatment of staff time, production, creative work, and shared business expenses. A cheap-looking campaign may simply leave its creative bill elsewhere.

Profit percentages also need their calculation. Profit divided by revenue is a different measure from profit divided by cost. Neither label should be left for the reader to guess. When the reference includes a formula, reproduce that basis in your internal comparison or clearly explain why you can't.

Once your own measure is defined, keep a short note with the report: what counts, what doesn't, the source record, and the reporting cutoff. This shouldn't require rebuilding data your systems already provide. The work is agreeing how to interpret those records and identifying exceptions that need attention.

If the benchmark uses an incompatible definition, there are two honest options. Recalculate on a common basis when the underlying data support it, or explain that the figures answer different questions. Don't estimate missing parts just to make the percentages line up. An unresolved comparison is more useful than a precise-looking conclusion built on a guess.

Check the sample and the voices it may miss

A large sample is encouraging only when you understand what it contains. Find the number contributing to the particular figure you're using. The report's cover might describe a broad research project, while a specific average uses a much smaller group that answered one question. Look for the base beside the table, not just the total in the introduction.

For survey evidence, check how participants were recruited, who could take part, and whether results were adjusted to reflect a wider population. The American Association for Public Opinion Research's survey research guidance recommends transparency about the population, sample construction, recruitment, question wording, and analysis. Those details let you judge what the finding supports.

Apply that scrutiny to your own attendee feedback too. People who reply may have different experiences from people who leave the message unopened. A survey sent only to purchasers can miss guests whose tickets somebody else bought. A survey handed out in the final session can miss people who left early. These are specific coverage limits to investigate.

A report based on platform customers has a different boundary. It describes the eligible records available through that platform under the publisher's inclusion rules. That can provide useful operational evidence, particularly when the measures are well defined. It doesn't automatically cover organizers using other systems or events that the publisher excluded.

Ask whether cancelled events, weak-selling editions, incomplete records, or unusual outcomes were removed. Some exclusions improve data quality. Others change the question being answered. A benchmark drawn only from completed events won't describe the risk of starting an event that never reaches opening day. Keep the inclusion rules attached to the result.

Response rate deserves attention, but it doesn't settle quality by itself. AAPOR's response-rate overview explains that response rates do not reliably distinguish accurate from inaccurate survey estimates. Read the other quality information and look for groups that may be missing. Don't treat a higher reply rate as proof that a result represents everyone.

There is no single minimum sample size that makes every event benchmark dependable. The useful amount depends on the question, variation, sampling method, and consequence of the decision. If a small reported difference could trigger a large commitment, ask for the publisher's uncertainty information or qualified analytical help before making a confident ranking.

You can still act on direct evidence of a problem. A documented access failure needs attention even if your survey is too small to rank you against the industry. Statistical uncertainty about the average doesn't cancel a specific obligation to deliver the experience you sold. Keep those decisions separate in the review.

Understand what the reported average gives weight to

An average can describe the typical event or the combined experience of all recorded ticket holders. Those are different questions. Before comparing, establish whether the publisher averaged event-level rates or added all the underlying counts together. The choice can change the result when event sizes vary.

Consider a separate, fictional attendance example. One event has ten valid tickets and all ten holders attend. Another has 990 valid tickets and 792 attend. Their event-level rates are 100% and 80%. Giving each event equal weight produces a 90% average. Combining the records gives 802 arrivals from 1,000 valid tickets, or 80.2%.

Neither calculation is a mistake. The first gives the small event as much influence as the large one. The second gives every ticket equal influence. If you run a small event and want to understand comparable operators, the pooled rate may be dominated by events unlike yours. If you're forecasting total attendance across a portfolio, the pooled count may answer the immediate question better.

Then check whether the published figure is a mean or a median. A mean adds the values and divides by their number. A median is the middle value when they are put in order, or the mean of the two middle values when the count is even. The US National Institute of Standards and Technology's guide to measures of location explains why extreme values can affect the mean more than the median.

In another invented example, five event rates are 50%, 55%, 60%, 65%, and 95%. Their mean is 65%, while the median is 60%. If your result is 62%, you sit above one summary and below the other. You haven't changed performance; you've changed the reference. Ask which calculation the report actually uses.

A range or distribution can be more useful than a single central value. Find out whether similar events cluster closely or differ widely. A percentile describes a position within the reported group, not a guarantee about what another business can achieve. Check the direction too. A high percentile for cost may describe an expensive result rather than a desirable one.

Don't reverse-engineer detail that wasn't published. A median doesn't reveal the highest and lowest result. A mean doesn't reveal how many events performed worse than yours. If the report only gives one figure, acknowledge that limited view and avoid presenting your event as a proven industry leader on that evidence.

Compare equivalent dates and reporting windows

A new publication date doesn't necessarily mean recent activity. Check when the data were collected and which event dates they cover. An annual report may contain events delivered under different prices, travel conditions, or measurement systems from the ones you're planning now. Record the activity period as well as the date on the cover.

Your own timing matters just as much. A ticket-sales comparison should use equivalent points before the event and account for time on sale. Sales four weeks before doors can mean something different when one edition launched months earlier and another only recently opened. Compare the actual opportunity to buy before deciding that demand improved or weakened.

For post-event results, use the same maturity. A financial report with outstanding invoices isn't directly comparable with one whose final costs and refunds are settled. An attendee survey open for a few hours isn't equivalent to one left open through several reminders. State the cutoff and replace provisional figures when the missing records become available.

Repeat purchasing needs an eligible opportunity too. A buyer can't purchase the next edition before it exists or goes on sale. If you compare retention across events, specify which buyers could return, which later events counted, and how long you observed purchases. Otherwise, differences in the calendar can look like differences in loyalty.

Measurement technology can change what appears in the report. Apple's Mail Privacy Protection explanation says protected Mail activity can download remote content in the background regardless of engagement. That is a concrete reason to be cautious about treating a tracked email open as a person choosing to read your message.

For email benchmarks, ask how opens and clicks are defined, what automated activity is filtered, and whether the methods changed during the comparison period. Don't invent an adjustment when the publisher doesn't provide one. If your systems measure different activity, acknowledge the difference and rely on other evidence for the spending decision.

With Loopyah attendee email, campaign reports show tickets and revenue associated with people who clicked, alongside send cost. Use those records to examine a commercial outcome beyond opens. You still need to understand the report's counting basis and refunds; recorded campaign sales alone don't prove that every purchase was caused by the email.

Keep a note whenever you change a question, report, tracking setup, or counting rule. Try to preserve a comparable view where possible. If you can't, mark the break and start the new series honestly. A clean-looking trend isn't worth much if the apparent improvement came from changing what you counted.

Work through an example before judging the headline

Let's use one fictional paid conference, Northbank, to connect these checks. All figures in this example are invented teaching assumptions, with money in USD. They aren't customer results, industry averages, or forecasts. Northbank wants to know whether its latest attendance result justifies leaving the arrival process unchanged for the next edition.

Both editions have 500 valid paid tickets after refunds. The organizer records one arrival per ticket and separates group-booked tickets from individually booked tickets. These are booking categories, not assumptions about why people attended. Complimentary tickets are excluded from every figure in this comparison, and the team has checked that the entry records are complete on the same basis.

In the earlier edition, 100 paid tickets belong to group bookings. Of those, 96 are used, giving a 96% attendance rate. The other 400 paid tickets are individual bookings, and 320 are used, giving an 80% attendance rate. Total recorded attendance is 416 out of 500, or 83.2%.

In the latest edition, the booking mix changes. Group bookings account for 400 paid tickets, of which 376 are used, giving a 94% attendance rate. Individual bookings account for 100 tickets, of which 78 are used, giving a 78% attendance rate. Total recorded attendance is 454 out of 500, or 90.8%.

The headline improves by 7.6 percentage points. Northbank welcomes 38 more paid ticket holders. Those are real outcomes in the example. Yet attendance falls within both categories: group bookings decline from 96% to 94%, while individual bookings decline from 80% to 78%. The stronger overall rate doesn't demonstrate a stronger arrival process.

Here's what's interesting. The latest event contains far more of the booking category with the higher attendance rate. That change in mix raises the combined result even though both category rates are lower. If the team stops at the headline, it can miss a pattern that deserves investigation before the next edition.

For an additional comparison, Northbank can calculate what the rates would look like under the same chosen mix. If both editions had an equal share of group and individual tickets, the earlier category rates would give 88%, and the latest would give 86%. That calculation holds the mix constant to make one difference easier to inspect.

Keep that adjusted comparison beside the actual totals. Northbank really did record 454 paid arrivals, not an adjusted headcount. The equal mix is an analytical choice, not a prediction of future bookings. It accounts for the two categories in this example, but it doesn't account for every difference between the events or establish the cause of the decline.

The useful conclusion is limited and actionable. Attendance within both recorded booking categories was lower, while the overall result improved because the mix shifted toward group bookings. Investigate what changed within each category before declaring the current process successful or deciding how many people will arrive next time.

What if an external report tells a different story?

Imagine Northbank also finds a published attendance benchmark. Before adding it to the comparison, the team needs the report's paid-ticket basis and booking mix. If the publisher supplies only a combined rate, Northbank can't tell whether the difference comes from performance within comparable groups or from the proportion of group bookings. An impressive sample size doesn't supply that missing breakdown.

The report might still be useful. If it covers similar paid conferences with complete entry records, the combined figure can provide broad context. Northbank should describe the remaining booking-mix difference beside it. If the report includes free registrations or counts repeat entries, the team needs a compatible result before using it to judge paid-ticket attendance.

Suppose a second source appears to disagree with the first. Read the definitions before choosing the figure you prefer. One may report the median conference, while the other pools all eligible tickets. One may cover local events and the other destination conferences. Those differences can explain disagreement without either publisher having made a calculation error.

Keep each source's conclusion separate. Don't average the published percentages to create a new industry figure when the populations, periods, and methods differ. That new number would have no clear population behind it. Record which source best matches the decision and which details remain unresolved. If neither matches closely enough, use Northbank's verified internal results for the immediate review and label the external evidence as context.

Build your internal comparison without rewriting history

Comparable internal history is often your best starting point because you can inspect the records and explain the operating conditions. It also has limits. Your own previous events can differ in price, venue, programme, audience, weather, or travel demands. Familiarity with the event doesn't make every edition a fair comparison.

Choose the characteristics before looking for the most flattering result. For Northbank, the paid-ticket basis, booking category, entry-record completeness, and reporting cutoff are essential. Other comparisons may need similar event duration, price bands, or purchase lead time. Write down why each characteristic matters to the measure instead of collecting every possible detail.

Keep a small set of genuinely comparable editions and show the important differences. Don't quietly remove a weak year because it makes the trend awkward. If you exclude an edition, explain the reason and keep its result available. A cancellation or disrupted event might not belong in a normal-delivery average, but it still matters to the business's risk history.

Preserve the raw totals as well as the rates. If a result looks odd, you need to see whether the numerator, denominator, or classification changed. Keep unknown categories visible. Guessing that an unclassified purchase belongs to the highest-performing group makes the report tidier while weakening the evidence behind your next decision.

For surveys, preserve the question and collection method where you want to measure change. The AAPOR guidance linked earlier advises keeping wording, framing, and methodology as similar as possible across comparisons. If you replace a question about overall satisfaction with one about value for money, label the new measure. Both questions can be useful without being points on the same trend.

Be realistic about small histories. A few editions can show recurring practical issues and help you plan, but they don't establish a universal range for your event type. Look at the conditions around unusual results and avoid presenting a narrow internal sample as a market-wide standard. Expand the record as new comparable events finish.

Use external evidence to challenge internal habits. If your team has accepted a delivery problem for years, comparing only with itself can make that problem look normal. A relevant peer reference, the promise on your ticket page, and direct attendee accounts can all expose a weakness. Internal history is a useful reference, not permission to repeat a known failure.

Turn the difference into an investigation and a costed choice

A benchmark comparison tells you where a result differs. It doesn't automatically tell you why. Northbank's lower attendance within both categories might relate to travel disruption, a different audience, unclear information, or incomplete assumptions about who would use group tickets. The example's rates alone can't choose between those explanations.

Start with evidence you can verify. Check whether the venue or start time changed, whether the arrival instructions reached the intended ticket holders, and whether the ticket description matched the practical journey. Read relevant attendee accounts alongside staff notes and entry records. Keep an explanation marked as a possibility until the evidence supports it.

HM Treasury's Magenta Book guidance on evaluation explains the role of a suitable comparison in estimating what would have happened without a change. Applied to your event, that means an improved result after a new reminder isn't by itself proof that the reminder caused it. Other conditions may have changed too.

You don't have to commission a research project before fixing an obvious error. Correcting an inaccurate venue address is justified by checking the address. For a more expensive or uncertain change, describe what you expect it to solve, what you'll spend, and what evidence you'll collect. Match the strength of the claim to the comparison you can actually make.

Suppose Northbank considers a fictional $600 addition to its arrival communications and staffing. The team should first verify what work is additional and whether that figure includes all associated costs. Then ask what benefit would justify it. More paid ticket holders using tickets they already purchased does not create another round of ticket revenue.

There may be other benefits, such as delivering the promised experience more reliably or avoiding documented support costs. Record those honestly. If the team expects later repeat purchases, it should identify the assumptions and observe actual eligible purchases over a defined period. Don't multiply extra arrivals by the ticket price and call the result new revenue.

For illustration, assume a genuinely additional later ticket would leave $60 after the costs tied to that sale, before continuing business overhead. Ten such purchases would recover $600 on that narrow basis. This is a requirement to test, not a forecast, and it doesn't establish that better arrival communications will produce those purchases or that future delivery capacity is free.

The team can decide to make the change for service reasons, test a smaller version, or seek better evidence before committing. What it shouldn't do is use an industry average to disguise an untested financial assumption. Put the full cost into the next budget and check that the business can fund it when payment is due.

Check what happens to costs at the next edition's expected attendance. An additional guest might use catering you've already paid for, or might trigger a new order, another staff shift, or a transport charge. Use the actual arrangement. The attendance comparison doesn't tell you which costs are fixed or what the next usable place costs to deliver.

There can also be a limit on what the event can accommodate. If Northbank plans for a higher turnout, its venue, session spaces, catering, and arrival arrangements must support that turnout. It shouldn't assume that unused places last year will be available under a different ticket mix this year. Confirm the operational conditions before treating a higher attendance expectation as an uncomplicated improvement.

This is where the event budget guide becomes relevant. Carry the supported assumptions into the whole event budget, including costs unrelated to the proposed change. A small initiative recovering its own expense doesn't demonstrate that the event as a whole will be profitable. Likewise, a worthwhile service correction still needs an owner and money available when the work is booked.

Copy a benchmark comparison record your team can use

Keep this record with the evidence used in your review. It can be a section in your existing report rather than another dashboard. The point is that someone else should understand the comparison, its limits, and the action without asking you to reconstruct it from several screenshots.

Decision and measure

Write the decision, the person responsible, and the date the decision is needed. Name the measure in full, followed by its calculation. State the included and excluded groups. For Northbank, the measure is unique recorded arrivals on valid paid tickets divided by valid paid tickets after refunds, assessed separately by booking category and in total.

Reference and coverage

Record the publisher, report title, source link, publication date, and activity period. Note who supplied the data, what unit was counted, the relevant sample size, and any exclusions. Identify whether the figure is an event-level mean, median, pooled rate, or another stated measure. Write "not supplied" where the source doesn't answer.

Your result and the fair comparison

Include the underlying counts, your reporting cutoff, and whether the figures are final. List the characteristics you matched and the differences that remain. If you recalculated either side, keep the method and inputs. If a common calculation isn't possible, say that the figures provide context rather than a direct performance comparison.

Interpretation and action

State what the evidence shows, what it doesn't establish, and what needs checking next. Describe the proposed action, its full additional cost, owner, approval, and review date. Explain what would make you continue, revise, or stop it. Keep service obligations visible even when a financial benefit can't yet be measured.

For Northbank, a completed conclusion would say that paid attendance rose overall, category rates fell, and the booking mix changed materially. The equal-mix calculation supports further investigation but doesn't identify a cause. The team will check communications and attendee accounts before approving the additional $600, with any confirmed service error corrected as soon as it is identified.

Keep the actual result and adjusted comparison on separate lines in that record. Write "454 paid arrivals from 500 valid paid tickets" for the latest event. Describe the 86% equal-mix result as a calculation using a chosen common mix, not observed attendance. This wording matters when the record reaches someone who wasn't in the analysis meeting. They should be able to distinguish what happened at the door from what you calculated to understand the difference.

To make that assignment usable, Northbank can name the operations lead as the owner and require the review before booking the next edition's extra staff. The lead will verify the latest instructions against the venue arrangements, check how group organizers distributed them, and collect voluntary accounts from both booking categories. Those are proposed checks in this fictional example, not explanations already established by the attendance figures.

The approval note should then say which problem was confirmed, how the proposed work addresses it, and whether the $600 includes every additional cost. If the investigation doesn't support that work, the team should revise the proposal. If it uncovers a different failure, such as inaccurate information, correct that failure and record the action. The earlier attendance comparison stays available so the next review can assess what changed without rewriting the starting point.

Attach this comparison to your post-event report, where it can inform the next budget and delivery decisions. Keep the original result visible when later evidence changes your interpretation. Explaining why you revised a conclusion helps the next reviewer understand the event better than quietly replacing the earlier version.

Ask the publisher when the missing detail could change your decision

Sometimes the report has the detail you need in an appendix. Sometimes you need to ask. A useful request is specific: "Does your attendance benchmark divide unique arrivals by valid paid tickets after refunds, and does the reported rate include complimentary admissions?" That is easier to answer than asking whether the data are relevant to your event.

Ask for the sample behind the particular figure, its event formats and markets, collection dates, exclusions, and calculation. If you're comparing costs, request the included cost categories. If you're comparing feedback, ask for the question, answer scale, collection method, and respondent group. Avoid requesting personal records when a definition or aggregate breakdown would resolve the issue.

You can explain the decision without sending confidential commercial information. Tell the publisher you run paid conferences and need a comparable reference for attendance. Ask whether an appropriate subgroup is available and how many events it contains. Don't assume a subgroup is dependable merely because the publisher can filter the data that narrowly.

If the answer never arrives, preserve the uncertainty. You can still use the report to identify a possible issue, discuss it with your team, and collect better internal evidence. Leave it out of a confident ranking or spending justification when the missing definition could reverse the conclusion. There is no requirement to manufacture a comparison just because the report looks authoritative.

Before your next budget review, attach the source's methods and your completed comparison record to the relevant proposal. Ask the person approving it to check the remaining differences, the full additional cost, and what you'll verify after delivery. If a missing definition could change that approval, resolve it before committing the money.