We lost our own event data. Here is what to keep, and who should own it.

We wanted to publish our own no-show rate.

Annex Creative Solutions has run more than 230 events over ten years — 3,500 people at KLCC, 800 across a hybrid summit, a long tail of corporate events between 100 and 600. Somewhere in that decade is a genuinely useful number: how many people who register for a Malaysian corporate event actually turn up.

We cannot get to it.

The records live in a shared Google Drive we no longer have access to, in spreadsheets owned by clients we finished with years ago, inside third-party platforms we rented for a single event, and on paper forms that were never typed up. Nothing was breached and no client data went anywhere it should not have. Access simply expired, quietly, one folder at a time, over ten years.

That is a normal outcome and almost every agency has some version of it. It is also completely avoidable, and this page is what we would tell you to do instead.

The five ways event data disappears

None of these look like data loss at the time. That is the whole problem — there is no moment where anything goes wrong.

The drive outlives the access. Someone leaves and their account is deprovisioned. A client closes a workspace. An agency is offboarded and every shared folder goes with it. The files still exist somewhere. You just cannot open them, and there is nobody left to ask.

Nobody agreed who owns the sheet. The client assumes the agency keeps it. The agency assumes the client has it. Both are half right for about a year, and then neither is.

The platform was rented for one event. A tool chosen for a single conference, on a plan that lapsed. The export you meant to do afterwards was never urgent enough to do that week, and then the login stopped working.

Paper that was never typed up. The walk-in sheet at the door, the sign-in book, the scribbled substitutions. It was real attendance data. It went into a box.

The spreadsheet that cannot be searched. This one still exists and is still useless. Five hundred rows, three spellings of the same company, names in inconsistent order, and a column someone free-typed into. Ctrl+F only finds what you can spell the same way twice.

What you are actually losing

Not sentiment. Four specific things, all of which have a use next year.

Your own no-show baseline. The most useful planning number there is, and the only one that is genuinely about your audience rather than someone else's. Two events give you a usable figure. We have run more than two hundred and cannot produce one.

The arrival curve. When people actually walked in, as opposed to when you told them to. It settles the annual argument about start times, and it is the input to how many check-in lanes you need.

Last year's list. Who came, who registered and did not, who was a walk-in. That is this year's invite list, and rebuilding it from memory costs more than the whole event software.

The evidence of what you delivered. When a client asks in eighteen months how the last one went, "well" is a worse answer than a number.

Who should own it

This is the part people get wrong by not deciding, rather than by deciding badly.

The client owns the attendee data. Always. They are the client's guests, invited on the client's behalf, and the client carries the relationship with them afterwards. An agency holding the only copy is not custody, it is a hostage situation nobody intended.

The agency should hold a working copy it can retrieve without asking permission. Not to own it — to do its job. You cannot plan next year's door from a folder you need someone else to unlock.

The platform holds it on behalf of both, and is neither. If your event tool is the only place a record exists, you have swapped one single point of failure for another. Whatever you use, the test is the same: can you get a complete export, today, without contacting support?

If any of this is going into a contract, the data questions procurement will ask covers the wording buyers actually use.

The handover, in the week it still works

Every item here takes minutes on the day and is nearly impossible six months later.

1. Export within 48 hours of the doors closing. Not at month end. The window where everyone still has access and still cares is about a week wide.
2. Export to a format that outlives the tool. CSV or a spreadsheet file. Not a share link, not a login, not a dashboard you have to be invited to. A link is not a copy.
3. Put it somewhere the organisation owns, not somewhere a person owns. A folder in the company's own drive, not in the account of whoever ran the event.
4. Two people have access, minimum. One is a single point of failure with a notice period.
5. Write down where it is. One line in the event's own folder saying what was exported, when, and where it went. This is the step everyone skips and the only one that survives staff turnover.
6. Type up the paper the same week. Walk-ins and door substitutions are the part of the attendance record that exists nowhere else. If they stay on paper they are already gone.
7. Record two numbers per event in one running sheet: registered, and attended. Nothing else. That single sheet is the baseline everything else on this site tells you to build, and it survives every tool you will ever change.

What to keep, and what to delete

Keeping everything forever is not the opposite of losing it. It is a different problem, and a worse one to explain to a client's legal team.

• Keep the aggregate permanently. Registered, attended, arrival pattern, ticket types. It is small, it is not personal data once aggregated, and it is the part with lasting value.
• Keep the attendee list for as long as the client says, and no longer. It is their data and their retention decision.
• Delete what you never needed. Every field you collected "in case" is a field somebody now has to protect. The registration form guide argues for collecting less in the first place, which is also the cheapest way to have less to lose.

The distinction matters: the numbers that make you better at this job are not personal data. You can keep those forever without keeping anyone's email address at all.

What we did about it

We built our own system, which is the honest reason AnnexHub exists. Not as a strategy — as a response to having a decade of events we could not reconstruct.

We are starting the baseline again from zero, from the next event, and publishing it when there is enough of it to mean something. That will take a year. It would have taken ten minutes an event for the last ten years.

That is the whole lesson and it is not a sophisticated one.

The short version

Export within 48 hours, to a file rather than a link, into a folder the company owns, with two people who can reach it and one line written down saying where it is. Keep registered and attended for every event in a single running sheet forever, and keep the personal data only as long as your client wants it kept.

Do that and in two years you will have the number we spent ten years not writing down.