The Second Record Nobody Meant To Create

A regular rings to book. Whoever answers cannot find him, so they create him. Now there are two of him.

One has his history, his preferences, the note about which room he likes. The other has tonight's booking.

⚠️ Neither record is wrong. Both were created by somebody doing their job properly. The problem is that they now describe the same person and neither knows about the other.

The short version

• Duplicates form when somebody cannot find what already exists
• The fix is almost never "be more careful" — it is making the search work
• A duplicate splits history, so both copies look less valuable than the truth
• Merging later is expensive and sometimes impossible
• The same shape applies to bookings, suppliers, shifts and invoices

Why "be careful" does not work

What you ask	What happens
"Check before creating"	They did check. They searched and found nothing
"Search first"	They searched for what they were told on the phone
"Use the phone number"	The number they gave is not the number on file

⚠️ Almost every duplicate is created by somebody who searched first and failed. Treating it as a discipline problem fixes nothing and annoys the person who did the right thing.

One: how the search fails

The causes are consistent and mostly fixable.

Spelled differently     - one from a card, one from hearing it
Different phone number  - a work mobile, a partner's phone
Booked under a nickname - or under the name of whoever booked
Search needs an exact match - so a partial name returns nothing

⚠️ The last one is the biggest and the least discussed. A search that only matches from the first character will not find somebody whose surname was typed into the first-name field.

The diagnostic

Take a regular you know is in the system.
Search for the middle of their name, not the start.
Do they come up?

⇒ If not, your staff cannot find people who were entered slightly differently, and duplicates are structurally guaranteed regardless of how careful anybody is.

Two: what a duplicate costs

The immediate cost is small. The compounding cost is not.

History splits      - neither record shows the real frequency
Value looks lower   - a regular appears as two occasional guests
Notes go missing    - the preference is on the copy nobody opened
Marketing misfires  - a "we miss you" to somebody who came last week

⚠️ The most expensive one is the third. A note that says "allergic" or "never seat near the stage" is useless if it is attached to the copy that was not used tonight.

And the one that embarrasses you

A regular of four years receives a message
addressed to a new guest.

⇒ They do not complain. They conclude you do not know who they are, which is the opposite of the entire point of keeping records.

Three: the merge, and why it is hard

Merging two records sounds simple and rarely is.

Which name is right?
Which phone number is current?
What happens to the bookings on each side?
What happens to points, credit or balances?

⚠️ Many systems cannot merge at all. The available action is to delete one, which throws away whatever history was on it.

What to do when merging is not possible

One   Pick the record with more history. That one survives.
Two   Copy the useful notes across by hand.
Three Rename the other: "DUPLICATE - see [name]"
Four  Do not delete it, in case something references it

⇒ Renaming rather than deleting sounds untidy and is far safer. A deleted record can leave bookings pointing at nothing.

Four: prevention, in order of effect

Ranked by what actually reduces the count.

One   Make search match partial and middle strings
Two   Search by phone number before name, every time
Three Show likely matches at the moment of creating
Four  Give one person the job of reviewing new records weekly
Five  Train the habit

⚠️ Training is last on purpose. It is the intervention people reach for first and the one with the weakest effect, because the failure is usually not knowledge.

Number three is the quiet winner

Type a name to create a new guest
⇒ the system shows three similar existing ones before you continue

⇒ If your system does this, duplicates drop sharply. If it does not, it is worth asking whether it can — this is a common feature, often switched off.

Five: duplicate bookings, which are worse

A duplicate guest splits history. A duplicate booking costs you a room.

Taken twice     - two staff, same call, both entered it
Held twice      - a provisional and a confirmed, never reconciled
Moved, not moved - rescheduled by creating a new one

⚠️ The third is the most common and the easiest to miss, because the original is still sitting in the old slot looking like a real booking.

The check that catches it

Before the shift: scan for two bookings with the same name,
the same night, different times.

⇒ Thirty seconds, once a day. Every one you find is a room you can sell.

⇒ On the related problem of a room that looks busy but is not: two cars, three bookings.

Six: how two people create the same thing

Nearly every duplicate involves two people who could not see each other's work.

One on the phone, one at the desk
One on a mobile, one on the system
One acting on a message, one on a call

⚠️ The gap is visibility, not competence. Two people acting on the same request, neither able to see that the other has started.

The fix that is not technical

Whoever takes it, says so out loud.

⇒ In a small room this removes most of them. It scales badly, which is why bigger operations need the record itself to show "being handled by —".

Seven: the same shape elsewhere

Once you see it, it appears everywhere.

Two supplier records      - invoices split across both
Two shifts for one person - one gets paid, one confuses the rota
Two invoices, one job     - the second one chased for months
Two systems, one truth    - the parallel spreadsheet

⚠️ The last is the parent of all the others. A second place where the same information lives guarantees the two will disagree, and nobody will know which is right.

⇒ On the specific version where an old record keeps giving instructions: the note that was true when you wrote it.

Eight: finding the ones you already have

You do not need a project. You need one list.

Sort all guests by name. Read it.
Duplicates sit next to each other.

⚠️ This catches the same-spelling cases in minutes. The different-spelling ones need the phone number: sort by number instead and look for repeats.

What to expect

A venue that has never checked usually finds
between 3% and 10% of records are duplicates.

⇒ That is a measured range from cleanup exercises, not a promise about yours. The point is that the figure is never zero, and the first check is always the most productive.

Numbers worth keeping

One   Total guest records
Two   How many share a phone number with another record
Three How many share a name with another record
Four  New records created last month, and by whom

⚠️ Four is diagnostic, not disciplinary. If one person creates most of the new records, it usually means they are the one who cannot find things — a search problem attached to a particular shift or device.

The card for the office

────────────────────────
  Before creating a new guest

  1. Search the phone number first.
  2. Search part of the name, not all of it.
  3. Try the other name field.
  4. If you create one anyway, say so out loud.

  Two records for one person is two
  half-histories, not one whole.
────────────────────────

Common objections

"Our staff are careful"

⚠️ They are, and duplicates are still forming, because the cause is a failed search rather than a skipped one.

"We'll clean it up when we change systems"

⇒ Migrations copy duplicates across faithfully. Cleaning before a move is work; cleaning after is the same work plus a worse mess.

"It's only a few"

⇒ It is a few of the records and a much larger share of your regulars, because the guests most likely to be duplicated are the ones who come most often.

"We can just merge them later"

⚠️ Check that your system can. Many cannot, and the alternative — deleting one — loses whatever was on it.

"We use phone numbers as the key"

⇒ Good, and it is the strongest single key available. It still fails when somebody books from a different phone, so it reduces duplicates rather than eliminating them.

What to do this week

One   Test the search: look up a regular by the middle of their name.
Two   Sort your guest list by name and read it once.
Three Sort by phone number and look for repeats.
Four  Merge or mark what you find. Keep the longer history.
Five  Ask whether your system can warn before creating a similar record.

Summary

• Duplicates come from failed searches, not carelessness
• The cost is split history, which makes regulars look occasional
• Merging is often impossible — mark, do not delete
• Partial-match search and a warning at creation do most of the work
• Duplicate bookings cost a room, not just tidiness
• Sorting the list by name, once, finds most of them

Nobody creates a duplicate on purpose. They create one because the thing they were looking for did not come up, which is a question about your search, not about them.