Established onAny spreadsheet, any versionExcel 365 and Google Sheets: published read-only viewsBoth: values-only paste into a new workbook
Attaching the workbook is the default and it is rarely the right artefact. The question that decides it is whether the recipient will use the numbers or work on the model, and those want different things.
What sending the file actually sends
The model, so every assumption is now visible to somebody who did not make them and may not understand which are load-bearing. Everything described in the previous three entries, including the hidden material and the cached copies. Permission to change it, and the implicit suggestion that changing it is expected. And a fork, immediately, in the sense of the entry on copies.
What a report sends
Figures, as of a stated date, that cannot be edited and are therefore the same figures for everybody who received them.
That last property is the one people underrate. Four recipients of a report are looking at one thing; four recipients of a workbook are looking at four things within a week.
The forms a report can take
A fixed document, which is what most people mean and which prints and attaches and cannot drift. A workbook containing values only, produced by copying and pasting values into a new file, which is useful where the recipient wants to sort or filter. And a published read-only view of the live document, where both parties are on the same service.
The third is the best where it is available, because it never goes stale, and it is available less often than one would like.
Why people send the file anyway
Three reasons, and all are legitimate. The recipient wants to filter or pivot it. The recipient wants to check the working. And it is one action rather than two.
The first two are answered by sending the report and offering the file on request, which almost nobody takes up, or by giving access to the source rather than a copy of it.
The as-of date
Every report carries the date its data represents, on the face of it, because a report without one will be used a month later by somebody who assumes it is current.
This is the same cell recommended in the entry on documentation, and it is the single most useful thing on any circulated figure.
The report that recalculates
A figure pasted into a document or a presentation with a live link updates when the source does, which sounds helpful and means the slide somebody presented in March shows different numbers in June.
Where a record of what was shown matters, paste as a picture or as values. Where currency matters, link. Deciding which, rather than accepting the default, is the whole recommendation.
What to do when somebody asks for the file
Ask what they want to do with it. Half the time the answer is a question you can answer in a sentence, and the other half is a genuine need to work on the model, which is a conversation about access rather than about attachments.
Sending a copy is the response that avoids that conversation and creates the fork.
The organisational version
Where the same report goes out on a schedule, the arrangement that survives is one source, one published output, and a list of who receives it. Every deviation from that is somebody's copy, and the copies are where the disagreements between departments come from.
What we cannot verify
The publishing and linking behaviours are documented by both vendors and were confirmed on current builds. The rest of this entry is a recommendation about practice, resting on the mechanisms described in the preceding entries of this part rather than on any new claim about the software.
The recipient who cannot open it
A workbook using the newer functions described in the second part opens with errors for anybody on an older version, and the recipient's first experience of your work is a page of complaints about names.
A fixed document has no version dependency at all, which is a further argument for it whenever the audience is outside the organisation and their software is unknown.
The one-line covering note
What the figures are, as of when, and who to ask. Three facts in the message body prevent most of the follow-up questions and all of the misdating.
What to do when the report is wrong
Reissue it with a new date and say what changed, rather than sending a corrected file with the same name. Recipients have the old one, and the only thing that tells them to stop using it is a message that says so.
In short
- The question is whether the recipient will use the numbers or work on the model.
- Sending the file sends the assumptions, the hidden material and a fork.
- Four recipients of a report see one thing; four recipients of a file see four.
- A published read-only view never goes stale and is available less often than one would like.
- A linked figure in a slide shows different numbers later than when it was presented.
- Ask what somebody wants the file for; half the time it is a question with a sentence for an answer.