What You Cannot Export

The difference between data and access

Exporting a report saves an answer to a question somebody asked.

What is being lost is the underlying record, against which any question could be asked.

So the export is a set of photographs rather than the thing photographed, and the questions you did not think to ask in June are unanswerable afterwards.

That is worth understanding before assuming an archive covers you.

What goes with the access

• Segments you did not build.
• Comparisons you did not run.
• Detail below what you exported.
• Combinations of two dimensions.
• Any question raised later.

The third is the practical one. Exporting your top thirty pages means the thirty-first is gone, and next year's question about a page that was not in the top thirty then has no answer.

Export a little more than seems necessary

Which is the only real defence.

Take the top hundred pages rather than the top thirty, since the extra rows cost nothing and the omission is permanent.

Take monthly rather than annual, for the same reason.

Take the whole date range rather than the last two years.

The marginal cost of a larger export is a slightly bigger file, and the marginal cost of a smaller one is discovering the gap when it matters.

This is the one area where over-exporting is the correct instinct.

The comparability problem is separate

And is not solved by any amount of exporting.

The new version measures differently, so its numbers describe something else even where the labels are similar.

A session, a user, and a pageview are all defined differently enough that placing the two series next to each other produces a step that is an artefact rather than a change.

So your archive of old figures cannot simply be joined to the new ones, and any chart doing that is misleading whoever reads it.

What it can do is answer questions about the past on its own terms, which is a real and narrower use.

Label the archive clearly as old-basis figures so nobody joins them later by accident.

A line at the top of the sheet saying these were measured differently and stop on the thirtieth of June is enough to prevent it.

A worked example

A business exported thoroughly in June and was satisfied.

The following spring they wanted to know how a particular page had performed for mobile visitors specifically over three years.

They had exported page totals and device totals separately, but never the two together, so the question could not be answered.

Nothing had been done wrong, and the combination simply had not occurred to anybody.

They now export the same reports from the new property annually, including two or three combinations they do not currently need.

Which is the correct lesson: the archive is for questions you have not had yet.

Some things were never in there anyway

Worth remembering, since it puts the loss in proportion.

Telephone calls, unless you tracked them, which most small businesses do not.

Referrals and word of mouth, which appear as direct or as branded search.

Anybody blocking the tracking, which is a meaningful and growing share.

And the reason anybody did anything, which no analytics has ever recorded.

So the record being preserved was always partial, and treating it as a complete history of the business overstates what is at stake.

Your own records are the better archive

Which is the argument this leads to.

Invoices, enquiry counts, quotes sent and won, and the note of what happened each year are a more complete history than any analytics export.

They are also in your possession, in a format nobody can discontinue, and they describe the business rather than the website.

A business with three years of enquiry counts and no analytics at all can answer more useful questions than one with the reverse.

If this deadline prompts one habit, that is the one worth forming.

Start the new archive now

Since the same loss will happen again.

Whatever tool you use next will also be replaced eventually, on somebody else's schedule.

Exporting the same six reports once a year, into the same spreadsheet, makes every future transition a non-event.

Put it in the annual routine alongside the other things done once a year.

Twenty minutes annually is the entire cost of never being in this position again.

The same applies to everything else you rent

Worth generalising, since this deadline is one instance of a recurring pattern.

Your email platform, your booking system, your listing, and your social profiles all hold data you did not create a copy of.

Each will eventually change its terms, its export options, or its existence, and each will do it on a schedule you are told about rather than consulted on.

Ask of each, once, what you would lose if it stopped tomorrow and whether an export exists.

That is an hour that turns several future emergencies into scheduled inconveniences, which is the most this kind of preparation can ever achieve.

The counter-case

This can be over-worried.

Most small businesses never ask a historical question more complicated than whether last year was better than this one, which the monthly totals answer completely on their own.

The combinations described above are the kind of question larger organisations with analysts ask, and small ones almost never do.

And an archive nobody opens is a cost rather than an asset, particularly one that takes a day to assemble.

Six reports and an hour is proportionate; three days of thorough extraction for a nine-page site is not.

Export more rows than seem necessary, label the archive as old-basis figures, keep your own enquiry records, and repeat the whole exercise annually.

What to do

1. Export top hundred, not top thirty.
2. Use monthly, not annual.
3. Take the whole date range.
4. Add two combinations you do not need.
5. Label it old-basis.
6. Keep your own enquiry counts.
7. Repeat annually from now on.

Step six is the one that matters longest, since your own records describe the business rather than the website and no company can discontinue them.

Why two tools disagree is covered in comparing two analytics tools that disagree.

Frequently asked questions

West Coast Media Solutions Inc. provides web design, web development, hosting, digital marketing, and business consulting to organisations across Canada, drawing on more than twenty-five years in the field.