Data That Stopped and Data That Started

The shape of what you now have

One series running from whenever you started until the end of June, on one basis.

A second running from whenever the new property was installed, on another.

They overlap for however many months both were running, which for most businesses is somewhere between two and twelve.

And they do not join, which is the fact to design around rather than to work around.

Draw the boundary explicitly

• Old series: start date to 30 June.
• New series: installation date onward.
• Overlap: the months both ran.
• The ratio measured in the overlap.
• A note saying they are not comparable.

Five lines written once at the top of the spreadsheet, before anybody builds a chart from it. The fourth is what makes an approximate comparison possible later and it can only be measured now, from the months you have.

The overlap is the only bridge

Which is why it matters more than anything else in the archive.

For the months both properties ran, you have the same reality measured two ways, which is the only calibration that will ever exist.

Take the totals for those months in each and work out the proportion.

That factor lets somebody in three years say roughly what a 2021 figure would have been on the new basis, which is otherwise unanswerable.

Write it as a range rather than a single number, since it varies month to month.

A business with no overlap at all cannot do this and has a hard boundary instead of a soft one, which is worth knowing rather than discovering.

A worked example

A business had eleven months of overlap and worked out that the new property reported between sixty-five and seventy-five per cent of the old for the same months.

They wrote that range at the top of the sheet with the months it was measured over.

Eighteen months later, comparing against 2022, they applied roughly seventy per cent to the old figures and got a comparison that was clearly approximate and clearly useful.

A second business had installed the new property in June and had one partial month of overlap, which was not enough to establish anything.

Their records have a hard break at the end of June with nothing spanning it.

The difference was entirely when the second property was installed.

Do not chart across the boundary

Which is the practical rule.

A single line running from 2019 to 2024 through a measurement change is a chart of two different things presented as one, and everybody who sees it will read the step as a business event.

Either chart the two series separately, side by side, or chart one line and mark the boundary clearly.

Where a continuous line is genuinely needed, apply the ratio and say on the chart that the earlier portion has been adjusted.

An adjusted series with a stated method is honest, and an unadjusted one presented as continuous is not.

The enquiry series has no boundary

Which is the argument this whole month has been making.

Enquiries counted in your own record run continuously through all of this, because nothing about the tools affected them.

So a business with three years of that has an unbroken series covering exactly the period the analytics cannot.

Put that column beside the two traffic series in the same sheet, and it becomes the spine that the other two attach to.

Anybody wanting to know whether 2021 was better than 2023 has an answer, on a consistent basis, regardless of what happened to any tool.

That is worth more than the entire archive of exported reports.

Expect this to recur

Since it is not a one-off.

The new property will itself be replaced eventually, and whatever replaces it will count differently again.

A business that has now built the habit of drawing boundaries and recording ratios will handle that in an afternoon.

One that treats this as a single disruption to get through will be in the same position in five years with nothing prepared.

The structure you build now is the reusable part rather than the figures.

A boundary drawn properly once teaches you exactly how to draw the next one, which is most of what this month was worth.

Keep the old numbers accessible

Rather than filed away.

An archive nobody can open is not an archive, so keep the exported spreadsheet in the same place as the current one and in the same format.

One workbook with a tab per year and a tab for each series is easier to use than eleven separate downloads in a folder.

And whoever maintains this next should be able to find it without asking, which is the same requirement as everywhere else.

Review it once a year and confirm it still opens, since file formats outlive tools and occasionally not by very much.

A plain spreadsheet format is the safest choice available, for the same reason it was the safest choice a decade ago.

The counter-case

This is more structure than many businesses need.

A business that never compares across years has no use for a bridge, and the boundary is simply where one set of numbers stopped.

The ratio is also approximate enough that somebody may reasonably decide it is not worth recording.

And for a business with a short history, there is very little on the far side of the boundary worth preserving at all.

In that case the honest answer is to start the clean series now and treat the old one as a curiosity.

Draw the boundary, measure the ratio from the overlap, never chart across it unadjusted, and keep the enquiry column running through.

The half hour

1. Write the two date ranges.
2. Identify the overlap months.
3. Work out the ratio as a range.
4. Note it at the top of the sheet.
5. Keep both series in one workbook.
6. Never chart across unadjusted.
7. Keep the enquiry column continuous.

Step three can only be done now, since it requires both properties to have run over the same months and one of them has stopped.

What was worth saving is covered in exporting the history you want to keep.

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.