Reading a Report You Did Not Set Up

Why this matters

An analytics account configured by somebody else, years ago, is a set of assumptions nobody in the business has examined.

Some of those assumptions will be wrong, and the report will present the resulting numbers with exactly the same confidence as correct ones.

Which means the first task with an inherited setup is not reading it. It is establishing whether it is measuring what it appears to measure.

That check takes about an hour and it frequently changes the numbers substantially.

The eight things to check

• Who has access, including people who have left.
• Whether internal traffic is filtered.
• Whether the tag is on every page.
• Whether it is on any page twice.
• Whether goals or events still fire.
• What filters are applied to the view you are reading.
• Whether the time zone is right.
• Whether more than one property exists.

The fourth produces double counting and is more common than expected, usually where a tag was added directly and again through a tag manager.

Access first

Before anything else, because it is a security question rather than an accuracy one.

An inherited account frequently lists a former agency, a designer from two rebuilds ago, and an employee who left in 2018.

Each of those can read everything, and in some cases change the configuration.

Check who has access, remove anybody who should not, and confirm that somebody in the business has full ownership rather than only viewing rights.

That last point catches businesses out: an account owned by a former supplier can be lost entirely if the relationship ends badly.

A worked example

A business reviewing an account set up by an agency in 2016.

Five people had access, three of whom had no current connection to the business.

Internal traffic had never been filtered, and the office network accounted for a noticeable share of sessions.

Two goals were configured and neither had fired since a form plugin change in 2018, so the reports showed zero conversions and everybody had assumed the site produced none.

And the property was set to a time zone three hours out, which shifted every daily figure.

Correcting all of that took an afternoon, and the resulting numbers were lower in traffic, higher in conversions, and for the first time believable.

Goals that stopped working

The most consequential failure and the least visible.

A goal configured against a thank you page stops working when the page address changes, which happens during any redesign or plugin update.

The report then shows zero, which is indistinguishable from nobody enquiring.

Which means a business can conclude its site produces nothing when the measurement broke two years ago.

The test is doing it yourself: submit the form and watch whether the goal registers. Two minutes, and it is the single most valuable check on this list.

Whether the tag is on every page

A gap that produces a specific and confusing symptom.

Where a tag is missing from some pages, those pages record no visits, and any visitor moving between a tracked page and an untracked one appears to arrive twice.

Which inflates session counts and makes some pages look unvisited when they are not.

The usual causes: a page built outside the normal template, a landing page created for a campaign, a checkout on a different system, or a page added after the tag was installed.

Checking is straightforward: view the source of a few pages and look for the tag, particularly on anything that looks different from the rest of the site.

Filters and views

A source of confusion worth understanding before comparing anything.

An account can hold several views of the same data, each with different filters applied.

Which means two people reading the same account can see different numbers and both be looking at what they were given.

Check which view your report uses and what filters it carries, particularly any excluding traffic, a country, or a section of the site.

Filters set years ago for a reason nobody recorded are common, and they silently remove data from every report afterwards.

The counter-case

Where a full audit is disproportionate.

An account nobody uses for any decision, where the honest answer is to check the access, leave the rest, and revisit if a question arises.

A business about to replace its site entirely, where the setup will be redone anyway and auditing the old one is effort spent twice.

Even then, the access check is worth doing immediately, because that is about control rather than measurement.

And the goal test is worth two minutes regardless, since a broken one is actively misleading.

Historical data you can now discard

The uncomfortable conclusion of an audit, and worth facing.

Where you discover that measurement was broken for a period, the data from that period is not merely imperfect. For the affected metric, it is wrong.

Which means comparisons spanning the break are meaningless, and any conclusion drawn from them should be revisited.

The practical response is noting the date the fault began and ending your comparable series there, rather than pretending the numbers connect.

That feels like losing something and it is more honest than a chart with a false continuity in the middle of it.

It also explains a good deal of what previously looked inexplicable, which is usually a relief.

Documenting what you find

The step that prevents doing this again in three years.

Write down what is configured: which property, which tag, what events exist, what filters are applied, who has access and what the time zone is.

One page, kept with your other business records.

That document is what a future supplier needs, what makes a handover possible, and what stops the next person inheriting the same unexamined assumptions.

It also records the date you verified everything, which tells anybody reading a report how much to trust it.

The hour

1. Check and clean up access.
2. Confirm you own the account.
3. Submit your own form and check the goal fires.
4. Check the tag is on every page and only once.
5. Check filters, views and time zone.
6. Write down what you found.

The third is the one that changes the most, because a business that discovers its conversion tracking broke in 2018 has been reading an entirely false picture since then.

None of the six requires technical knowledge beyond knowing where to look, and an afternoon spent on them determines whether everything you read afterwards means anything.

What to do once the numbers are trustworthy is covered in a report your accountant would accept.

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.