Analytics, Chat and Three Pixels

The arithmetic nobody does

Each tag is approved individually, and individually each one is defensible: a fifth of a second for a measurement somebody wanted.

Nobody adds them up, because they arrive months apart and are requested by different people.

Six of those is more than a second before the page can get on with itself, which nobody would have approved as a single decision.

Which makes this a question of accounting rather than of technology.

What is usually on there

• An analytics tool, sometimes two.
• An advertising pixel per platform.
• A chat widget.
• A review or badge widget.
• A heat mapping or recording tool.
• A consent banner, which loads before the rest.

The fifth is worth checking specifically. Session recording tools are among the heaviest things a small site loads, are usually installed for a one-off investigation, and are almost never removed afterwards.

Measure the total, not the items

The exercise that makes the decision obvious.

Load the page with the tags present and note the figure.

Then load it with third-party scripts blocked, which every browser's developer tools can do, and note that figure.

The difference is what your measurement and marketing tools are costing every visitor.

That single number is what should be weighed, rather than the fifth of a second attributed to each addition.

It takes five minutes and it is frequently between one and three seconds.

A worked example

A business found their pages took about five seconds and their own content accounted for about one and a half of that.

Blocking third-party scripts brought the page to under two seconds.

Listing them produced an analytics tool, a second analytics tool nobody used, two advertising pixels for campaigns that had ended, a chat widget nobody monitored, a recording tool installed for a study in 2021, and a reviews badge.

They kept the analytics tool, one pixel, and the reviews badge, and removed the rest.

The page came to about two and a half seconds and nothing was lost that anybody could name.

The removed tools had been costing every visitor for two years to serve nobody.

Ask what the output is used for

Which is the question that decides most cases.

For each tag, ask who looks at what it produces and when they last did.

Where nobody can answer, the tag is a cost with no return and removing it is straightforward.

Where somebody does look, it is earning its place and should stay.

That is a better test than any judgement about the tool itself, since a good tool nobody reads is worth less than a modest one somebody acts on.

Ask the question by name rather than in general, since a general question produces a general defence of measurement.

Load them later

Where a tag is genuinely wanted.

Nothing on this list needs to run before the page is visible, since none of them affect what the visitor sees.

Setting them to load after the content means the page appears at its own speed and the measurement happens afterwards.

Some tools resist this, claiming they will miss data, and for most small businesses the difference is a rounding error against the visitors lost to a slow page.

Ask whoever maintains the site to defer everything on this list, which is usually a setting in a tag manager or one attribute in the page.

The consent banner is a special case

Since it has to run first by design.

A consent tool blocks other tags until a choice is made, which means it sits at the top of the chain and everything waits on it.

A heavy consent banner therefore costs more than its own weight, because it delays everything behind it as well.

Choose a light one, load it from your own site rather than an external service where possible, and check what it costs specifically.

Where a business has few tags to consent to, the banner may be costing more than the tags it is managing.

The ones that load on every page

A distinction worth drawing, since not all tags are equally expensive.

A tag needed only on the checkout or the thank you page is being loaded on every page on most sites, because it was installed once in the site-wide header.

Moving it so it runs only where it is needed removes it from every other page at no loss to what it measures.

Conversion tracking, booking widgets, and payment scripts are the common cases, and all three are frequently site-wide when they need not be.

Ask whoever maintains the site which tags are site-wide and whether each genuinely needs to be, which is a shorter conversation than removing anything.

Review it annually

Since this accumulates in exactly the way nothing removes.

Once a year, list every third-party tag with who asked for it and who reads the output.

Remove anything failing both, one at a time, checking nothing broke.

Record the page speed before and after, which turns the exercise into evidence and makes it easier to repeat.

The best moment is when a campaign ends or a supplier relationship changes, since that is when tags become orphaned.

They cost more than time

Worth noting, since speed is only one of the reasons to keep the list short.

Every third-party tag is code from somebody else running on your pages, with whatever access that implies, and you are trusting each supplier's security as well as your own.

Each also involves your visitors' data reaching another company, which is something you are responsible for explaining in your privacy notice.

A tag nobody can account for is therefore an obligation nobody can account for either, which is a second reason to remove it.

Check that your privacy notice lists what is actually running, since the two drift apart exactly as this accumulates.

The counter-case

Measurement is worth paying for.

A business with no analytics is making decisions blind, which costs considerably more than a fraction of a second.

Advertising pixels are also required for the advertising to be measurable, and removing one to save time makes the spending unaccountable.

And a chat widget that produces enquiries is earning its weight comfortably.

Measure the total, ask who reads each output, defer everything, and remove what nobody owns.

The five minutes

1. Test the page normally.
2. Test it with third-party scripts blocked.
3. Note the difference.
4. List every tag by name.
5. Ask who reads each output.
6. Remove what nobody owns.
7. Defer everything that remains.

Step three is what changes the conversation, since a total of two seconds is a decision and six separate fifths of a second never were.

Finding what is there in the first place is covered in third-party scripts you forgot about.

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.