Check whether the visits came from one source, one country, or one page, and whether time on site collapsed. Bots, referral spam, and a scan all produce a rise with no change in enquiries.

The pattern that means it is not real

Traffic up sharply. Nothing published. No advertising running. Enquiries exactly as before.

Those four together are close to diagnostic, because genuine interest in a small business does not arrive without a cause and does not leave the phone silent.

Real growth is gradual, is traceable to something, and produces at least a few more enquiries.

A sudden vertical rise with a flat business underneath it is almost always a measurement artefact, and it is worth ten minutes to identify which one.

The four checks

Any one of those pointing somewhere unusual identifies the cause, and they take about two minutes each.

Referral spam

The oldest of these and still common.

Visits appear from a site you have never heard of, frequently with a name designed to make you curious enough to visit it.

That is the entire purpose: the traffic is generated to appear in your reports so that you look the sender up.

The tell is a referring domain you do not recognise contributing a meaningful share of a spike, with no time on site and no interaction.

Do not visit the referring address. Filter it out and move on, since visiting is the outcome it was created to produce.

Bots and scanners

The larger category, and much of it is not aimed at you at all.

Automated tools crawl the web constantly: search engines legitimately, and alongside them scanners looking for vulnerabilities, scrapers copying content, and tools checking whether sites exist.

Most of that does not appear in analytics, because analytics generally runs when a page is rendered and many bots do not render pages.

What does appear is anything sophisticated enough to run the page properly, plus anything specifically designed to appear.

The signature is many pages hit in a short period, near-zero time on each, and frequently a single unfamiliar device or browser repeated throughout.

Your server logs show the whole picture where analytics shows a fraction, which is worth remembering when the two disagree.

The other explanations

Not every unexplained rise is unwanted traffic.

Somebody sharing your page in a large group or forum produces a genuine spike from one source with poor engagement, which looks similar and is real people.

A supplier or developer working on the site produces concentrated visits from one place.

Tracking installed twice doubles everything and shows a step rather than a spike, with bounce rate collapsing at the same moment.

And a page being linked from a local news article or a community group is exactly the thing you would want, arriving in the same shape as the things you do not.

Which is why the check matters rather than the assumption.

A worked example

A business saw visits roughly triple over four days and briefly believed something had gone very well.

Enquiries were unchanged.

The source breakdown showed almost all of it from a single referring domain nobody recognised, with an average time on site of a second or two.

Geography showed it arriving from a country they had never sold to.

They added a filter to exclude it, did not visit the referring site, and their figures returned to normal.

The useful outcome was a rule they adopted afterwards: any rise without a corresponding change in enquiries gets checked before it gets celebrated, which has since saved them from two similar episodes.

What to do about it

Less than people expect, and mostly at the reporting end rather than the site.

The newer analytics filters a good deal of known automated traffic automatically, which handles the majority without any configuration.

For what remains, exclude the specific referrer or the specific pattern in your filtering settings rather than attempting anything at the server.

Blocking at the server is possible and is usually disproportionate for a small site, since the traffic costs you very little beyond distorted figures.

The exception is genuine load: if a scanner is hitting your site hard enough to slow it down, that is a hosting conversation rather than an analytics one.

Note it so future-you knows

The step that pays off a year later.

Write down the date, what the spike was, and that it was excluded, in whatever note you keep about the site.

Otherwise somebody looking at an annual chart next winter sees an unexplained mountain in March and either investigates it again or tries to work out what you did to cause it.

That is a real cost: businesses regularly spend an afternoon trying to reproduce a spike that was never real.

One line prevents it, and the same note should record any period where filtering changed, since that is a discontinuity somebody will otherwise misread.

The counter-case

Over-filtering does more harm than the noise.

Excluding an entire country can remove genuine customers, particularly for a business with any cross-border trade, and a broad rule entered carelessly is difficult to notice afterwards because it removes data silently.

Chasing every unexplained fluctuation is also a poor use of attention on a small site, where variation is normal and most spikes resolve themselves.

There is also a version where somebody becomes so suspicious of their figures that they dismiss a genuine improvement as bots.

Check the four things, filter the obvious, note what you did, and let ordinary variation alone.

The ten minutes

  1. Check whether enquiries moved at all.
  2. Look at the source breakdown for the period.
  3. Check geography against where you sell.
  4. Look at time on site for those visits.
  5. Do not visit an unfamiliar referring address.
  6. Filter the specific source, not broadly.
  7. Write one line in your notes about it.

Step one is the fastest test available, since real interest in a small business almost never arrives without producing at least a few more enquiries.

Reading sources properly is covered in where your visitors came from.

Frequently asked questions

What does a rise with no change in enquiries mean?

Usually that it is not real. Genuine interest in a small business is gradual, traceable to a cause, and produces at least a few more enquiries.

What should I check?

Whether most of it came from one referrer, whether it came from a country you do not serve, whether time on site collapsed, and whether it hit one page or all of them.

What is referral spam?

Visits generated to appear in your reports so you look up the sender. The tell is an unfamiliar referring domain with no time on site. Do not visit it, filter it.

Do bots show up in analytics?

Some do. Most do not, because analytics runs when a page renders and many bots do not render pages. Your server logs show the whole picture where analytics shows a fraction.

Could a spike be genuine?

Yes. Somebody sharing your page in a large group produces the same shape with real people, as does a supplier working on the site. That is why the check matters.

What should I do about it?

Filter the specific source in your settings rather than blocking at the server, since the traffic costs little beyond distorted figures. Then note the date so nobody misreads the chart later.

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.

Traffic tripled and the phone did not ring?

Check the source and the time on site. Those two answer it in about four minutes.

Start a Conversation