Open the source and landing page reports for the day of the jump, check whether the visitors behaved like people, and if they did, make the page they landed on useful today, while the attention lasts.

The jump in the graph

The visits line in Analytics runs flat for months, then one day it leaps to several times its usual height.

The first reaction is pleasure. The second is suspicion, especially when nobody did anything to cause it.

Almost always, a spike like that comes from one place: somebody linked to you, mentioned you or sent you visitors in a single burst. Occasionally the somebody is a program.

Either way it is worth an hour on the day it happens. A real spike is short-lived, and most of its value lies in what you do before it fades.

Where spikes come from

The last needs ruling out first. A spike of fake visits looks exactly like a real one on the overview graph, and acting on it wastes an afternoon.

Finding the source

In Analytics, set the date range to the day of the jump and compare it with the same weekday a week earlier, so the comparison is fair.

Then open Acquisition, All Traffic, Source/Medium. The row that has grown is usually obvious: a website you have never heard of under referral, or a sudden pile under direct or social.

Referrals name the site. Social shows the network but rarely the post. A jump in direct visits often means links opened from email or from apps that do not pass on where the visitor came from.

While the spike is still running, the Real-Time report shows the sources of people on the site at that moment, which is often the quickest answer.

Finding the landing page

The source tells you who sent them. The landing page tells you what they came for.

Open the Landing Pages report for the same day and add Source as a secondary dimension. One page will usually account for most of the extra visits, and it is often not the home page.

Read that page the way a stranger arriving from the link would. Check that it answers what the link promised and shows them where to go next.

Real visitors or not

Before celebrating, check that the visitors behaved like people.

Real visitors read. They stay on the landing page for a while, some go on to a second page, and they come from places that make sense for the link. Fake ones arrive and vanish, often all from one city.

A referral spike from a site with an odd name, where every visit lasts no time at all and none goes past the first page, is almost certainly spam. Do not visit the site to find out.

One more check settles most cases: add Hostname as a secondary dimension. Visits recorded against a domain that is not yours never touched your website.

A worked example

A Persian restaurant in a suburban plaza saw visits jump on a Thursday afternoon, with nothing changed on the site and no advertising running.

Source/Medium showed one referral: a local food writer's blog, which had published a piece on where to find good kebabs and rice dishes in the area, linking to the restaurant's menu page.

The visitors were plainly real. They came from the surrounding cities, spent minutes on the menu, and a good number went on to the contact page for the address.

The menu page, though, had no hours, no phone number and no mention of reservations. It had been written for people who already knew the place.

That evening the owner added the hours, the phone number and a line about booking for weekends to the top of the menu page, and put a note in Analytics marking the date and the cause.

Several customers that weekend mentioned the article. The extra visits faded within a week, as they do.

What to do while it lasts

A spike is attention you did not pay for, and it lasts days at most. Use it.

Make the landing page answer the visitor's next question, whether that is price, hours, location or how to book. Add a clear link to the page that turns interest into an enquiry.

If the link came from a discussion where people are asking questions, a short, helpful reply from the business, signed as the business, is usually welcome. A sales pitch is not.

And make sure the phone is answered and the inbox watched. A burst of interest that meets a full voicemail box is simply wasted.

Mark it for next year

Mark the spike with an annotation in Analytics, naming the source. It takes a minute.

Without it, next February's comparison will show a strange fall, and somebody will spend an hour working out why this year looks worse than last.

The same habit is worth keeping for anything that moves the numbers: a radio spot, a newsletter, a change to the site, a day the site was down.

The counter-case

Not every spike is worth chasing. A link from a forum on the other side of the world may bring real, interested visitors who can never become customers of a local business.

Some spikes are predictable, such as the day your own newsletter goes out, and need nothing more than the annotation.

And the effort should match the site. For a site with a handful of visits a day, an hour of investigation is plenty.

For the rest, the routine holds: find the source and the landing page, confirm the visitors are real, and improve that page while they are still arriving.

The same day

  1. Compare the day with the same weekday before.
  2. Open Source/Medium to find the sender.
  3. Open Landing Pages with source as a secondary dimension.
  4. Check the visitors behaved like people.
  5. Read the landing page as a stranger would.
  6. Add what they need to that page today.
  7. Annotate the date and the cause.

Step four is the one that saves wasted effort. A spike of referral spam looks like success on the graph, and only the behaviour behind it shows whether anyone was there at all.

Keeping automated visits out of the figures in the first place is covered in bot visits counted as customers.

Frequently asked questions

What usually causes a sudden spike in visits?

One source sending visitors in a burst: a forum thread, a news or blog mention, a newsletter, a shared social post, an ad campaign, or bots and referral spam.

How do I find where the visits came from?

In Analytics, open Acquisition, All Traffic, Source/Medium for the day of the jump and compare it with the same weekday a week earlier. The grown row is usually obvious.

How do I tell if the visitors were real?

Real visitors stay on the landing page, some visit a second page, and they come from places that make sense. Fake ones arrive and vanish, often all from one city.

Can a spike be referral spam?

Yes. A referral from an odd site where every visit lasts no time and the hostname is not yours is spam, however good it looks on the graph.

What should I do while the spike lasts?

Make the landing page answer the next question, such as hours, price or how to book, link it to your enquiry page, and make sure the phone is answered.

Why annotate a spike in Analytics?

So next year's comparison makes sense. Without a note naming the cause, the same month next year looks like a fall, and someone wastes time explaining it.

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.

Visits jumped and nobody knows why?

Send us the date. We will find the source, check the visitors were real and say what to change.

Start a Conversation