Referral traffic: links from other sites, spam and AI assistants

How referral traffic works

When someone clicks a link from one website to another, their browser usually tells the new site where they came from. That piece of information is called the referrer. Your analytics tool reads it on the first page of the visit and asks a simple question: is this a search engine, a social network, one of my own pages, or somewhere else? If it’s somewhere else, the visit goes in the referral pile.

So in one sense it’s a leftover category. It’s every link-driven visit that isn’t search or social. In practice it’s often the most interesting pile in your reports, because it’s made up of people who read about you on a site you don’t control and chose to click.

Say you run a WooCommerce shop selling hand-poured candles. A home-décor blogger writes about autumn table settings and links to your fig-and-cedar candle. Everyone who clicks that link arrives with the blog as their referrer, and your report lists the blog as a referral source.

What counts as a referral

The exact rules vary a little between tools, but this is the usual split:

Where the click happened	Usually reported as
A blog post, article or online magazine	Referral
A forum, Q&A site or community board	Referral (some tools class big communities as social)
A directory, review site or “where to buy” page	Referral
A partner or affiliate site	Referral
Google, Bing, DuckDuckGo and other search engines	Organic search
Facebook, Instagram, Pinterest, LinkedIn and similar	Social
An email app, messaging app or PDF	Often direct, because no referrer is sent
An ad or link you tagged with campaign parameters	Whatever the tags say
Another page on your own site	Not a new source at all

People often ask whether referrals count as organic. It’s unpaid, so it’s organic in the everyday sense, but analytics tools keep “organic” for unpaid search results. If someone finds you on Google, that’s organic traffic. If they find you through a link in someone’s article, that’s referral.

Reading your referral traffic sources

A list of referring sites sorted by visits is a start, but it can mislead you. Volume and value are different things.

Here’s a made-up example. The candle shop gets 4,000 visits in October, and 320 of them are referrals from 25 different sites. Sorted by visits, a large craft forum sits at the top with 110 visits and the décor blog is second with 70. Sort the same list by orders, though, and the picture flips: the forum thread produced one order, the blog post produced nine.

That’s the pattern to look for. A forum thread where someone mentioned you in passing brings curious browsers. A blogger who recommends a specific product to readers who trust their taste brings buyers. If you only look at visit counts, you’ll thank the wrong people and chase the wrong links.

A few habits make this report more useful:

• Look at a longer date range. A good article can keep sending visitors for years. Three months of data hides that.
• Check the landing page. If a referral lands on a product that’s out of stock or a page you’ve deleted, you’re wasting a gift.
• Expect domains, not exact pages. Modern browsers usually send only the referring site’s domain to other sites, so you’ll see the blog’s address but not always which post. Search that site for your name to find the link.

If you use Analyzine, referral visits sit next to organic search, direct and social in the same wp-admin report, and each order is tied to the source that earned it. That makes the “volume versus value” comparison a quick look rather than a spreadsheet job.

Referral spam and other junk

Not every referral is a real person. Referral spam is fake traffic designed to plant a domain name in your reports, hoping you’ll get curious and visit. The site at the other end usually sells something dubious or worse.

Signs that a referrer is spam:

• An odd domain name offering “free traffic”, SEO services or share buttons
• A sudden burst of visits from a site you’ve never heard of
• Every visit sees one page and leaves straight away
• When you check (carefully, without clicking through from a work machine), there’s no link to you anywhere on that site

Some spam never touches your site at all: tools that accept data sent directly to their own servers can be fed fake hits. Trackers that record visits on your own server are harder to fool that way, but a bot can still load a real page and send a made-up referrer. Either way, the fix is the same. Exclude the domain from your reports, block persistent offenders with your security plugin or firewall, and don’t waste time “disavowing” them. They aren’t links to you.

How to earn more referrals

Good referrals are earned slowly, one relationship at a time. Some practical places to start:

1. Find who already links to you. Pull your referral report for the past year. Thank the sites that send buyers and fix any links that point to old URLs with a redirect.
2. Get listed where your customers look. A yoga studio belongs in local wellbeing directories, the council’s activities page and the venue listings of any events it runs. A shop that sells wholesale should be on its stockists’ “where to buy” pages.
3. Make something worth linking to. A candle burn-time guide, a free printable class timetable, a sizing chart or a simple calculator gets linked far more often than a product page.
4. Show up in communities as a person. Answer questions on forums in your niche. Link to your site only when it actually answers the question.
5. Work with partners. Guest posts, joint offers with a complementary business and podcast appearances (with a link in the show notes) all bring visitors who arrive pre-warmed.
6. Tag the links you control. When you place a link yourself, add UTM parameters so you know exactly which post or listing it came from, even if the referrer gets trimmed.

Direct or referral, and where AI assistants fit

The difference between the two is simple on paper. A referral arrives with another site’s address attached. Direct traffic arrives with nothing attached at all.

In real life, plenty of clicks that started on another site end up labelled direct, because the referrer was removed along the way. Links in email apps, chat apps, documents and some privacy-focused browsers often lose it. So treat your referral numbers as a floor, not the full count.

AI assistants are the newest wrinkle. When a chatbot or AI search tool shows a link and the person clicks it, the visit may arrive as a referral from the assistant’s domain, or with no referrer at all, depending on the app and browser. If visits from AI tools matter to you, look for those domains in your referral list and accept that some of it will hide under direct.

For how referral fits alongside search, social, email and paid visits, see traffic source.

Referral data and privacy

Knowing which site sent a visitor doesn’t require knowing who the visitor is. The referrer is a page address, not a person, and you can record it without cookies or a profile of anyone. That’s one reason cookieless analytics can still give you a clear referral report.

One thing to watch: full referring URLs sometimes carry more than you’d expect, such as search terms or account details in a query string. If your tool stores referrers, the domain and path are enough. When you review your report each month, spend five minutes on the top ten referrers by orders, not by visits, and send each of those sites a short thank-you.

Frequently asked questions