What Is Dark Social? Why Shared Links Show Up as Direct Traffic

Dark social is the sharing of links through private channels, such as WhatsApp, Slack, text messages, DMs and email, where the click reaches your site without any referral information. Because nothing says where the visitor came from, analytics tools like GA4 file the visit under Direct. The traffic and the sales are real; only the source is invisible.

This guide explains what dark social is, why the referrer disappears, how much of your Direct traffic it really is, and how to estimate it and put a value on it with a worked example. It also covers the newest dark channel, AI assistants, and the jargon people mix it up with.

What Is Dark Social?

Picture a reader who finds your pricing page, copies the URL from the address bar and pastes it into a team Slack channel. Three colleagues click it. On your side, three new visitors land directly on a deep pricing URL with no source at all. They did not type that URL or bookmark it. A person recommended you, privately, and your reports have no row for that.

That is dark social. Three things define it:

• The sharing is private. It happens one to one or in small groups, not on a public feed where you could see likes, shares or comments.
• The click carries no referrer. The browser does not tell your site which page or app the link was clicked in, and the URL usually has no UTM tags because people share the plain address.
• Analytics misfiles it. The visit lands in Direct, next to typed URLs and bookmarks, so word of mouth looks like brand loyalty or noise.

Dark social is not the dark web and has nothing to do with hidden or illegal content. “Dark” only means the source is dark to your analytics.

Dark social channels

Some web versions of these tools do pass a referrer, depending on how they open links. Check your own referral report before assuming any single app is always dark.

Where the Term Comes From

Alexis Madrigal coined “dark social” in an October 2012 article for The Atlantic, “Dark Social: We Have the Whole History of the Web Wrong”. Working with the analytics firm Chartbeat, he noticed that a big part of the site’s “direct” traffic landed on deep article URLs that nobody would type from memory. His conclusion: people had been sharing links privately through email and chat long before Facebook and Twitter, and analytics had simply never seen it. The Atlantic revisited the idea in 2013.

Some current posts credit the concept to B2B marketers who popularized it around 2021. They made it a common phrase in demand generation, but the term and the measurement idea are from 2012.

Why Dark Social Has No Referrer

Your analytics tool learns a visit’s source from two places: the Referer header the browser sends with the request (the page the link was clicked on), and any UTM parameters in the URL. Dark social visits usually have neither. Here is why:

• Apps are not web pages. When you tap a link in a native app, the app hands the URL to a browser or an in-app browser. There is no previous web page, so there is often no Referer to send.
• Browsers trim referrers on purpose. Under the default policy, a browser sends only the origin (for example, https://example.com/) on cross-site clicks, and sends nothing at all when a link goes from an HTTPS page to an HTTP page. Sites can also choose a stricter policy. MDN documents the rules in its Referrer-Policy reference.
• People share clean URLs. Someone copying from the address bar shares whatever is there. If your own campaign UTMs were in it, they now tag a share with the wrong source, which is its own problem.

With no referrer and no UTMs, GA4 has nothing to work with. Its default channel group puts a session in Direct when the source is exactly “(direct)” and the medium is “(none)” or “(not set).” That is where dark social lands. If you want the full mechanics of referrers, read what is a referrer.

Dark Social vs Direct, Referral and Social Traffic

The short version: dark social is a subset of Direct, and not all of Direct is dark social. For how referral traffic differs, see what is referral traffic.

Why Dark Social Matters

• It is where trust lives. A link from a colleague or a friend carries more weight than an ad. Private recommendations often decide which vendors make a shortlist or which product gets bought.
• It distorts your channel reports. The channel that started the conversation (a social video, a blog post, a podcast) gets no credit, and Direct looks bigger and more valuable than it is. Cut the “non-converting” channel and the Direct sales it was feeding can shrink with it.
• It is large in B2B. Buying groups pass links around Slack, Teams and email threads for weeks. The final visit that converts often has no source.
• It drives a lot of ecommerce growth. Products spread through group chats, especially in markets where messaging apps are the main way people communicate. A store can see its Direct orders rise without any change in its marketing.
• It is growing. Privacy features in browsers and operating systems, and the move of conversations from public feeds to private groups, both push more traffic into the dark.

Those big percentages

Most articles on dark social quote a figure for how much sharing is private. The figures differ from page to page, rarely link to a method, and many trace back to surveys by ad-tech vendors from the mid-2010s. Treat them as a sign that the effect is large, not as a benchmark for your site. Your own Direct traffic, segmented as shown below, is a better number than any industry average.

How to Measure Dark Social

You cannot read private messages, and you should not try. You can measure the footprint dark social leaves. Use these five methods together; each one covers a weakness of the others.

1. Split Direct by landing page. In GA4, open Reports > Engagement > Landing page and add a filter for Session default channel group = Direct. Homepage and short, memorable URLs (login, account, cart) are mostly genuine direct visits. Long, deep URLs such as a specific blog post or product variant are rarely typed, so Direct sessions on them are mostly shared links.
2. Keep new visitors only. Add New / established (or new vs returning users). Returning visitors on deep pages may be using bookmarks or history. New visitors on deep pages with no source are the strongest dark social signal you have.
3. Make sharing traceable. Add share buttons for WhatsApp, email, Slack and copy-link that append UTM tags (for example, utm_source=whatsapp&utm_medium=share). Most people still copy the address bar, so this captures only part of the sharing, but it shows which pages and channels people share to. Referral codes and per-customer links do the same for ecommerce.
4. Ask. Add an optional “How did you hear about us?” field to lead forms and post-purchase pages, with choices like “A friend or colleague sent me a link” and “A group chat or community.” Self-reported answers are imperfect, but they catch what no tag can.
5. Watch the proxies. Track branded search impressions in Search Console, mentions in public communities, and the gap between what you spend and what Direct produces. A rise in all three after a campaign is a dark social signal. For bigger budgets, a social listening tool or a brand recall survey of your audience adds a view that does not depend on clicks at all.

Worked Example: Turning Direct Traffic Into a Dark Social Estimate

The numbers below are illustrative, for a B2B software site in one month. The goal is a defensible estimate and a dollar figure, not precision.

That gives 2,000 sessions: 22% of Direct (2,000 / 9,000) and 5% of all traffic (2,000 / 40,000). Now attach outcomes. Suppose those 2,000 sessions produced 30 demo requests (a 1.5% conversion rate) and each demo request is worth $80 to the business, based on close rate and deal size. Dark social brought in about 30 × $80 = $2,400 of pipeline value that month, none of which any channel report credited.

Two follow-ups make the number useful. First, look at which deep pages lead the list: those are the pages people share, so they deserve share buttons and updates. Second, compare the estimate month over month after a campaign. If a webinar or a social video is followed by a rise in dark social sessions on related pages, that campaign is getting credit it would otherwise never see. To set a value per conversion, use our guide on how to calculate conversion value.

Not All Direct Is Dark Social: Failure Modes to Rule Out First

Before you call a jump in Direct “dark social,” check whether your own setup is losing sources. These causes look identical in reports and are fixable.

AI Assistants: The Newest Dark Channel

AI assistants behave like dark social in two ways. First, some clicks from them arrive with a referrer such as chatgpt.com or perplexity.ai, but others arrive with none, for example from mobile apps or when someone copies a link out of an answer. Those visits land in Direct. Second, an assistant can recommend your brand without any click at all; the buyer then searches your name on Google and the sale is credited to branded search or Direct.

The fix is the same mix as for dark social: report the AI referrals you can identify, keep the rest in Direct instead of guessing, add “an AI assistant such as ChatGPT” to your “how did you hear about us” options, and watch branded search. SEOConversion works this way: it identifies AI visits from referrers and supported UTM signals and leaves sourceless visits in Direct rather than inventing attribution. For the setup, read AI conversion tracking, and for organic search, the pillar guide on SEO conversion tracking.

How to Get More (and Better) Dark Social

• Make pages worth forwarding. Clear comparison tables, calculators, original data and checklists are what people paste into a team chat. One useful page beats ten generic ones.
• Make sharing one tap. Put share buttons where people decide to share (after a key result, under a table), and have them carry UTM tags so you learn from them.
• Give people a reason to pass it on. Referral programs and shareable discount codes reward private sharing and make it measurable at the same time.
• Show up where the conversations happen. Brand-run communities, employee advocacy and creators people trust all seed the private groups you cannot buy your way into. Act like a member, not an advertiser.
• Judge upstream channels fairly. Before cutting social, content or podcasts, check whether your dark social estimate and branded search moved with them. A last click report alone will undercredit them.

Other Terms People Confuse With Dark Social

Searches for dark social often turn up other social media jargon. Most of these terms have no official definition, so the meanings below are the common ones.

FAQ