Skip to content
New storyWe built Skald to count visitors. It caught an AI-key thief.
SkaldEvery visit, truly told.
Live demo Start free trial

Features · pages and sources

Website traffic sources: where visits came from, and what they read

Website traffic sources and pages, counted from the log your server already writes. Sources, campaigns and ad networks, then each page's readers, exits and story since it was published.

Website traffic sources are where a site's visits come from: a search engine, an AI assistant, a social app, another site, a campaign link, or nothing at all (Direct). Skald reads them from the referrer and the tags of each log line, and never guesses one the log did not carry. Part of Skald features.

Start your 30-day free trial Open the live demo

30-day free trial · no card · we set it up within one working day · free to self-host

access.log · sample · where visits came from6 lines
  1. 09:14 "GET /blog/" ref "google.com"search engine
  2. 09:15 "GET /docs/" ref "com.openai.chatgpt"AI assistant · ChatGPT app
  3. 09:16 "GET /pricing/?gbraid=…"paid search · Google Ads
  4. 09:17 "GET /?utm_source=newsletter&utm_medium=email"campaign · email
  5. 09:18 "GET /" ref "com.whatsapp"social · WhatsApp
  6. 09:19 "GET /pricing/" ref "-"Direct

Sample lines. The network comes from the click ID; its value is never kept.

[01] sources

Source types, referring sites and campaigns in one grid

Traffic › Sources puts source types, referring sites and campaigns side by side, each with its visits, its leads and its lead rate. Pick a source type to see how one log line is read.

tap one

"GET /blog/" ref "google.com"

people · source: search engine · google.com

A search engine's address in the referrer. A Google click whose address carries udm=50 reads google.com (AI Mode), still under search.

A new source starts a new visit

Someone who opens the site directly and comes back from ChatGPT minutes later is two visits, the second from ChatGPT. An AI assistant's visit is never hidden inside an earlier one.

Referring sites with their icons

Each referring site shows its own small icon, fetched by your server and kept there.

Leads beside every source

A Leads column and a lead rate on Sources, Pages and Places & devices, with first and last touch side by side.

All Skald featuresAI referral trafficlead attribution

[02] tag problems

UTM campaign tag problems, flagged for you

Badly tagged links split one campaign into several, or hide where visitors came from. Skald finds four kinds, with how many visits had each, up to three example landing addresses and the pages they landed on.

  1. Source, no medium

    The link has utm_source but no utm_medium, so no tool can tell the channel. Add one, such as email or cpc.

  2. A medium no tool knows

    An unusual utm_medium lands in "other". Use a usual name: cpc, email, social, referral.

  3. Ad click, no tags

    A gclid or fbclid and no utm_ tag: the network is known, the campaign is not. Turn on auto-tagging.

  4. Tags on your own links

    A utm_ tag on a link between your own pages starts a new source mid-visit. Never tag internal links.

The same section can list campaigns seen in your leads but never in your server log, which often means a redirect dropped the tags. Above it, UTM tags lists visits by source, medium, campaign, term and content.

[03] pages

Top pages, entries, exits and exit rate

Traffic › Pages is the page leaderboard: the last 30 days with their change, or today. Each page with its readers, entries, exits and exit rate.

Published between lists the pages published in any dates, this month or last, with their first crawl and first visitor. Calls under /api/ never count as pages, and a page shown from the browser's cache still counts, keeping its visit's source.

Traffic › Pages · sample · 30 daysexit rate
  1. /pricing/ · 1,204 readers · +12%exits 31%
  2. /docs/setup/ · 862 readers · +40%exits 22%
  3. /blog/server-logs/ · 640 readers · newexits 58%
  4. /book-a-demo/ · 118 readers · +5%exits 64%

Sample data.

page story · sample · /docs/setup/since published
  1. Wed 2 Sep · published
  2. Wed 2 Sep 11:20 · first Googlebot
  3. Thu 3 Sep 08:02 · first person
  4. Fri 4 Sep 02:41 · first AI crawler
  5. Wed 9 Sep 16:05 · first AI fetch
  6. Mon 21 Sep · first lead

[04] page story

Page story: one page since it was published

Select any page for its story: the first crawl, the first AI fetch, the first person, the first lead, and every day since, with where its readers came from.

Compare this page puts its last week against its first. The publish date comes from your website's page list when it gives one; otherwise it is the first day the log saw the page, and the story says so. A day without figures says so too, never 0.

AI crawler tracking

[05] where and when

Countries, cities, devices and the time of day

Places & devices has countries, cities, devices, browsers and systems, each with its leads. Time shows when readers arrive, day by hour.

  • 🇨🇦 CAN: a flag and a three-letter code
  • full name on hover and in the CSV
  • cities from a free city file
  • a day-by-hour heatmap

A country comes from the log's own field or a GeoIP file your server admin sets. A city is counted for people only, at the first page of the visit. The heatmap uses the report's time zone.

Skald Places and devices with sample data: a Cities table with each city's flag and visitors
Sample data. Places & devices, Cities.

[06] what changed

Insights: what is working, dropped, new or gone

Overview › Insights says what changed in plain words, the five biggest first. One sentence, one number and one link per card, to the tab that shows it.

  1. Biggest first

    Size is how much moved and what it is worth: a lead counts more than a visit, an AI fetch or a crawl less.

  2. One card per subject

    A site that is both a referrer and an AI assistant is one card; the rest is under more about this.

  3. Checked days only

    Yesterday against the 14 days before, or 7 or 30 days against the days before. Today lists only what is new today.

See your own sources, 30 days free Open the live demo

What it does not do

  • Many AI apps strip the referrer, so those clicks arrive as Direct. Likely visits from AI are shown apart, never added to any source.
  • A click from Google AI Overviews looks like any google.com search click. Only AI Mode, with udm=50, is told apart.
  • A click ID names the ad network, never the campaign. Untagged ads show up under tag problems.
  • Counted tabs keep each breakdown on its own: "Canada and Chrome" cannot be crossed there.
  • Countries and cities need a field in the log or a GeoIP file. Without one, they say Not recorded.
content groupingGoogle AI Mode traffic

keep reading

More on website traffic sources

FeaturesVisitors and journeysEach visit in order, first touch, a replayable timeline, flow and retention.See the journeysAI trafficAI referral trafficPeople who clicked an AI answer: a floor, with likely visits kept apart.See AI referralsPricingSkald pricingPer site, a 30-day free trial with no card, or free to self-host.See the prices

questions

Website traffic sources FAQ

Straight answers, including what Skald does not do.

Read the docsAsk us a question
What does Direct mean in Skald?

Direct means the visit came with no referrer and no campaign tag. Many Direct visits are people coming back who first found you somewhere else, so under each one Skald shows where that visitor first came from, when it knows. A value the log did not carry at all reads Not recorded, never unknown.

Does Skald read UTM tags and ad click IDs?

Yes. UTM source, medium, campaign, term and content are read from each landing address, and utm_source decides before utm_medium. Ad networks are told from their click IDs, including Google Ads on iPhone (gbraid, wbraid), TikTok and X. Skald keeps which network it was, never the click ID's value.

Are ChatGPT and Claude phone apps counted as AI traffic?

Yes. The ChatGPT and Claude phone apps count as AI assistants, and the YouTube, WhatsApp and Telegram apps count as social. A visit from Google whose address carries udm=50 reads google.com (AI Mode).

What is a page's story?

Select any page on Traffic › Pages to see it since it was published: its first Googlebot and Bingbot fetch, its first AI crawler and AI fetch, its first person, its first lead, and every day since. Compare this page puts its last week against its first.

What are campaign tag problems?

Badly tagged links that split one campaign or hide a source: a utm_source with no utm_medium, a medium no tool knows, an ad click ID with no tags, and tags on links between your own pages. Each shows how many visits had it and example landing addresses.

Why do my countries say Not recorded?

Skald takes a visit's country from your log, when the log has one (Cloudflare's country, or nginx's cc= field), or from a GeoIP country file your server admin sets; DB-IP's free file works. Without either, the country stays Not recorded. Cities need a city file the same way.

$ tail -f access.log | skald

See what AI does with your site. Free for 30 days.

No card. We set up hosted Skald for you within one working day, or you self-host it free. Every line of your log, sorted and added up.

Start your free trial Explore the live demo Run the free site check
Start free trial Live demo