Reading the digest
The digest arrives on a schedule for each project, at the address registered with it. It is the primary surface: everything in the dashboard and the API is also in here, and most weeks reading it is enough.
Windows are daily, weekly and monthly. Each block compares against the previous window of the same length.
Traffic overview
## Traffic Overview
+ 1,204 events
+ 12 bot events
+ Bot ratio: 1.0% (normal)
+ Top devices: desktop (71%), mobile (24%), tablet (5%)
+ Top regions: GB (58%), US (22%), DE (8%)
Events counts every recorded interaction, including automatic page views, scroll milestones and clicks. It moves with traffic and with how heavily your site is instrumented, so compare it against your own history rather than against anyone else’s.
Bot ratio is the proportion classified as automated from the user agent. Between 10 and 40 percent is ordinary for a public site. A sudden move in either direction is worth a look: upward usually means a crawler discovered you, downward sometimes means a change broke your own instrumentation.
Bot events are excluded from the referrer and session analyses below, so those numbers describe human traffic even when this one does not.
Session overview
## Session Overview
+ 89 sessions (previous window: 74)
+ Devices: desktop 63, mobile 21, tablet 5
+ Regions: GB 51, US 20, DE 7
A session is one visit: the events sharing an identifier, in order, with timings. Sessions rather than pageviews are the unit here, because a journey is the thing worth measuring.
Device and region are taken from the session’s own events, so a session counts once regardless of how many pages it covered.
Referrers
## Referrer
direct 41
google.com 23
news.ycombinator.com 14
bayinformationsystems... 8
linkedin.com 3
Where human sessions came from, taken from the first event of each session and normalised to the sending domain.
direct covers typed URLs, bookmarks, links from applications that strip the
referrer, and any browser or extension suppressing it. It is usually the
largest single row and it is not a channel.
This is the block that most often changes what you do next, because it is the only one that says where to spend attention.
Landing pages
## Landing Pages
/ 38
/pricing.html 19
/docs/rules.html 12
/blog/analytics-rules... 9
The first page of each session. A landing page with many sessions and few onward actions is the clearest signal the digest produces.
This block depends on your action names carrying the path. With the default
client configuration they do. A site emitting a bare page_view for every page
will see this block collapse to a single row.
Tag distribution
## Tags
visited 89
bounce 44
engaged 21
pricing 19
converted 3
One row per tag written by your rules in this window. This is the part of the digest you designed, and it is the part worth reading first.
A session carries every tag that matched, so these do not sum to the session
count. bounce and engaged as defined in the rules documentation are
mutually exclusive; pricing and converted overlap with both.
Ratios between tags are the useful reading. Nineteen pricing visits producing three conversions is a number you can act on. Nineteen pricing visits on its own is not.
Transition insights
## Transition Insights
Top landings: page_view.home (38), page_view.pricing (19)
Top exits: page_view.pricing (22), page_view.home (17)
Top paths: page_view.home -> page_view.pricing -> [exit] (14)
page_view.home -> page_view.docs_rules (9)
Computed from the action sequences of every session in the window as a transition matrix: how often each action follows each other action.
Top exits is the drop-off report. An action appearing frequently as an exit and rarely as a step towards anything else is where people stop. When that action is your pricing page or the middle of a signup, you have found the thing to fix.
The full matrix is available through the API and rendered in the dashboard as a graph. The digest shows the summary.
Comparison with the previous window
Each block compares against the previous window of the same length, stored from the last digest. That comparison is the reason the digest is more useful than the same numbers fetched on demand: it remembers, and you do not have to.
What to do with it
Most weeks the answer is nothing, and that is the correct outcome for a report that arrives whether or not anything happened.
Three things are worth acting on when they appear:
- A new referrer with real session counts. Someone linked to you. Find out where and what they said.
- A tag ratio that moved. Pricing visits up and conversions flat means the page changed or the traffic did.
- A new top exit. Something started stopping people. It is usually a deployment.
For anything the digest raises and does not answer, the sessions behind any tag are one command away:
clientlog --origin https://example.com --project myapp/home \
sessions list --limit 20