A referral visit becomes useful to a publisher when it helps a reader find something worth reading or doing. Counting visits is a starting point; a repeatable review connects the source, the reader's action, and the effort needed to maintain the channel.
Use the workflow below to measure referral traffic from related websites. It works as a small editorial review, with Google Analytics 4 examples for publishers who already use that tool.
Define the decision before opening a report
Write one question, such as: “Are visits attributed to these related publications helping US readers discover our newsletter?”
Record the review period, reporting timezone, source sites, intended audience, and one meaningful action. A 28-day period is a practical example, not a promised time to see results. Compare it with another complete period of the same length and note unusual publishing or promotional activity.
If the question concerns US readers, apply the same United States country condition to both periods. GA4 supports report comparisons based on dimensions such as Country. Keep a copy of the comparison definition so another reviewer can reproduce it.
Start with session sources
In GA4, open the Traffic acquisition report. Select Session source / medium to inspect the reported origin and medium of sessions, then examine the sources relevant to your review.
Use a consistent scope. Google's comparison of acquisition reports distinguishes session-scoped Traffic acquisition dimensions from user-scoped dimensions such as First user source. Mixing them changes the question being answered.
Build a source list from evidence you can check. For each site, record its reported source value and the public link or placement you observed. A referral source identifies a website in the report; it may still be unclear which of that website's links led to a visit.
For Congrid, recommendations can appear on other member websites, as described on the homepage. Review the relevant referring sites rather than assuming every visit will carry congrid.net as its source. Label placement attribution as uncertain when the available evidence only identifies a domain.
Verify the reader action you plan to measure
Choose an action that reflects your publication's goal: a completed newsletter signup, a useful download, or another clearly defined interaction that your site can measure.
Check what the event actually records. Opening a signup form and successfully joining a newsletter represent different steps. Give the review a precise definition, such as “the configured event fires after the subscription success response.”
Google's key-event reporting guide describes checking collection in Realtime or DebugView. Verify your chosen event through your existing measurement workflow before using it to judge a traffic source.
If collection is missing or unreliable, mark the action metric as unavailable. A blank measurement cannot establish that readers took no action.
Keep session rates and event counts distinct
For a session-based action rate, use sessions containing the chosen action as the numerator and sessions in the same source, geography, and date cohort as the denominator. Google's metric definitions define Session key event rate using sessions in which a key event happened divided by total sessions. Specify which key event the review uses.
Check the key-event counting method as well. Repeated triggers can produce multiple event counts within one session, depending on the setting. An event total and a count of sessions containing that event answer different questions.
Here is a hypothetical calculation, included only to illustrate the method:
| Measure for one defined cohort | Example count |
|---|---|
| Sessions attributed to the reviewed source | 24 |
| Sessions containing a verified signup action | 3 |
| Total signup-event triggers | 5 |
The session-based signup rate is 3 divided by 24, or 12.5%. Dividing five event triggers by 24 would calculate a different quantity. These figures are not observed Congrid results, benchmarks, or forecasts. With only three sessions containing the action, a few additional visits could change the rate substantially.
Use a scorecard another editor can reproduce
Keep the review small enough to repeat:
| Field | What to record |
|---|---|
| Cohort | Dates, timezone, geography, source values, and report scope |
| Arrival | Attributed sessions and the relevant destination pages |
| Reader action | Event definition, session-based rate, and underlying counts |
| Confidence | Collection gaps and uncertainty about the originating placement |
| Maintenance | Time spent on integration checks and resolving problems |
| Decision | Continue observing, improve a reader journey, or fix measurement first |
Compare the underlying counts alongside rates. If your event definition, analytics configuration, or geography condition changed, record the break before comparing periods.
Investigate missing or surprising attribution
GA4's (direct) / (none) guidance explains that this label includes traffic without a clear referral source. Keep that uncertainty visible rather than assigning unexplained direct traffic to a recommendation network.
When a result looks surprising, check the reported source, the public referring link, the destination page, and the measurement setup. Record what each check establishes. A live link shows a possible route; a session report shows attributed activity; neither by itself proves how many recommendation impressions were served.
For disappointing reader actions, inspect the destination experience too. A relevant visit may reach an unclear newsletter offer or a broken form. The content similarity review guide offers a separate way to assess topic, audience, and task fit.
Review traffic alongside the work it requires
Include the time needed to maintain participation. For Congrid publishers, the current onboarding guide explains keeping the homepage accessible and its member badge consistent with the registered domain and wallet. The website verification checklist can support that maintenance review.
Maintain separate records for referral reader outcomes, organic search performance, and protocol rewards. A verified publisher status or reward entry does not establish a referral visit, and an increase in total visits does not establish an organic ranking gain.
Finish each review with one evidence-based next action and a review date. Continue collecting when the sample is small, repair measurement when it is incomplete, and improve a reader journey when a specific obstacle is visible. Use the same scorecard next time so the decision rests on comparable observations.