Reading Referral Reports for Launch Traffic
Traffic · 10 min read ·
A referral report shows who sent visitors, not who sent good ones. How to read sources, pair them with behaviour and avoid the traps in the table.
Open the referral report after a launch and the table looks like the answer to everything. Here are the places that sent people. Here is the biggest. Here is the order. You are tempted to thank the top row and move on.
Do not move on too fast. A referral report tells you where visits came from. It does not, by itself, tell you which sources were worth having. This guide explains how the report is built, where it misleads and how to pair sources with behaviour so that you learn something you can act on.
How a referral is recorded
When someone clicks a link on one page and arrives at another, the browser may tell the destination where the click came from. This information is known as the referrer. Analytics tools use it, together with campaign tags you add to links yourself, to group visits into sources.
Two mechanisms are at work.
The referrer. The page the visitor came from. Useful, but sometimes missing or reduced.
Campaign tags. Labels you add to a link's address, such as a source and a medium. Google's analytics help describes how these parameters are read. Because you choose them, they survive where referrers do not.
The best reports use both: tags for the links you control, referrers for the rest.
Where the data goes missing
Every referral report undercounts something. Know the main causes.
Direct traffic is a grab bag. Visits that appear as direct include people who typed the address, used a bookmark and also many whose source was lost: links from some messaging apps, from emails read in certain programs, from documents and from secure pages in some situations. A large direct figure after a launch usually means some of your shared links were untagged.
Referrers can be stripped. Some sites and apps set rules that hide the originating page. Privacy features in browsers may reduce the information passed.
Tagging is only as good as your discipline. A single link shared without tags is invisible. A tag with a typo splits one source into two.
Platforms rewrite links. Some services wrap links in their own, which may replace or hide the original source.
Counting differs between tools. Different tools define a visit differently, so numbers will not match.
None of this makes the report useless. It means you should read it as a good estimate, not an exact count.
Read the table in three passes
Pass one: the shape
Look at the whole table first.
- How many sources are there?
- Is one source dominant, or are visits spread?
- How large is direct or unknown?
- Are there surprising entries: a site you did not know mentioned you?
A launch that depends on one source is fragile. A spread suggests resilience. A huge unknown share suggests untagged sharing.
Pass two: the quality
Add behaviour to each row. For every major source, find:
- Visits.
- Actions taken, such as sign-ups or enquiries.
- Conversion rate: the share of visits that took the action. A conversion rate is simply the proportion of visitors who complete a chosen action.
- Engagement: whether visitors stayed, looked at more than one page or returned.
Now compare. A source with many visits and no actions is different from one with few visits and many actions. The second may be your best audience.
Pass three: the story
Add context from your notes. What did you do that day? Where did you post? Who mentioned you? Connect the numbers to events.
A useful way to rank sources
Rank sources on three measures, not one.
- Reach: visits.
- Value: actions and their eventual worth.
- Fit: engagement and return rate.
A source that is top on reach but bottom on value is a window-shopper crowd. A source that is modest on reach but top on value and fit deserves attention.
Typical patterns and what they suggest
High visits, very low actions. The audience may not match the product, the link's promise may differ from the page, or the page may be slow or confusing for that crowd. Check the landing page as those visitors would see it.
Low visits, high actions. A small, well-matched audience. Find more places like it.
High bounce, short time. Visitors left quickly. Possibly mismatch, possibly a slow page, possibly the answer was on the first screen and they were satisfied. Combine with action data to tell which.
Spiky source. A burst on one day and nothing after is typical of a post or mention. A steady stream suggests a lasting listing or search.
Many sources, each small. Typical of a healthy long tail. Watch for a few that grow.
Sudden new source. Someone linked to you. Find out who and say thanks.
Referral sites versus campaigns
Separate two kinds of source in your head.
Referrals you do not control, such as another site mentioning you. They arrive with whatever referrer the visitor's browser passes. Treat them as discoveries.
Campaigns you control, such as your posts, emails and listings, which you tag. Treat them as experiments.
Reading the two together tells you both what the world is saying about you and how well your own efforts are working.
Use tags that you can read
A tag scheme only helps if you can read it a year later. Keep it simple and written down.
- Source: the place, in lower case, with no spaces. For example the name of the community or newsletter.
- Medium: the type: post, listing, email, directory.
- Campaign: the launch or period.
Check the tags on a sample of your own links before you share them, and test that a click produces the expected row.
Beware the biggest-number trap
The top row is not always the best row. Large numbers tempt us to do more of what already produced them, even when quieter sources performed better. Always check the action column before you decide where to spend your next hour.
Account for time
Some sources act quickly and some slowly. A post may produce most of its visits within a day. A listing or a search result may produce a few each day for months. Compare like with like:
- Look at the first day, the first week and the first month for each.
- Do not judge a slow source on its first day.
- Do not expect a fast source to last.
From report to action
Turn what you read into specific steps.
- Double down on the one or two sources with the best value and fit. Post there again, with a new angle, when the rules allow.
- Fix the landing page for the sources that send many visits but few actions.
- Thank new referrers and, where fitting, return the favour.
- Tag better wherever the report shows an unexplained direct share.
- Drop sources that cost time and bring nothing.
- Ask the people from your best source what they were looking for.
Write the actions next to the table, with a date.
Keep it honest
When you report numbers to others, say what they are and are not.
- Say "visits from tagged links" not "total reach".
- Say how many sources were untraceable.
- Show counts beside percentages.
- Date every figure.
- Do not credit a source for actions you cannot connect to it.
On this site
A listing here links to your website, so you can look for visits that arrive from it in your own referral report, and the launches page shows how entries appear to visitors. The developers page describes the free API and MCP server for reading listing data, the leaderboards page labels each ranking by what it measures and the submit page starts an entry.
An example, read aloud
Imagine a week-after report with six rows. A community post sent a large number of visits and a handful of sign-ups. A directory listing sent a modest number and a good share signed up. A newsletter mention sent a small number, nearly all of whom started a trial. A search result sent a trickle. Direct held a large share. One row, from a site you did not recognise, sent a few visits with high engagement.
Read it aloud. The community post gave reach but poor fit, so check the page for that crowd and consider a different angle next time. The directory gave steady value, so keep its listing fresh. The newsletter gave excellent fit, so write to the editor with thanks and ask what readers asked. The search result is early and slow, so leave it alone. The large direct share suggests some links went out untagged, so tag the next batch. The unknown site deserves a look and a thank-you. Six rows became six decisions, which is the entire point of a report.
A short checklist for each report
Before you close the report, check that you have done five things. You have noted the date range and the tool. You have paired each major source with an action count. You have written down the top three decisions. You have noted any untraced share and a plan to reduce it. You have saved a copy of the table, so next week's report has something to compare against. Five minutes of discipline here makes every later report more useful than the last.
The short version
A referral report is built from referrers and your own tags, and it always undercounts something, especially direct traffic. Read it in three passes: shape, quality and story. Pair visits with actions and engagement, rank sources on reach, value and fit, tag links consistently, compare over matching time windows and turn each reading into a dated action. The biggest row is not always the best row.
Questions and answers
- What is a referral report?
- A table showing which sites, channels or campaigns sent visitors to your pages, usually with counts of visits and actions.
- Why do some visits show as direct?
- Some links lose their source information, for example from certain apps, private messages or secure-to-insecure transitions, so they appear with no referrer.
- How do I tell good sources from bad?
- Pair visit counts with what visitors did: how many took the action, stayed, returned or replied.
- What should I do with a source that sends many visits but no actions?
- Check whether the audience matches your product, whether the message on the link matched the page and whether the page loads well for those visitors.
- How often should I read the report?
- Daily in launch week and weekly afterwards, looking for patterns rather than single days.