Substack Traffic Sources Explained: Direct, Direct to App, Google and the Network

Substack traffic sources describe the path the platform could identify, not every step a reader took before arriving. Google and a named website provide a relatively clear referrer. Substack onboarding, trackbacks, and other Substack categories identify parts of the platform’s network. Direct can include a typed URL, bookmark, or missing referral data. Direct to App means an outside link opened in the Substack app without preserving the exact external source. Keep those ambiguous categories unassigned rather than crediting them to a recent campaign. Narrareach’s Substack analytics workflow helps compare the available sources with publishing activity and subscriber movement while preserving that uncertainty.

The useful question is not simply which row is largest. It is which source is known well enough to support your next decision.

Substack source labels at a glance

Substack uses source labels in more than one reporting context. Individual post statistics show where views of a post came from. Publication-level Growth and traffic reports show visitors, subscriptions, revenue where applicable, and the sources Substack could identify over a chosen period. A subscriber record can also carry a free-subscription source.

These reports answer related questions, but their totals and categories should not be treated as interchangeable.

Substack’s current metrics guide defines the Growth labels, while its post-source guide explains Email, Direct, Other, and named referring sites for individual posts.

What Direct traffic really means

Direct is the most easily overinterpreted row. A reader may type your address, use a bookmark, or open a saved link. Those are direct visits in the everyday sense. Substack also places visits in Direct when the original source is unknown, including some anonymous browsing.

Referral information can disappear when a reader moves between apps, uses privacy tools, copies a link into a message, or passes through a redirect that does not preserve the source. The resulting visit is real; its origin is incomplete.

This is why a large Direct number does not prove that a large loyal audience returns from memory. Some of those readers may be loyal. Others may have arrived through private messages, documents, untagged links, or another path that lost its referrer.

Suppose Direct rises from 120 to 420 visits during the week you promote an article in three places. You cannot divide the additional 300 among LinkedIn, a private group, and a partner email merely because all three ran. Record the timing, but keep the native source as Direct unless another tracked system provides stronger evidence.

The practical response is to improve future measurement, not rewrite the past. Use distinct links where the destination supports them, keep a campaign log, and compare those records with the native source report. Unknown traffic can still guide a test, but it should not appear as confirmed channel performance.

Direct to App identifies a handoff, not the original channel

Direct to App is more specific than Direct in one respect: Substack knows an external link opened in its app. The missing part is where the external link came from.

A reader might tap a newsletter link in a text message, social app, browser, or another email and be handed into Substack. If the exact outside source does not survive that handoff, the visit appears as Direct to App.

Do not rename Direct to App as “Substack app discovery.” The reader began outside the app. Do not assign it to Instagram, LinkedIn, Google, or email without supporting evidence. The defensible statement is: an external link opened in the app, and the original source was unavailable.

You can still use the row. If Direct to App rises during a coordinated promotion, record the association and repeat the campaign with cleaner link tracking. If it remains high across every week, treat it as a persistent measurement limitation when evaluating external channels.

Google is clearer, but it does not answer every search question

A Google source gives you a stronger acquisition clue: the reader came through Google. It does not, by itself, tell you the search query, ranking position, whether the result was organic or another Google surface, or every page the person viewed before subscribing.

Use Google traffic to find the posts responsible for search discovery. Compare each post’s views and attributed free or paid subscriptions where Substack reports them. A post with 3,000 Google visits and 12 subscriptions plays a different role from one with 400 visits and 10 subscriptions.

The first produced more subscribers. The second produced 10 ÷ 400 × 100 = 2.5 observed subscriptions per 100 recorded visits, compared with 12 ÷ 3,000 × 100 = 0.4 for the first. That does not create a universal conversion probability because views may include repeats and attribution can miss earlier touchpoints. It does suggest that the smaller page attracts readers whose search need aligns more closely with the publication.

Connect Google Analytics or another supported outside tool when you need deeper web analysis. Substack’s Analytics settings documentation says GA4 can report page views, traffic sources, and conversions on publication pages. The resulting GA4 dashboard has its own scope and attribution rules, so it should deepen the investigation rather than be forced to reproduce Substack’s totals.

How to read Substack’s network categories

“The Substack network” is an umbrella, not a single reader path. Substack reports several internal surfaces separately because each represents a different discovery context.

Substack onboarding records subscriptions that occur through algorithmic recommendations shown as someone joins the platform. The reader may select several publications during setup. Evaluate these subscribers by later activity rather than assuming they are highly engaged or low quality.

Substack other covers internal website or app paths such as a profile, inbox, or direct message when a more precise category is not displayed. You know the movement happened inside Substack, but you may not know which internal encounter mattered most.

Substack trackbacks represent links from another Substack location, including a post, comment, or homepage sidebar. This can show that another publication sent readers, but the label does not prove an endorsement or reveal every context without inspecting the referring page.

Recommendations and Notes can also appear as distinct growth sources. Keep them separate when the interface does. A recommendation shown after subscribing, a Note found in the feed, and a profile visit are different acquisition experiences even though all occurred inside Substack.

For paid publications, Substack’s Network section can also categorize paid subscribers based on platform surfaces and account conditions, including the app, other network features, saved payment cards, existing accounts, imports, and new accounts. Those categories describe aspects of the paid subscription path. They should not be added mechanically to a traffic-source table that measures visits.

Traffic source is not the same as subscriber source

Traffic asks where recorded visits came from. Subscriber attribution asks which source or content was credited when somebody joined. The largest traffic source may not produce the most subscribers.

Consider a constructed 30-day report:

Direct provides the most visits and the most subscriptions in raw numbers. Trackbacks produce only 80 visits but six attributed subscriptions, or 7.5 per 100 recorded visits. Google sits between them in volume and observed efficiency.

The table supports three different decisions. Direct contributes meaningful scale but remains too ambiguous for a channel-specific claim. Trackbacks deserve investigation because a small identified source produced a relatively strong outcome. Google may justify more work on the search topics that generated both traffic and subscriptions.

Do not rank sources from one table alone. Repeat views affect the denominator, source mixes change, and 80 visits create more statistical volatility than 380. Compare at least several equal periods and inspect the articles, Notes, or recommendations behind the source.

Use a source-confidence ladder

Before acting on a source, classify the evidence:

1. Named and directly reported: Google, a recognizable domain, a specific post, or another platform-reported source.
2. Identified Substack surface: onboarding, Notes, recommendations, trackbacks, or another explicit network category.
3. Partly identified: Direct to App confirms an external-to-app handoff but not the external origin.
4. Unidentified or grouped: Direct may contain missing referral data; Other can combine individually small sources.

Higher confidence permits a more specific decision. If a named publication repeatedly sends subscribing readers, deepen that relationship or write a relevant follow-up. If Direct rises, improve measurement and inspect timing before changing the editorial plan.

Confidence does not equal value. An ambiguous source can bring excellent readers. The ladder tells you how precisely you can explain the path, not whether the readers matter.

Run a four-week source audit

Choose one consistent four-week window and record visits, free subscriptions, paid subscriptions, and revenue where relevant for each source. Keep raw counts beside any calculated rate.

Then follow this sequence:

1. Separate named external sources, identified Substack surfaces, Direct to App, Direct, and Other.
2. Find the top article or publication page associated with each major source.
3. Calculate attributed subscriptions per 100 recorded visits only where the numerator and denominator cover a comparable period and scope.
4. Compare the new subscribers’ later activity rather than assuming the highest-acquisition source produced the strongest readership.
5. Select one source to expand, one to measure more cleanly, and one to leave unresolved.

Narrareach can bring available traffic-source rows, article performance, subscriber movement, Notes, and supported cross-platform publishing activity into the same review. The workflow then moves from source evidence to a specific next draft, its destination version, review, scheduling, and another comparable measurement period.

Substack remains the source of its native traffic and subscriber labels. Narrareach cannot restore referral information that Substack did not receive, identify the origin hidden inside Direct to App, or turn a same-week correlation into a confirmed conversion.

Open your last 30 days of Growth reporting. Add Direct and Direct to App together only to understand how much acquisition remains ambiguous, never to name a channel. Then find the smallest identified source that produced at least one subscription. Inspect the exact article and reader promise behind it. That source may give you a clearer next experiment than the largest row in the chart.