Where Are Your Newsletter Subscribers Actually Coming From?

To learn where newsletter subscribers come from, start with Substack's Growth report and the Subscription source (free) field in the subscriber dashboard. Then keep two distinctions: a reported source is the platform's attribution label, not a complete history of everything that influenced the reader; and the source that produces the most signups may not produce the most active subscribers later.

My view is that writers should stop searching for one perfectly true source. Keep a small source ledger with three layers: the source Substack reported, the content path you can verify, and the reason the reader gives when they choose to tell you. Narrareach can keep that evidence beside the next editorial decision, while Substack remains the authority for native subscriber records.

Use three kinds of source evidence

One subscriber might see your name in a podcast, read two articles from Google, notice a Note, and finally subscribe after a trusted publication recommends you. A dashboard still needs to assign a usable label. That label can be accurate within the platform's rules without describing the whole journey.

Treat the evidence in layers:

The first layer is your operating record. The others add context. Do not replace native attribution with a guess because the guessed story feels more satisfying.

Start with Substack's native records

Substack currently gives writers several views of acquisition.

The Growth report connects visitors, subscriptions, and revenue to source categories and a publishing timeline. It can show whether subscriber movement coincided with an article, campaign, or broader change in traffic. The Subscribers dashboard can display a free subscriber's source, signup information, recent activity, email opens, subscription type, paid upgrade date, and revenue where applicable. You can filter or segment readers and export the available columns.

Individual content reports add a narrower view. Post statistics can include new subscribers and subscriber sources. A Note's View stats can include attributed free and paid subscriptions, followers, clicks, shares, and revenue. These reports help identify a content path, but they should not be added together casually. The same reader journey can touch several pieces before one subscription is recorded.

Begin with a fixed period, such as the last 30 days. Record the publication-level number of new free and paid subscribers. That total is the control. Then inspect source and content breakdowns without assuming every table is a mutually exclusive decomposition of the same number.

Understand what common source labels can support

Substack's current metrics guide distinguishes sources such as Google, Instagram, email, Substack onboarding, Substack trackbacks, other Substack surfaces, Direct, and Direct to App.

Use the labels at the confidence they deserve:

• Named external source: evidence that a visit or subscription was attributed to that domain or platform. It does not prove which specific post, profile, or person created the interest unless the report supplies that detail.
• Named Substack publication or recommendation: evidence of a relationship within the network. It does not prove that the recommending writer personally sent the reader.
• Substack onboarding or other: a platform discovery path, not a private explanation of the recommendation system.
• Direct: typed URLs and bookmarks can appear here, but so can visits without usable referral information. Treat it as partly unknown.
• Direct to App: an outside link opened in the app without an exact external origin. Do not assign it to the campaign you most recently ran.
• Email: a click from an email. That may be your own newsletter, a forwarded message, or another email context depending on the report.

If the labels are unfamiliar, use the fuller plain-English guide to Substack traffic sources. The practical rule is to preserve uncertainty. “Direct: 18 subscriptions” is useful. “Our podcast caused all 18 Direct subscriptions” is a hypothesis until you have supporting evidence.

Traffic volume is not subscriber acquisition

A channel can send many visits and few subscribers. Another can send little traffic and a high number of signups relative to those visits. Ranking channels by traffic alone answers where attention came from, not where the subscriber list grew.

Current Narrareach product data makes this distinction visible. On September 21, 2026, Narrareach held freshly cached all-time Growth reports for 10 connected Substack publications. Those reports contained 92 source rows. Only 43 rows had a non-zero subscriber total, and together those rows contained 3,193 platform-attributed subscriptions.

These are self-selected publications with different ages, audiences, and acquisition histories. The totals are not a benchmark, market share, or channel ranking. A source row can record traffic without a subscription. The decision-relevant point is narrower: source presence and subscriber contribution are different facts.

Narrareach's Substack analytics workspace can preserve those source counts beside articles, Notes, subscriber movement, and the publishing queue. It cannot recover a referrer that was never reported or reconstruct a reader's private journey.

Build a 30-day subscriber-source ledger

You do not need a complicated attribution model. Create one row per source category for a shared 30-day window.

Track:

1. reported new subscriptions;
2. relevant visits where the denominator is available;
3. the number of newly acquired readers who show the selected later activity after the same elapsed period;
4. the hours or money spent on the channel;
5. attribution confidence;
6. the next editorial or distribution decision.

Do not compare groups of unequal age. A subscriber acquired yesterday has had fewer chances to open an issue than someone acquired four weeks ago. Either wait until every group has reached the same subscriber age or describe the activity measure as a current rolling snapshot.

Substack's subscriber dashboard guide says its activity field reflects email opens and web views in the last month. That field is useful for segmentation, but it is not automatically a fixed 30-day follow-up from each signup date.

A worked example: the biggest source may not be the best next investment

Suppose a newsletter records 100 new subscribers during a 30-day acquisition period. After every group has had the same follow-up opportunity, the ledger looks like this:

The arithmetic is straightforward: Notes activity is 18 ÷ 40 = 45%; recommendations are 18 ÷ 28 = 64.3%; Google is 9 ÷ 12 = 75%; and the unresolved group is 8 ÷ 20 = 40%.

Several interpretations can be true at once:

• Notes produced the most subscribers.
• Notes and recommendations produced the same number of later active readers.
• Google produced the smallest named subscriber group but the largest active share.
• Recommendations produced 4.5 later active readers per hour of distribution work; Notes produced 1.5.
• The unresolved group cannot be assigned a cost or strategy honestly.

This is a constructed example, not a Narrareach result or a benchmark. It shows why “Where did most subscribers come from?” is only the first question. The next allocation depends on volume, later behavior, cost, and whether the channel is strategically repeatable.

Add reader-reported context without corrupting attribution

A short welcome question can capture a different kind of evidence: “Where did you first hear about this newsletter?” or “What made you subscribe today?”

Keep the answer separate from Substack's reported source. If Substack says Google and the reader says a podcast, both may be useful. The podcast may have created awareness while Google supplied the final discoverable path.

Use a short list plus an optional text field:

• another newsletter or recommendation;
• Substack Notes or the app;
• search;
• a podcast, event, or community;
• social media;
• a friend or forwarded email;
• other.

Do not force every new subscriber to answer. A voluntary response rate leaves gaps, and memories are imperfect. The purpose is to discover repeated influences that native attribution cannot name, not to overwrite the platform record.

Substack also supports private subscriber tags. Tags can preserve information that lives outside the platform's source field, such as attendees from a workshop or members of a course. Use a clear naming rule—event-toronto-sep-2026, for example—and never apply a tag that implies causation you did not verify.

Diagnose each source before scaling it

For every meaningful source, ask what happened at four stages.

1. Discovery

Did enough relevant people encounter the publication? Low traffic may mean the channel is too small, the distribution effort is weak, or the content does not travel there.

2. Subscription

Did visitors understand the continuing promise? High traffic with few subscriptions can point to weak audience fit, a one-off article, or unclear publication positioning. It does not automatically mean the call to action is the problem.

3. First issues

Did new subscribers receive what the acquisition path implied? Someone who joined after a narrow practical guide may disengage if the next three issues cover an unrelated personal theme. The solution may be a better welcome path or more accurate promise, not more promotion.

4. Retention or upgrade

Did the group remain active or become paid where that is part of the model? A source that produces cheap free signups can still be costly if few readers continue. A smaller source may be valuable if the relationship persists.

These stages turn a source label into an editorial diagnosis. They also stop a channel from being judged on a metric it cannot control.

Know when outside analytics helps

Google Analytics 4 can add detail about web sessions, landing pages, and acquisition on pages where its tag loads. Advertising pixels can report campaign events according to their own systems. These tools are useful when you need more web-path detail.

They are not replacements for Substack's native subscriber and billing records. A subscription can happen in the app or another native surface that an external web tag never observes. GA4's session source and Substack's subscription source can legitimately disagree because they describe different scopes.

Use outside analytics to answer a named web question: Which landing page did search visitors enter? Which campaign produced a tagged web conversion? Do not force its total to equal the platform's full native audience.

Turn the source audit into one publishing decision

Choose one source whose evidence is strong enough to act on. Write a one-sentence diagnosis:

“Recommendations produce fewer signups than Notes, but the resulting readers remain more active after the same follow-up window.”

Then choose one response. Study the publications producing aligned recommendations and draft an article that serves the shared reader problem. If Google brings a small but durable cohort, deepen the search-led topic and strengthen internal links. If Notes produces volume but weak continuation, align the Notes more closely with the newsletter promise before increasing frequency.

In Narrareach, keep the source evidence beside that diagnosis, draft the response, review the reader promise and links, then schedule or publish through the appropriate workflow. After the next fixed period, compare the same measures again. Narrareach supports the measure-to-decision-to-draft cycle; it does not claim that timing proves causation or that a larger source is a better audience.

Open the last 30 days of Growth reporting and write down every source with a non-zero subscription count. Mark Direct and Direct to App as unresolved. Choose the largest source you can identify and compare its new readers after an equal follow-up period. Your next article should respond to what those readers came for, not merely to the channel name that delivered them.