diff --git a/Readme.md b/Readme.md
index 8bc1923..08fc323 100644
--- a/Readme.md
+++ b/Readme.md
@@ -75,7 +75,17 @@ causes network problems because it seems `--network=host` is also not supported
 in dind (docker-in-docker). So we instead copy an explicit setup to the existing
 postgres, which boots up 10x faster than docker anyway.
 
-1. Copy the setup script
+1. Set up the workflow trigger. Include both `pull_request` (for PR analysis) and
+   `push` to your main branch (to establish a baseline for comparison).
+
+```yaml
+on:
+  pull_request:
+  push:
+    branches: [main]
+```
+
+2. Copy the setup script
 
 ```yaml
 jobs:
@@ -111,7 +121,7 @@ you can change `sudo -u postgres createuser -s -d -r -w me` to create a new user
 with a name of your choosing and `sudo -u postgres createdb testing` to create a
 db with a different name.
 
-2. Run your migrations and seed scripts. This is just an example showing that
+3. Run your migrations and seed scripts. This is just an example showing that
    the migrations should target the postgres instance that was set up with the
    previous command
 
@@ -127,9 +137,9 @@ jobs:
           POSTGRES_URL: postgres://me@localhost/testing
 ```
 
-3. Run your test suite against the same database. You can do this with any tool
+4. Run your test suite against the same database. You can do this with any tool
    and use any query builder or ORM you like.
-4. Run the analyzer. `GITHUB_TOKEN` is needed to post a comment to your PR
+5. Run the analyzer. `GITHUB_TOKEN` is needed to post a comment to your PR
    reviewing the indexes found in your database. `SITE_API_ENDPOINT` is the
    Query Doctor API endpoint used to fetch per-repo configuration.
 ```yaml
@@ -154,7 +164,7 @@ jobs:
           POSTGRES_URL: postgres://me@localhost/testing
 ```
 
-5. Add `pull-request: write` permissions to your job to allow
+6. Add `pull-request: write` permissions to your job to allow
 
 ```yaml
 jobs:
diff --git a/src/main.ts b/src/main.ts
index d34543e..7e032c8 100644
--- a/src/main.ts
+++ b/src/main.ts
@@ -80,6 +80,15 @@ async function runInCI(
       comparisonBranch,
       runId ?? undefined,
     );
+    reportContext.comparisonBranch = comparisonBranch;
+    if (!previousRun) {
+      log.info(
+        "main",
+        `No baseline found on branch "${comparisonBranch}". Comparison will be skipped. ` +
+          `To establish a baseline, run the analyzer on pushes to "${comparisonBranch}" ` +
+          `(add "push: branches: [${comparisonBranch}]" to your workflow trigger).`,
+      );
+    }
   }
   if (previousRun) {
     reportContext.comparison = compareRuns(
diff --git a/src/reporters/github/success.md.j2 b/src/reporters/github/success.md.j2
index 9018fb4..b5224f7 100644
--- a/src/reporters/github/success.md.j2
+++ b/src/reporters/github/success.md.j2
@@ -7,7 +7,9 @@
 {{ queryStats.total | default('?') }} queries analyzed
 {%- if newQueryCount > 0 %} | {{ newQueryCount }} new quer{{ "ies" if newQueryCount != 1 else "y" }}{% endif %}
 {% else %}
-{{ queryStats.total | default('?') }} queries analyzed — no previous run to compare against.
+{{ queryStats.total | default('?') }} queries analyzed — no baseline found to compare against.
+
+> **No baseline on `{{ comparisonBranch }}`** — the analyzer cannot detect regressions without a previous run. To establish a baseline, add a `push` trigger for your comparison branch so the analyzer runs on merges to `{{ comparisonBranch }}`. See the [CI integration guide](https://docs.querydoctor.com/guides/ci-integration/#workflow-trigger) for setup instructions.
 {% endif %}
 
 {% if displayImproved.length > 0 %}
diff --git a/src/reporters/reporter.ts b/src/reporters/reporter.ts
index e777739..165296f 100644
--- a/src/reporters/reporter.ts
+++ b/src/reporters/reporter.ts
@@ -81,6 +81,7 @@ export interface ReportContext {
   metadata: ReportMetadata;
   error?: Error;
   comparison?: RunComparison;
+  comparisonBranch?: string;
   runUrl?: string;
   queryBaseUrl?: string;
 }