marketing-mindset 1.12.0 → 1.14.0

package/README.md

@@ -17,7 +17,7 @@ npx marketing-mindset install              # copy the skill into your agent's sk
 
 `npx marketing-mindset jd --level senior` prints the JD with the artifacts due by day 90,
 the sample-size floors, the kill rule and the screening questions. Web version:
-https://axelfreeman.github.io/marketing-mindset/tools/marketing-engineer-jd-generator.html
+https://axelfreeman.github.io/marketing-mindset/tools/marketing-engineer-jd-generator.html?utm_source=npm&utm_medium=readme&utm_campaign=mindset
 
 ## Floors that do not move
 
@@ -30,11 +30,11 @@ https://axelfreeman.github.io/marketing-mindset/tools/marketing-engineer-jd-gene
 
 ## Links
 
-- Skill page: https://axelfreeman.github.io/marketing-mindset/
+- Skill page: https://axelfreeman.github.io/marketing-mindset/?utm_source=npm&utm_medium=readme&utm_campaign=mindset
 - Source: https://github.com/axelfreeman/marketing-mindset
-- Free test-size calculator (browser, no signup): https://axelfreeman.github.io/marketing-mindset/tools/email-test-planner.html
+- Free test-size calculator (browser, no signup): https://axelfreeman.github.io/marketing-mindset/tools/email-test-planner.html?utm_source=npm&utm_medium=readme&utm_campaign=mindset
 - MCP server, same method over stdio: https://www.npmjs.com/package/marketing-mindset-mcp
-- `llms.txt` for agents: https://axelfreeman.com/llms.txt
+- `llms.txt` for agents: https://axelfreeman.com/llms.txt?utm_source=npm&utm_medium=readme&utm_campaign=mindset
 
 ## Need this done for you?
 
@@ -43,8 +43,8 @@ sending setup that passes a deliverability checklist, tests with a written kill
 measurement inside your own analytics property — is sold as fixed-scope packages, with the
 scope and the prices published before you get on a call:
 
-**https://axelfreeman.com/marketing-engineer.html** — Sprint $900 · Engine $1,900/month · Full build $2,900.
-Scope: https://axelfreeman.com/scope.html · Pricing: https://axelfreeman.com/pricing.html
+**https://axelfreeman.com/marketing-engineer.html?utm_source=npm&utm_medium=readme&utm_campaign=mindset** — Sprint $900 · Engine $1,900/month · Full build $2,900.
+Scope: https://axelfreeman.com/scope.html?utm_source=npm&utm_medium=readme&utm_campaign=mindset · Pricing: https://axelfreeman.com/pricing.html?utm_source=npm&utm_medium=readme&utm_campaign=mindset
 
 MIT © Axel Freeman
 
@@ -55,10 +55,30 @@ npx marketing-mindset aeo          # 23 weighted checks, grouped, with the max s
 npx marketing-mindset aeo --json   # same data for a script or an agent
 ```
 
+### Tag every outbound link before it ships
+
+A channel you cannot attribute is a channel you cannot kill, so the rule is mechanical:
+links that leave for your own property carry UTM, internal links never do.
+
+```
+npx marketing-mindset utm https://axelfreeman.com/offer.html --source devto --medium article --campaign mindset
+npx marketing-mindset utm --file article.md --source devto --medium article --campaign mindset
+```
+
+`--file` rewrites the file in place and reports how many links it tagged; a second run tags nothing
+(idempotent). Third-party links are left alone, existing query params are kept, and sentence
+punctuation after a URL stays where it belongs. The convention it writes:
+
+| parameter | what it holds | examples |
+|---|---|---|
+| `utm_source` | the channel that sent the click | `devto`, `github`, `npm`, `telegram`, `habr`, `reddit` |
+| `utm_medium` | the placement | `article`, `readme`, `package`, `post`, `profile` |
+| `utm_campaign` | the product or the cluster | `mindset`, `tapac`, `hire`, `lyzhi`, `engine` |
+
 The same checklist is a browser tool with a live score:
-<https://axelfreeman.github.io/marketing-mindset/tools/aeo-readiness-checklist.html>.
+<https://axelfreeman.github.io/marketing-mindset/tools/aeo-readiness-checklist.html?utm_source=npm&utm_medium=readme&utm_campaign=mindset>.
 If you want the plumbing in that checklist built for you (offer page, query-intent pages, schema,
-`llms.txt`, IndexNow, the tracked event), packages are published at https://axelfreeman.com/marketing-engineer.html.
+`llms.txt`, IndexNow, the tracked event), packages are published at https://axelfreeman.com/marketing-engineer.html?utm_source=npm&utm_medium=readme&utm_campaign=mindset.
 
 ### Sending ramp: `warmup`
 
@@ -66,7 +86,7 @@ If you want the plumbing in that checklist built for you (offer page, query-inte
 npx -y marketing-mindset@1.4.0 warmup --start 2026-09-21 --target 200 --days 21 --mailboxes 4
 ```
 
-Day-by-day ramp for a new sending domain: starts at 20/day (or the target, if smaller), grows by half a day, never above the target, and prints the four stop rules (bounce >2%/day, complaints >0.1%, any reply ends that sequence, opens halving with flat sends -> drop back two ramp days). `--json` gives the machine-readable schedule. Work with me: https://axelfreeman.com/marketing-engineer.html
+Day-by-day ramp for a new sending domain: starts at 20/day (or the target, if smaller), grows by half a day, never above the target, and prints the four stop rules (bounce >2%/day, complaints >0.1%, any reply ends that sequence, opens halving with flat sends -> drop back two ramp days). `--json` gives the machine-readable schedule. Work with me: https://axelfreeman.com/marketing-engineer.html?utm_source=npm&utm_medium=readme&utm_campaign=mindset
 
 ### Test budget: `budget`
 
@@ -74,14 +94,14 @@ Day-by-day ramp for a new sending domain: starts at 20/day (or the target, if sm
 npx -y marketing-mindset@1.5.0 budget --base 0.03 --lift 0.2 --cpl 4 --daily 500
 ```
 
-The money side of the same floor `sample` computes: contacts needed in both arms, cost per arm and cost to read the test at your cost per contact, days to the floor at a given daily volume, and the spend guard (below the floor the difference is noise, not a result). `--json` for agents. Work with me: https://axelfreeman.com/marketing-engineer.html
+The money side of the same floor `sample` computes: contacts needed in both arms, cost per arm and cost to read the test at your cost per contact, days to the floor at a given daily volume, and the spend guard (below the floor the difference is noise, not a result). `--json` for agents. Work with me: https://axelfreeman.com/marketing-engineer.html?utm_source=npm&utm_medium=readme&utm_campaign=mindset
 ### Dated plan: `plan`
 
 ```
 npx -y marketing-mindset plan --base 0.03 --lift 0.2 --daily 500 --weeks 4
 ```
 
-The floor from `sample`/`budget` laid out on a calendar: contacts per week, cumulative volume against the floor, the first date the test can honestly be read, the mechanics-only first week, and an honest verdict when the volume cannot reach the floor inside the plan. `--json` for agents. Work with me: https://axelfreeman.com/marketing-engineer.html
+The floor from `sample`/`budget` laid out on a calendar: contacts per week, cumulative volume against the floor, the first date the test can honestly be read, the mechanics-only first week, and an honest verdict when the volume cannot reach the floor inside the plan. `--json` for agents. Work with me: https://axelfreeman.com/marketing-engineer.html?utm_source=npm&utm_medium=readme&utm_campaign=mindset
 
 ## `brief` — the plan as markdown you can send
 
@@ -93,7 +113,7 @@ One page: the floor in contacts per arm and in total, what reaching it costs at
 the dated read, a week-by-week table with the percentage of the floor covered, the three gates, and the
 kill rule. `--json` returns the same object for an agent or a script.
 
-Method, packages and the paid version of this work: https://axelfreeman.com/marketing-engineer.html
+Method, packages and the paid version of this work: https://axelfreeman.com/marketing-engineer.html?utm_source=npm&utm_medium=readme&utm_campaign=mindset
 
 ## `mde` — what a volume you can afford can actually read
 
@@ -200,3 +220,28 @@ The floor comes from the same formula as `sample`, and the arms split the earlie
 so a unit chosen here, a queue planned with `queue` and a result read with `verdict` stay consistent.
 Two things it refuses to hide: the earliest step reads soonest and owns the least, and money always reads
 one step behind whatever unit you chose.
+
+## `fit` — which channel can actually fund a verdict
+
+Two channels can both be worth testing and only one of them can be read this quarter. `fit` takes
+`name:contacts_per_month:cost_per_contact` for each channel, applies the same floor as `sample`, and
+says which channel reaches that floor inside the horizon and what one readable test costs there.
+
+```bash
+npx marketing-mindset fit --channels email:9000:0.02,linkedin:1200:0.35,seo:30000:0.005
+```
+```
+email: 9,000 contacts a month at 0.02 per contact -> floor 13,914 per arm, 4,500 events per arm a month, 4 month(s) to a verdict, one test 556.56, fits: yes
+linkedin: 1,200 contacts a month at 0.35 per contact -> floor 13,914 per arm, 600 events per arm a month, 24 month(s) to a verdict, one test 9739.8, fits: no
+seo: 30,000 contacts a month at 0.005 per contact -> floor 13,914 per arm, 15,000 events per arm a month, 1 month(s) to a verdict, one test 139.14, fits: yes
+first to run: seo - 139.14 for one readable test in 1 month(s)
+```
+
+At a 3% base rate and a +20% relative lift the floor is 13,914 contacts per arm on every channel, so the
+channel is not chosen by a better conversion rate but by how fast it can deliver those contacts and what
+each contact costs. `--json` returns the same object: one entry per channel with `floor_per_arm`,
+`events_per_arm_per_month`, `months_to_verdict`, `cost_per_arm`, `cost_per_test` and `fits`, plus
+`first_to_run`, `unreadable_channels` and three gates. `first_to_run` is the fitting channel with the
+lowest `cost_per_test`, ties broken by fewest months; when no channel fits it is a sentence saying so.
+Two things it refuses to hide: the floor is set by the base rate and the lift rather than by the channel,
+and a cheap contact list that cannot reach the floor inside the horizon funds nothing.

package/bin/cli.js

@@ -7,6 +7,7 @@
  *   marketing-mindset verdict        read a finished test: interval, floor, honest verdict
  *   marketing-mindset queue          volume into a queue of honest tests with read-out dates
  *   marketing-mindset cost           the cost of one readable verdict, and funded tests per budget
+ *   marketing-mindset fit            which channel can actually fund a verdict inside a horizon
  *   marketing-mindset unit           which funnel step can fund a verdict, and in how many months
  *   marketing-mindset sections       list every section of SKILL.md
  *   marketing-mindset read <text>    print one section by fuzzy title match
@@ -65,6 +66,107 @@ function readSkill() {
   return fs.readFileSync(SKILL, "utf8");
 }
 
+/* --- UTM tagging: every OUTBOUND link carries source, medium, campaign -------------------
+ * Rule: tag links that leave for our own domains; never touch internal site links,
+ * never tag a link twice, never drop existing query params.
+ */
+const OUR_HOSTS = [
+  "axelfreeman.com", "axelfreeman.ru", "axelfreeman.github.io", "tapacapi.com", "tapac.ai",
+  "gymcamanalytics.com", "releazer.online", "dublicator.online", "spbilo.ru", "avtootkliki.ru",
+];
+
+function utmTag(url, source, medium, campaign, content) {
+  if (!/^https?:\/\//i.test(url)) return url;
+  let host;
+  try {
+    host = new URL(url).hostname.replace(/^www\./, "");
+  } catch (e) {
+    return url;
+  }
+  if (OUR_HOSTS.indexOf(host) === -1) return url; // only our own destinations
+  if (/[?&]utm_source=/i.test(url)) return url; // idempotent
+  const parts = [];
+  if (source) parts.push("utm_source=" + encodeURIComponent(source));
+  if (medium) parts.push("utm_medium=" + encodeURIComponent(medium));
+  if (campaign) parts.push("utm_campaign=" + encodeURIComponent(campaign));
+  if (content) parts.push("utm_content=" + encodeURIComponent(content));
+  if (!parts.length) return url;
+  return url + (url.indexOf("?") === -1 ? "?" : "&") + parts.join("&");
+}
+
+function runUtm(opts) {
+  const source = opts.source ? String(opts.source) : null;
+  const medium = opts.medium ? String(opts.medium) : "referral";
+  const campaign = opts.campaign ? String(opts.campaign) : null;
+  const content = opts.content ? String(opts.content) : null;
+
+  // Tag links in prose, but never inside fenced blocks or `inline code` — those are examples.
+  function tagText(text) {
+    let n = 0;
+    let inFence = false;
+    const lines = text.split("\n").map((line) => {
+      if (/^\s*(```|~~~)/.test(line)) {
+        inFence = !inFence;
+        return line;
+      }
+      if (inFence) return line;
+      const parts = line.split("`");
+      return parts
+        .map((seg, i) => {
+          if (i % 2 === 1) return seg; // inline code — leave as is
+          return seg.replace(/https?:\/\/[^\s)>"'*`]+/g, (u) => {
+            const tailMatch = /[.,;:!?]+$/.exec(u); // sentence punctuation is not part of the url
+            const tail = tailMatch ? tailMatch[0] : "";
+            const core = tail ? u.slice(0, -tail.length) : u;
+            const t = utmTag(core, source, medium, campaign, content);
+            if (t !== core) n++;
+            return t + tail;
+          });
+        })
+        .join("`");
+    });
+    return { text: lines.join("\n"), count: n };
+  }
+
+  if (opts.file) {
+    const p = String(opts.file);
+    const src = fs.readFileSync(p, "utf8");
+    const res = tagText(src);
+    if (res.count) fs.writeFileSync(p, res.text);
+    console.log("tagged links: " + res.count + (res.count ? " in " + p : " (nothing to tag)"));
+    console.log("utm_source=" + (source || "-") + " utm_medium=" + medium + " utm_campaign=" + (campaign || "-"));
+    return 0;
+  }
+
+  const url = opts.url ? String(opts.url) : (opts._ ? String(opts._) : null);
+  if (!url) {
+    console.log(`utm — tag every outbound link, once, before it ships
+
+  Usage:
+    npx marketing-mindset utm <url> --source devto --medium article --campaign mindset
+    npx marketing-mindset utm --file article.md --source devto --medium article --campaign mindset
+
+  Convention:
+    utm_source    the channel that sent the click   devto | github | npm | telegram | habr | reddit | colony
+    utm_medium    the placement                      article | readme | package | post | comment | profile | bio
+    utm_campaign  the product or cluster             mindset | tapac | hire | lyzhi | engine
+
+  Rules:
+    - internal links of your own site carry NO utm — they would restart the session
+    - tag only links that leave for our own domains; a third-party link stays clean
+    - never tag twice: a url that already has utm_source is returned untouched
+    - existing query params are kept`);
+    return 1;
+  }
+  const tagged = utmTag(url, source, medium, campaign, content);
+  console.log(tagged);
+  if (tagged === url) {
+    console.error("(not tagged: internal/no-own-domain/already tagged — see rules)");
+    return 1;
+  }
+  return 0;
+}
+
 
 /* --- agent-readiness / AEO checklist (23 weighted checks) -------------------------------
  * The same checklist that runs in the browser tool, as data: agents can read it and score
@@ -730,6 +832,125 @@ function runUnit(opts) {
   return 0;
 }
 
+var FIT_ZA = { "0.05": 1.9599639845, "0.01": 2.5758293036, "0.1": 1.6448536270 };
+var FIT_ZB = { "0.8": 0.8416212336, "0.9": 1.2815515655 };
+function fitRound2(x) { return Math.round(x * 100) / 100; }
+function fitFloor(rate, lift, power, alpha) {
+  const p1 = Math.min(Math.max(rate, 0.000001), 0.999999);
+  const p2 = Math.min(p1 * (1 + lift), 0.999999);
+  const pbar = (p1 + p2) / 2;
+  const za = FIT_ZA[String(alpha)] || FIT_ZA["0.05"];
+  const zb = FIT_ZB[String(power)] || FIT_ZB["0.8"];
+  const num = Math.pow(za * Math.sqrt(2 * pbar * (1 - pbar)) + zb * Math.sqrt(p1 * (1 - p1) + p2 * (1 - p2)), 2);
+  return Math.ceil(num / Math.pow(p2 - p1, 2));
+}
+function planFit(input) {
+  const channels = input.channels || [];
+  const base = Number(input.base !== undefined ? input.base : 0.03);
+  const lift = Number(input.lift !== undefined ? input.lift : 0.2);
+  const power = Number(input.power !== undefined ? input.power : 0.8);
+  const alpha = Number(input.alpha !== undefined ? input.alpha : 0.05);
+  const horizon = Number(input.horizon !== undefined ? input.horizon : 6);
+  const problems = [];
+  if (channels.length === 0) { problems.push("channels: pass --channels email:9000:0.02,linkedin:1200:0.35,seo:30000:0.005"); }
+  channels.forEach(function (c) {
+    if (!(Number(c.contacts_per_month) > 0)) { problems.push("contacts per month for '" + c.name + "' must be a positive number of contacts a month"); }
+    if (!(Number(c.cost_per_contact) >= 0)) { problems.push("cost per contact for '" + c.name + "' cannot be negative"); }
+  });
+  if (!(base > 0) || !(base < 1)) { problems.push("base rate: a number between 0 and 1 (0.03 = 3%)"); }
+  if (!(lift > 0) || !(lift < 5)) { problems.push("lift worth detecting: a fraction, 0.2 = +20% relative"); }
+  if (!FIT_ZB[String(power)]) { problems.push("power must be 0.8 or 0.9"); }
+  if (!FIT_ZA[String(alpha)]) { problems.push("alpha must be 0.05, 0.01 or 0.1"); }
+  if (!(horizon > 0)) { problems.push("horizon: months you are willing to wait, a positive number"); }
+  if (problems.length) { return { problems: problems }; }
+
+  const floor = fitFloor(base, lift, power, alpha);
+  const rows = channels.map(function (c) {
+    const perMonth = Number(c.contacts_per_month);
+    const cost = Number(c.cost_per_contact);
+    const events = perMonth / 2;
+    const months = Math.ceil(floor / events);
+    const costArm = fitRound2(floor * cost);
+    return {
+      channel: c.name, contacts_per_month: perMonth, cost_per_contact: cost,
+      floor_per_arm: floor, events_per_arm_per_month: events, months_to_verdict: months,
+      cost_per_arm: costArm, cost_per_test: fitRound2(costArm * 2), fits: months <= horizon
+    };
+  });
+  let best = null;
+  rows.forEach(function (c) {
+    if (!c.fits) { return; }
+    if (!best || c.cost_per_test < best.cost_per_test ||
+        (c.cost_per_test === best.cost_per_test && c.months_to_verdict < best.months_to_verdict)) { best = c; }
+  });
+  const none = "none of the listed channels can finish a verdict inside " + horizon + " month(s)";
+  const unreadable = rows.filter(function (c) { return !c.fits; }).map(function (c) { return c.channel; });
+  const gates = [
+    "the floor does not depend on the channel: " + fmtNum(floor) + " contacts per arm in every row, because it comes from the base rate and the lift",
+    best
+      ? "fund " + best.channel + " first: " + best.cost_per_test + " for one readable test (" + best.months_to_verdict + " month(s) at " + best.cost_per_contact + " per contact)"
+      : none + " - move the horizon or find a channel with more contacts",
+    unreadable.length
+      ? "out of reach inside " + horizon + " month(s): " + unreadable.join(", ")
+      : "every channel listed reaches a verdict inside " + horizon + " month(s)"
+  ];
+  return {
+    base_rate: base, lift: lift, power: power, alpha: alpha, horizon_months: horizon, floor_per_arm: floor,
+    channels: rows,
+    first_to_run: best ? best.channel : none,
+    first_to_run_cost_per_test: best ? best.cost_per_test : 0,
+    first_to_run_months: best ? best.months_to_verdict : 0,
+    unreadable_channels: unreadable, gates: gates
+  };
+}
+
+function runFit(opts) {
+  const spec = opts.channels;
+  const base = Number(opts.base !== undefined ? opts.base : 0.03);
+  const lift = Number(opts.lift !== undefined ? opts.lift : 0.2);
+  const power = Number(opts.power !== undefined ? opts.power : 0.8);
+  const alpha = Number(opts.alpha !== undefined ? opts.alpha : 0.05);
+  const horizon = Number(opts.horizon !== undefined ? opts.horizon : 6);
+  const problems = [];
+  let channels = [];
+  if (typeof spec !== "string" || !spec.trim()) {
+    problems.push("channels: pass --channels email:9000:0.02,linkedin:1200:0.35,seo:30000:0.005");
+  } else {
+    spec.split(",").forEach(function (part) {
+      const bits = part.trim().split(":");
+      if (bits.length !== 3 || !bits[0].trim() || !bits[1].trim() || !bits[2].trim() ||
+          !isFinite(Number(bits[1])) || !isFinite(Number(bits[2]))) {
+        problems.push("channels: '" + part.trim() + "' must look like name:contacts_per_month:cost_per_contact");
+      } else {
+        channels.push({ name: bits[0].trim(), contacts_per_month: Number(bits[1]), cost_per_contact: Number(bits[2]) });
+      }
+    });
+  }
+  if (problems.length) { console.error("fit: " + problems.join("; ")); return 1; }
+  const out = planFit({ channels: channels, base: base, lift: lift, power: power, alpha: alpha, horizon: horizon });
+  if (out.problems) { console.error("fit: " + out.problems.join("; ")); return 1; }
+  if (opts.json) { console.log(JSON.stringify(out, null, 2)); } else {
+    const head = ["channel", "contacts/mo", "cost/contact", "floor/arm", "events/arm/mo", "months", "cost/test", "fits"];
+    const rows = out.channels.map(function (c) {
+      return [c.channel, fmtNum(c.contacts_per_month), String(c.cost_per_contact), fmtNum(c.floor_per_arm),
+              fmtNum(c.events_per_arm_per_month), String(c.months_to_verdict), String(c.cost_per_test),
+              c.fits ? "yes" : "no"];
+    });
+    const widths = head.map(function (h, i) {
+      return Math.max(h.length, Math.max.apply(null, rows.map(function (r) { return r[i].length; })));
+    });
+    function pad(s, i) { return s + " ".repeat(widths[i] - s.length); }
+    console.log(head.map(pad).join("  "));
+    console.log(widths.map(function (w) { return "-".repeat(w); }).join("  "));
+    rows.forEach(function (r) { console.log(r.map(pad).join("  ")); });
+    console.log(out.first_to_run_cost_per_test
+      ? "first to run: " + out.first_to_run + " - " + out.first_to_run_cost_per_test + " for one readable test in " + out.first_to_run_months + " month(s)"
+      : "first to run: " + out.first_to_run);
+    out.gates.forEach(function (g) { console.log("gate: " + g); });
+  }
+  return 0;
+}
+
 function fmtNum(n) {
   return String(n).replace(/\B(?=(\d{3})+(?!\d))/g, ",");
 }
@@ -866,6 +1087,8 @@ if (cmd === "sample") {
   process.exit(runQueue(flags(rest)));
 } else if (cmd === "unit") {
   process.exit(runUnit(flags(rest)));
+} else if (cmd === "fit") {
+  process.exit(runFit(flags(rest)));
 } else if (cmd === "cost") {
   process.exit(runCost(flags(rest)));
 } else if (cmd === "aeo") {
@@ -893,6 +1116,8 @@ if (cmd === "sample") {
     process.exit(1);
   }
   console.log("## " + hit.title + "\n" + hit.body.join("\n").trim());
+} else if (cmd === "utm") {
+  process.exit(runUtm(flags(rest)));
 } else if (cmd === "install") {
   require(path.join(__dirname, "install.js"));
 } else {
@@ -913,6 +1138,8 @@ Usage:
   npx marketing-mindset verdict         was the difference real, and was the volume ever enough?
   npx marketing-mindset queue           how many honest tests your volume can finish, and when each reads
   npx marketing-mindset cost            what one readable verdict costs to finish, and how many a budget funds
+  npx marketing-mindset fit             which channel can actually fund a verdict, and what one test costs there
+      --channels email:9000:0.02,linkedin:1200:0.35,seo:30000:0.005 [--base 0.03] [--lift 0.2] [--power 0.8] [--alpha 0.05] [--horizon 6] [--json]
   npx marketing-mindset unit            which step of your funnel can fund a verdict, and how many months it needs
       --steps visits:12000,signups:900,activations:270,paid:40 [--lift 0.2] [--power 0.8] [--alpha 0.05] [--horizon 6]
       --cost 0.02 --daily 500 [--budget 5000] [--base 0.03] [--lift 0.2] [--power 0.8] [--alpha 0.05]
@@ -926,6 +1153,8 @@ Usage:
       --level senior --mode fulltime --surfaces seo,out,analytics
   npx marketing-mindset aeo             agent-readiness / AEO checklist (23 weighted checks)
       --json                            machine-readable
+  npx marketing-mindset utm             tag every outbound link, once, before it ships
+      --source devto --medium article --campaign mindset [--url <url> | --file article.md]
   npx marketing-mindset sections        every section of the skill
   npx marketing-mindset read numbers    print one section
   npx marketing-mindset install         copy the skill into your agent's skills dir

package/package.json

@@ -1,13 +1,16 @@
 {
   "name": "marketing-mindset",
-  "version": "1.12.0",
-  "description": "A marketer's operating mindset as an installable agent skill and CLI: test-size floors, kill rules, honest verdicts on finished tests, which funnel step can fund a verdict, the lift a volume can read, a queue of readable tests with read-out dates, the cost of one readable verdict. npx marketing-mindset sample|kill|verdict|unit|queue|cost|mde|budget|brief|plan|warmup|aeo|jd|sections|read|install.",
+  "version": "1.14.0",
+  "description": "A marketer's operating mindset as an installable agent skill and CLI: test-size floors, kill rules, honest verdicts on finished tests, which funnel step can fund a verdict, which channel can actually fund a verdict, the lift a volume can read, a queue of readable tests with read-out dates, the cost of one readable verdict, and UTM tagging on every outbound link. npx marketing-mindset sample|kill|verdict|unit|fit|queue|cost|mde|budget|brief|plan|warmup|aeo|utm|jd|sections|read|install.",
   "keywords": [
     "agent-skill",
     "skill",
     "marketing",
     "growth",
     "aeo",
+    "utm",
+    "attribution",
+    "link-tagging",
     "claude-code",
     "codex",
     "llm",
@@ -15,6 +18,7 @@
     "sample-size",
     "test-queue",
     "cost-per-verdict",
+    "channel-fit",
     "funnel-unit",
     "ab-test-verdict",
     "minimum-detectable-effect",

package/skill/SKILL.md

@@ -236,6 +236,25 @@ Sharp, honest, zero fluff. No corporate-speak, no generic templates, no hedging.
 - Using paid data sources when free ones exist
 - Working for free — zero-exchange tasks, "for the portfolio", or treating training time as free
 
+## Tagging: Every Outbound Link Carries UTM
+
+A campaign you cannot attribute is a campaign you cannot kill. The rule is mechanical:
+**every link that leaves for your own property carries UTM; internal links never do.**
+
+- `utm_source` — the channel that sent the click: `devto`, `github`, `npm`, `telegram`, `habr`, `reddit`, `colony`.
+- `utm_medium` — the placement: `article`, `readme`, `package`, `post`, `comment`, `profile`, `bio`.
+- `utm_campaign` — the product or cluster: `mindset`, `tapac`, `hire`, `lyzhi`, `engine`.
+
+Rules that keep the data clean:
+
+- Tag only destinations you own. A third-party link stays clean.
+- **Never tag internal site links.** An internal link with UTM restarts the session, so one visit looks like two and the page's own report becomes wrong.
+- Never tag twice: a URL that already has `utm_source` is done. Tagging happens before the text ships, not after.
+- Keep existing query params — append, never rebuild the URL.
+- Run it: `npx marketing-mindset utm --file article.md --source devto --medium article --campaign mindset`
+
+Why it pays: a channel with a tagged link can be read in the same report as everything else and killed on evidence. An untagged one only feels like it is working.
+
 ## Pitfalls
 
 - The temptation to answer from general marketing knowledge is strong — resist it; check fresh, recent context first.