Autopilot Guide
End-to-end SEO content pipeline for kidkarma.app
Research
→
finds keywords
Compete
→
analyzes competitors
Reddit
→
monitors discussions
Content
→
generates & publishes blogs
Promote
→
images, social, newsletter
Intelligence
↩
tracks quality & rankings
Research Tab
Discovers and scores keywords for content generation. This is the starting point — works from zero, no prior data needed.
Discover
→
Enrich
→
Approve
→
Content
How Discover Keywords Works
Pulls keywords from 3 dynamic sources via DataForSEO API (does not visit any website):
1. Competitor gaps — Asks DataForSEO: "what keywords does busykid.com rank for?" Returns keyword lists from their database. Uses the same competitor list you manage on the Compete tab
2. Reddit trends — Takes high-engagement post titles already fetched on the Reddit tab, extracts topic phrases, and asks DataForSEO for related keyword suggestions
3. People Also Ask — Extracts PAA questions from existing SERP data and expands them into keyword suggestions
The competitor list is shared: Research uses it to find keyword ideas (API query), Compete uses it to scrape actual pages (HTTP). Same list, different purpose.
Cost-safe: Every step checks local DB first. Skips API calls for keywords already discovered, competitors already queried within 7 days, and SERP data already fetched. No duplicate spending.
Buttons
- Discover Keywords (~$0.50-1.50) — Pulls from competitor gaps, Reddit trends, and Google PAA (see above)
- Enrich & Score (~$0.10-0.25) — Fetches difficulty scores, intent, and SERP data for unscored keywords
- Clean & Classify (free) — Removes off-topic keywords, reclassifies funnel stages
- Approve / Skip (free) — Approve sends keyword to content queue. Bulk actions via checkboxes
- Active Funnels (free) — Bulk approve/revoke by funnel stage (TOFU/MOFU/BOFU)
Table Columns
- Volume — Monthly Google searches. Target 100-1000 for low-competition wins
- Diff — Keyword difficulty (0-100). Under 40 is ideal for new sites
- CPC — Cost Per Click ($) — what advertisers pay per Google Ads click. High CPC ($1+) = high commercial intent = valuable traffic. Informational only, does not affect scoring
- Opportunity — Combined score (0-100) weighing volume, low difficulty, and business relevance
- SERP — Competition signal. Green = weak (can rank), Yellow = moderate, Red = dominated by big sites
- Book — Whether knowledge base has matching content for E-E-A-T backing
Statuses
researched Discovered, not scored →
analyzed SERP scored →
approved Ready for content →
content_created Draft exists →
published Live •
skipped Removed
Auto-cleanup: stale keywords (>90d) get skipped; old skipped (>60d) get deleted. Runs Sunday 3 AM.
Compete Tab
Monitors competitors, finds content gaps, and tracks your SERP rankings over time.
Add Competitors
→
Discover
→
Review
→
Content
→
Rank Check
↩
Buttons
- Run Discovery (free) — Scrapes competitor pages, compares structure, generates recommendations
- Check Ranks (~$0.06) — Checks Google position for top 30 keywords, reports gains/drops
- + Competitor (free) — Add/remove competitor domains to track
- Weekly Toggle (~$0.06/wk) — Auto runs discovery + rank check every Monday 6 AM
- Monthly Toggle (~$0.50/mo) — Deep analysis on 50 keywords, auto-drafts for gaps, 1st of month
How Discovery Works
Prerequisite: Requires keywords from the Research tab. Run Discover + Enrich on Research before using this.
1. Pick your keywords — Takes your already analyzed/approved keywords from Research (up to 30)
2. Find competitors in SERP — Uses SERP data (from Research's Enrich step) to check which competitor domains appear in Google top 10 for each keyword
3. Scrape their pages — Fetches each competitor page and extracts: word count, heading structure (H1-H3), FAQ section, JSON-LD schema, embedded video, data tables, internal/external link counts. Cost-safe: skips pages already scraped within 14 days, uses cached data instead
4. Compare & recommend — Generates actionable recommendations:
- Critical: No content exists but competitors rank for this keyword
- High: Your article is <70% of competitor word count, or competitor covers topics you don't
- Medium: Missing FAQ section, schema markup, or data tables that competitors have
- Low: Competitor has video or more internal links than you
How Rank Tracking Works
1. Takes your top 30 keywords (by opportunity score) that have content
2. Queries Google via DataForSEO SERP API (~$0.002/keyword) for current top 10 results
3. Finds your domain in results and records the position
4. Compares to last check and reports:
- GAIN: Position improved by 3+ spots
- DROP: Position worsened by 3+ spots
- LOST: Was in top 10, now gone
- NEW: First time appearing in top 10
History is stored per keyword so you can track trends over time.
Statuses
pending Action needed •
done Content created •
dismissed Ignored or stale
Auto-cleanup: recs auto-resolve when content is created. Stale pending (>90d) auto-dismissed.
Reddit Tab
Monitors parenting subreddits, generates AI replies linking to your blog, and flags potential new blog topics.
Discover
→
AI Analyze
→
Copy Reply
→
Paste on Reddit
Buttons
- Discover Posts (free) — Scans monitored subreddits for recent relevant posts
- Analyze (~$0.01) — AI generates a helpful reply with blog link for each post
- Copy Reply (free) — Copies AI reply to clipboard for pasting on Reddit
- + URL (free) — Manually add any Reddit post URL to analyze
- + Sub (free) — Add/remove subreddits to monitor
Backlink Strategy
When AI analyzes a post, it checks if an existing published blog matches the topic. If yes, the generated reply naturally includes a link like "I wrote more about this at kidkarma.app/blog/...". If no matching blog exists, it flags the topic for a new draft instead.
Rules: Reply must be genuinely helpful first. The link feels like sharing a resource, not advertising. Never mentions the app name directly. Stays under 150 words. Sounds like a real parent.
Manual process: Replies are drafts only — you copy and paste them on Reddit yourself. This is intentional: Reddit bans accounts that only self-promote.
Statuses
fetched Discovered →
analyzed Reply generated →
draft_created Blog draft made •
archived Fully processed •
skipped Not relevant
Auto-cleanup: processed posts auto-archive. Stale fetched (>30d) auto-archived.
Content Tab
Generates AI blog drafts, lets you review/approve/reject, and publishes to kidkarma.app. Max 2 articles/week.
Generate → Review → Approve → Publish to Dev → Review URLs → Deploy to Prod
On approve: auto-generates hero image + social drafts. On publish: deploys to dev first (kidkarma-revamp.pages.dev).
Buttons
- Generate Drafts (Claude API) — Creates 3-4 drafts from compete gaps + Reddit posts + approved keywords
- Publish Approved (fal.ai ~$0.04/img) — "Publish Only" or "Publish + Promote" (+ image + social + newsletter). Max 2/week
- Deploy to Production (free) — After reviewing dev URLs, pushes to kidkarma.app
- Discover pSEO (Claude) — Finds programmatic SEO patterns, registers keywords for future drafts
- Archive All (free) — Hides all content to start fresh
- Clean Up Empty (free) — Deletes drafts with no body text
Deploy Flow
1. Publish Approved → builds + deploys to kidkarma-revamp.pages.dev (dev)
2. Review each article on dev preview URLs
3. Deploy to Production → rebuilds + deploys to kidkarma.app
Both repos live on GitLab. Docker on .17 runs autopilot. Wrangler deploys to Cloudflare Pages.
Statuses
draft AI-generated →
in_review Under review →
approved Ready to publish →
published Live on dev •
rejected Not suitable •
archived Hidden
Draft sources: Compete gaps (2) → Reddit posts (1) → Approved keywords (fills rest). Limits: 7 drafts/week (soft), 2 publishes/week (hard).
Promote Tab
Three sections: Images (fal.ai hero images), Social Media (Publer), and Newsletter (VBOUT). Most runs automatically via "Publish + Promote".
Publish
→
Hero Image
→
Social Posts
→
Newsletter
Sections
- Images — Hero image gallery. "Generate All Images" batch-creates missing ones (~$0.04/image)
- Social Media — Publer posts (Instagram + LinkedIn). "Sync from Publer" pulls latest status
- Newsletter — Blog alerts (auto) + weekly digest (manual). "Send Weekly Digest" shows preview first. "View Subscribers" for contact list
What Runs Automatically
- On Approve: Social post drafts created (Instagram + LinkedIn)
- On Publish + Promote: Hero image → push social to Publer → blog alert newsletter drafted
- On Publish Only: Content deployed, no promotion
Settings Tab
Three read-only sections for configuration visibility.
- Writer Profile — Author name, bio, E-E-A-T signals. Read-only; managed in website source code
- Budget & Costs — Total API spend, per-service breakdown, monthly trends
- Connected Services — Which APIs are configured and their env var status
Intelligence Tab
Content quality insights across three sections.
- Performance — Published count, avg GEO score, top 5 and bottom 5 content
- AI Visibility — GEO score distribution (high 70+, medium 40-69, low <40)
- Self-Healing — Auto-detects: no body, no image, low GEO score
GEO Score
Measures how likely AI assistants (ChatGPT, Perplexity) are to cite your content. Based on: named author, citations, structured data, practical advice. Higher = better AI visibility.
Logs Tab
System activity log with cost tracking. Filter by module or status.
- Cost Strip — Total entries, success/failed counts, total API cost
- Module Filter — Filter by keyword_research, content_gen, publisher, social, etc.
- Status Filter — Filter by success, failed, or skipped
🔬
Pipeline: Discover Keywords → Enrich & Score → Approve best → auto-feeds into Content → Generate Drafts. Auto-cleanup: Keywords auto-update to "content_created" when a draft is generated, and "published" when content goes live. Stale researched keywords (>90 days) are auto-skipped. Skipped keywords >60 days are auto-deleted.
Actions
Active FunnelsiClick a funnel chip to bulk-approve or revoke all keywords in that funnel stage. Approved keywords become available for draft generation in the Content tab. TOFU = broad awareness topics, MOFU = comparison/guide content, BOFU = high-intent download/buy keywords.
Loading...
Funnel
Status
SERP
Book
Search
0 selected
| Keyword | VolumeiMonthly Google search volume. Higher = more traffic potential. Target 100-1000 for low-competition wins. | DiffiKeyword difficulty (0-100). Lower = easier to rank. Under 40 is ideal for new sites. Over 60 needs strong domain authority. | CPCiCost Per Click ($) — what advertisers pay per Google Ads click for this keyword. High CPC = high commercial intent. Shows how valuable the traffic is. $1+ means people are willing to pay to reach this audience. | SourceiWhere this keyword was discovered. Gap = competitor keyword gap, Reddit = trending Reddit topic, PAA = People Also Ask, Seed = original seed expansion. | FunneliTOFU = awareness (broad topics), MOFU = consideration (comparisons, guides), BOFU = decision (ready to buy/download). Mix all three. | OpportunityiCombined score (0-100) weighing volume, low difficulty, business relevance, and SERP signal. Higher = better keyword to target. | SERPiSearch results signal. Green = weak competition (can rank), Yellow = moderate, Red = dominated by big sites. Based on domain authority of top 10 results. | BookiWhether our knowledge base (3 parenting books, 117 chunks) has matching content for this keyword. Book-matched articles score higher for E-E-A-T. | Status | Actions |
|---|
📝
Pipeline: Generate Drafts (from Compete gaps + Reddit posts + approved keywords) → Review → Approve → Publish (2/week max, 7 drafts/week max). Auto-cleanup: When a draft is generated, the source keyword updates to "content_created" and related Compete recs are auto-resolved. On publish, keyword updates to "published" and linked Reddit posts are auto-archived.
🌐 Deploy Review
▶
How it works: "Publish Approved" deploys to dev (kidkarma-revamp.pages.dev). Review your articles there. When satisfied, click "Deploy to Production" to push to kidkarma.app.
| Article | Type | Published | Dev Preview | Production |
|---|
Type
Status
| Title | TypeiContent type: blog (keyword), reddit (from Reddit discussions), comparison (vs pages), pseo_persona (age-specific), faq, use_case. | Keyword | WordsiTotal word count. Target 1200-2000 for blogs. Competitor articles averaging higher? The Compete tab will flag it. | SEOiSEO score (0-100). Checks keyword usage, heading structure, word count, FAQ section, meta description. Aim for 80+ before approving. | GEOiGEO score (0-100). Measures LLM/AI citability — self-contained FAQ answers, BLUF opening, structured data readiness. Aim for 75+. | Book %iPercentage of content backed by knowledge base books. Higher = more research-backed. 0% means no book match was found for this keyword. | Status | Actions |
|---|
Promotion Pipeline
When you Publish + Promote: images are auto-generated, social posts are pushed to Publer, and a newsletter is drafted for your review before sending.
Images
▶| Content | Type | Preview | Status | Created |
|---|
Social Media
▶Loading connected accounts...
| Platform | Content | Caption | Status | Scheduled | Publer |
|---|
Post Preview
Newsletter
▶| Type | Subject | Recipients | Sent At | Actions |
|---|
Newsletter Preview
Publish Content
Subscribers
| Name | Status | Subscribed | |
|---|---|---|---|
| Loading... | |||
Compete
Tracked Competitors
iAdd competitor domains to track. Discovery will scrape their pages and compare keywords, word count, structure, and rankings against yours. Use domains only (e.g. busykid.com), not full URLs.
💡
Competitor analysis and content gaps. Fill gaps via Content → Generate Drafts. Auto-cleanup: When you generate a draft or publish content for a keyword, all related recommendations here are auto-resolved. Done/dismissed recs older than 30 days are auto-deleted. Stale pending recs (>90 days) are auto-dismissed.
| Competitor | OverlapiHow many of your target keywords this competitor also ranks for on Google. Higher overlap = more direct competition. | Their Avg PosiCompetitor's average Google position across shared keywords. Position 1 is best. If lower than yours, they outrank you. | Our Avg PosiYour average Google position for the same keywords. Compare to theirs — if yours is higher (worse number), you need to improve content. | Beats UsiNumber of keywords where this competitor ranks higher than you. These are the keywords to focus your content improvements on first. | PagesiNumber of competitor pages we've analyzed. More pages = better data for recommendations. |
|---|
CONTENT GAPSiKeywords where you have NO content but competitors do. These are missed opportunities — competitors get traffic you don't. "Generate Next Draft" auto-fills the top priority gap.
—
THEY BEAT USiKeywords where you HAVE content but competitors rank higher. Use "Improve" to expand word count, add FAQ sections, or fix structure to overtake them.
—
PENDINGiTotal recommendations waiting for action. Each tells you exactly what to do — create missing content, expand a short article, or add FAQ/tables.
—
TRACKEDiCompetitor domains being monitored. Discovery scrapes their pages weekly to find what keywords they rank for that you don't.
—
Recommendations
💬
Generate replies and copy-paste to Reddit for traffic. Replies include your blog links automatically. High-engagement posts without a blog auto-feed into Content → Generate Drafts. Auto-cleanup: Posts with reply + draft are auto-archived. Replied posts >14 days and stale posts >30 days are auto-archived. Posts whose linked blog is published are auto-archived. Use "Show archived" to see them.
Monitored Subreddits
iAdd subreddits to monitor for parenting discussions. Only relevant posts about chores, habits, and rewards are shown.
All
Writer Profile
▶
Bhagyesh Patel
ACTIVE WRITERParenting & Family Life Editor
A veteran teacher turned parenting writer whose work has appeared in The Atlantic and The New York Times. Translates child-development research into practical, guilt-free advice through personal anecdotes and warm storytelling.
linkedin.com/in/bhagyeshonline
Child Development
Parenting Research
Family Routines
Chores & Responsibility
Positive Discipline
E-E-A-T Signals (Google Trust Factors)
| Signal | What It Means | Status |
|---|---|---|
| Experience | Author has first-hand parenting experience | ✓ "mom of three" in bio |
| Expertise | Author has relevant professional credentials | ✓ "early-childhood teacher" credential |
| Authority | Named author with consistent byline across articles | ✓ Byline on every article |
| Trust | Content cites research, no misleading claims | ✓ Research-backed + knowledge DB |
Why this matters: Google ranks content higher when there's a named, credentialed author. Only S'moresUp (Reeves Xavier, CPO) does this among KidKarma's competitors — everyone else uses anonymous "admin" or brand-only attribution.
Writing Voice & Tone
| Dimension | Do This | Avoid |
|---|---|---|
| Tone | Warm, conversational, encouraging — like a knowledgeable friend | Clinical, academic, preachy, guilt-inducing |
| Perspective | First-person plural ("we") or second-person ("you") — empathetic, alongside the parent | Third-person detached, talking down |
| Evidence | Cite child development research naturally, paraphrase from knowledge base | Direct book quotes, academic citations, "studies show" overuse |
| Structure | Practical tips, age-specific advice, actionable steps, real examples | Vague generalities, walls of text, no clear takeaways |
| Brand Voice | Mention KidKarma naturally where relevant, not in every paragraph | Hard-selling the app, forced product mentions |
How the Writer Profile Is Used
1. Article Byline: Every published article shows "Written by Bhagyesh Patel, Parenting & Family Life Editor at KidKarma"
2. AI Prompt: The tone description is injected into the AI writing prompt so articles match this voice consistently
3. SEO Schema: JSON-LD Person schema is generated for the author on each blog page — helps Google Rich Results
4. GEO Score: Named author attribution adds 8 points to the GEO (AI-citability) score
Note: This writer profile is statically managed in the marketing website source code (
website/). To change the author name, bio, or photo, update the Next.js website directly and redeploy. The autopilot reads this profile for AI content generation but does not modify it.
Budget & Costs
▶| Service | Job Type | Usage | Last Used | Total Cost |
|---|
Monthly Breakdown
| Month | Service | API Calls | Cost |
|---|
Connected Services
▶Tenants
▶Performance
▼Top Content (by GEO Score)
Needs Improvement
AI Visibility (GEO Scores)
▶| Article | GEO Score | Word Count | Status |
|---|
Self-Healing
▶Cost by Tool
Cost by Session (Day)
| Time | Module | Action | Status | Details | CostiAPI cost for this action in USD. Most are $0 (free). DataForSEO ~$0.002-0.075/query, Claude AI ~$0.05/article, fal.ai ~$0.04/image. |
|---|