X9Elysium Admin Dashboard — Tableau Integration Plan

Goal: A single private admin dashboard at x9elysium.com/admin that shows everything Darsh needs to run the agency in one view — inbound leads, SEO performance, website traffic, Core Web Vitals, marketing campaigns, and internal tasks.

Audience: Darsh + Adhvait only. Never indexed. Never linked from public nav, sitemap, or footer.

Built with: Tableau (Cloud or Public, see §3) embedded into the X9Elysium static site.

1. The 4 dashboards (what your friend builds)

One Tableau workbook with four sheets / tabs, each answering one question:

Tab 1 — Leads & Pipeline

"What's in the funnel and what should I work on today?"

KPI tiles: Leads MTD · Leads last 30d · Avg first-response time · Open pipeline value · Closed-won MTD · Lead → discovery call conversion %
Lead funnel chart: New → Qualified → Discovery booked → Proposal sent → Won
Leads over time: stacked bar by source (organic, LinkedIn, referral, direct, paid)
Lead table: sortable list of recent leads with status, score, days-since-contact, next action
Lost-reason breakdown: pie chart (budget, timing, competitor, ghosted, not a fit)
Source ROI: which channel produced the highest closed-won value

Data source: leads.csv

Tab 2 — SEO Performance

"Are we ranking for the queries that matter, and is organic traffic growing?"

KPI tiles: Total ranking keywords · Top-10 keywords · Avg position (top 50) · Organic sessions MTD · Organic conversions MTD · Featured snippets owned
Keyword position heatmap: keywords × time, color = position (green good, red bad)
Top-moving keywords: biggest gainers + biggest losers in last 30 days
Top landing pages: organic sessions per URL, sortable
Click-through trend: impressions vs clicks vs CTR over time
Country split: Canada vs US vs other

Data sources: seo-rankings.csv, seo-traffic.csv

Tab 3 — Website Health

"Is the site fast, are people converting, and where are they coming from?"

KPI tiles: Sessions today · Sessions MTD · Avg LCP · Avg INP · CLS · Conversion rate · Bounce rate
Core Web Vitals trend: LCP / INP / CLS over time, green/yellow/red thresholds
Traffic over time: line chart, sessions + users
Top pages by views
Acquisition mix: donut by channel (organic, direct, social, referral, email)
Geo map: sessions by country

Data sources: website-analytics.csv, core-web-vitals.csv

Tab 4 — Tasks & Operations

"What's on my plate this week, how did we do this quarter, and what's slipping?"

KPI tiles (current): Open tasks · Overdue · Due this week · Hours estimated this week · Hours logged
Quarterly performance row (filterable by quarter, defaults to current):
- Tasks completed (QN) · Completion rate (Done / Total created in quarter) · On-time rate (Done before due / Done) · Avg cycle time · Estimate accuracy (actual / estimated, target 1.0)
- Q1 quick read example: if 80 tasks were created Jan 1 – Mar 31 and 62 closed by Apr 1, that's a 78% completion rate. Use a colored badge: ≥85% green, 70–84% amber, <70% red.
Task burndown: tasks created vs closed per week (quarter-bounded line chart with cumulative totals)
Quarterly comparison: small bar chart — completed tasks Q4 2025 vs Q1 2026 vs Q2 2026 (in progress)
By project: stacked bar, open tasks per client/project
By assignee: workload split Darshan vs Adhvait (open + completed in quarter)
By category: how many Build / Design / Sales / Audit tasks completed in quarter
Slipping tasks: list of overdue with days late
Throughput: avg cycle time (created → done) per category

Data source: tasks.csv

Tab 5 — Finance & Operations

"How much did we earn, how much did we spend, who owes us money, and what's our runway?"

KPI tiles: Revenue MTD · Revenue QTD · Outstanding A/R · Overdue A/R · Expenses MTD · Net cash flow MTD · Monthly recurring SaaS cost · DSO (Days Sales Outstanding)
Cash flow line chart: monthly revenue (paid invoices) vs monthly expenses, last 6 months
A/R aging bucket: stacked bar — 0–30 / 31–60 / 61–90 / 90+ days outstanding
Invoice status funnel: Draft → Sent → Viewed → Paid (drop-off visible)
Top clients by paid revenue (bar chart, USD)
Expense breakdown: treemap or donut by category (Software/SaaS, Hosting, Marketing, Travel, Meals, Office, Contractor, Education)
Recurring vs one-off expenses: stacked bar by month — what's the always-on baseline?
Unbilled reimbursables alert: list of billable_to_client = Yes expenses with no matching invoice line item
Sample invoice: see sample-data/sample-invoice.md for the standardized X9Elysium invoice template

Data sources: invoices.csv, expenses.csv

"Is the X.com posting cadence producing reach and engagement, and which thoughts land?"

KPI tiles: Posts last 30d · Total impressions 30d · Avg engagement rate 30d · Top post (likes + RTs + replies) · Posts in queue (rough — derived offline from data/x-thoughts.md count) · Days since last post
Posts per day timeline: sparkline / bar — confirm the 2x/day cadence is actually firing
Engagement-rate trend: line chart — engagement-rate per post over time, with a rolling 7-day average overlay
Top 10 posts by engagement: sortable table (text snippet + impressions + likes + RTs + replies + engagement_rate + url-out to x.com)
Format breakdown (when we add the column later): single-thought vs question vs list — which structure performs best
Post-time analysis: heatmap weekday × hour-of-day, color = avg engagement_rate (helps validate / change the 09:13 + 17:13 PT cadence)

Data source: x-posts.csv

Notes:

engagement_rate is precomputed in the CSV: (likes + retweets + replies) / impressions. Empty when impressions is null (Free tier sometimes returns null).
impressions may be missing on the Free X API tier. Treat as best-effort. Engagement counts (likes/RTs/replies) are always populated.
text field is the full tweet — useful for tooltips on every chart so a click can show the actual post.
Source of truth for posting cadence is the GitHub Actions cron (.github/workflows/x-post.yml) — if the timeline shows missed days, check the Action log not the dashboard.

Optional Tab 7 — Marketing Campaigns if you want to track LinkedIn / paid: use campaigns.csv. Instagram is intentionally excluded from the brand strategy (see docs/marketing/x-automation-plan.md).

2. Where the real data comes from (production pipeline)

Sample CSVs are for him to design with. In production, a single Google Sheets workbook acts as the unified store, with each tab populated automatically:

Tableau dataset	Real source	Pipeline
Leads	Web3Forms (contact form)	Web3Forms webhook → Cloudflare Worker → Google Sheets API (`Leads` tab). Manual columns (status, score, deal value, lost reason) edited in Sheet directly.
SEO rankings	Google Search Console + a rank tracker (Ahrefs / SEMrush / SerpRobot)	Daily Google Apps Script pulls GSC API → `SEO_Rankings` tab. Rank tracker has its own GSheet export.
SEO traffic	Google Search Console (clicks, impressions, CTR, position per page/query)	Daily GSC API pull via Apps Script → `SEO_Traffic` tab.
Website analytics	GA4 OR Plausible OR Cloudflare Web Analytics	GA4 → BigQuery export OR Plausible API → Apps Script → `Web_Analytics` tab.
Core Web Vitals	PageSpeed Insights API or CrUX (Chrome UX Report)	Weekly Apps Script that hits PSI for ~10 key URLs → `CWV` tab.
Tasks	Asana or Jira (see §9) — recommend Asana for this scale	API → Apps Script (or Make.com) → `Tasks` tab nightly.
Invoices	Tally Create Invoice form + Update Invoice Status form	Tally → Google Sheets API → `Invoices` tab. PDF generated from the Sheet row via Apps Script + Google Docs template.
Expenses	Tally Log Expense form (mobile-friendly — log on the go)	Tally → Google Sheets API → `Expenses` tab. Receipt photo uploaded to Google Drive, link stored in row.
Campaigns (optional)	LinkedIn Ads / Meta Ads / Instagram insights	Manual entry weekly, or platform → Sheets via Zapier / Make.

Why Google Sheets as the middle layer:

Tableau has a first-class Google Sheets connector (5-min setup)
You can hand-edit qualitative columns (status, lost reason, lead score)
Every API in the universe writes to Sheets
Free, versioned, shareable
No infra to babysit

Refresh cadence:

Tableau Cloud auto-refreshes from Google Sheets daily (or hourly on paid tiers).
Apps Script time-driven triggers run nightly at 2am local time to repopulate.

2.5 Forms-driven entry layer (the "everything connected by forms" piece)

For all data that doesn't have an upstream API (invoices, expenses, manual notes), use Tally forms as the entry point. Tally writes directly to Google Sheets, which Tableau already reads. No middleware, no copy-paste, no spreadsheets-as-databases-edited-by-hand.

Why Tally specifically:

Free tier covers everything you need (unlimited forms, unlimited submissions).
Native Google Sheets integration (one click, no Zapier needed).
Conditional logic, file uploads, calculated fields, hidden fields, password protection.
Works on phone — log expenses from your wallet at the cafe.
Embeddable into the admin dashboard page so all 3 forms live alongside the dashboards.

Alternatives if you don't like Tally: Fillout (similar feature set, slightly nicer UI, free tier weaker), Google Forms (free, but ugly + no conditional logic + no file uploads to Drive in one step), Typeform (paid, beautiful, overkill).

The forms inventory

Form	What it captures	Writes to	Triggered by
Public Contact Form (already exists — Web3Forms on `/contact`)	Lead name, email, company, topic, budget	`Leads` Sheet (via Cloudflare Worker webhook)	Public site visitor
Lead Status Update (admin-only Tally form)	Lead ID, new status, score, deal value, notes	Updates an existing row in `Leads` Sheet	Darsh / Adhvait, after a discovery call
Create Invoice (Tally)	Client, project, line items (repeatable group), terms, currency	New row in `Invoices` Sheet + triggers PDF generation via Apps Script	After a deal is signed
Update Invoice Status (Tally)	Invoice ID, new status, paid amount, paid date, payment method	Updates existing row in `Invoices` Sheet	When client pays / partially pays
Log Expense (Tally — mobile-first)	Vendor, category, amount, currency, payment method, billable, receipt photo upload	New row in `Expenses` Sheet	Anytime you spend money for the business
New Task (only if NOT using Asana/Jira)	Title, project, assignee, due date, priority, est hours	New row in `Tasks` Sheet	Skip this if you adopt Asana/Jira — use their UI instead

Form → Sheet → Tableau wiring

Tally Form
   │
   │  (native Tally → Google Sheets integration, 1-click setup)
   ▼
Google Sheet tab (Invoices / Expenses / Leads / etc.)
   │
   │  (optional: Apps Script trigger on edit — auto-set status,
   │   compute derived fields, send notification email)
   ▼
Tableau Cloud (refreshes from Sheet daily)
   │
   ▼
Embedded in app/admin/page.tsx via Tableau Embedding API

Embedding the forms next to the dashboards

The admin page can show the dashboards on top, and below them a small "Quick Actions" row:

┌─ X9Elysium Admin ──────────────────────────────────────────┐
│  [Tab 1: Leads]  [Tab 2: SEO]  [Tab 3: Web]  [Tab 4: Tasks]│
│  [Tab 5: Finance]                                          │
│                                                            │
│  ▼ Embedded Tableau workbook                              │
│  ────────────────────────────────────────                 │
│                                                            │
│  Quick Actions:                                            │
│  [+ Log Expense]  [+ New Invoice]  [+ Update Status]      │
│  (each opens a Tally embed in a modal)                     │
└────────────────────────────────────────────────────────────┘

Tally provides <iframe> and modal-popup embeds out of the box — drop them straight into app/admin/page.tsx.

Optional — Apps Script automation hooks

Worth adding once Tally → Sheets is wired:

Auto-set invoice status = Overdue for any row where due_date < today AND status NOT IN (Paid, Void). Runs nightly.
Email reminder to AP contact when an invoice flips to Overdue.
Email Darsh + Adhvait a Monday morning digest: open A/R, expenses last week, upcoming due invoices.
Auto-render PDF from the Invoices Sheet using a Google Docs template on row creation.

All of these are 30–60 lines of Apps Script each. Build them later — get the dashboard live first.

3. Tableau hosting — pick one

Option	Cost	Auth	Verdict for X9Elysium
Tableau Public	Free	None — dashboards are publicly viewable if you have the URL	❌ Don't use. Lead names + emails would be exposed.
Tableau Cloud (Creator)	~US$75 / user / month, billed annually	Tableau-managed SSO, embed tokens (JWT)	✅ Recommended. Cheapest legitimate path for a private admin dashboard. 1 Creator seat covers Darsh, Adhvait gets a Viewer seat at ~$15/mo.
Tableau Server	Self-hosted, infra cost + license	Yours	❌ Overkill for 2 users.
Free alternatives	$0	Varies	If Tableau cost stings: Looker Studio (free, GA4-native, embeddable, weakest for custom data shapes), Metabase Cloud (~$85/mo Starter, beautiful, great for SQL teams), Hex (free tier, notebook-style).

Recommendation: Start with Tableau Cloud. If your friend already has a workbook built, this is the path of least resistance. ~US$90/mo total for both founders, all-in.

4. Embedding into x9elysium.com

The X9Elysium site is a static export on Cloudflare Workers — no Node runtime in production. So:

Route

Create app/admin/page.tsx (App Router, client component)
Not linked from anywhere public
Excluded from app/sitemap.ts
Excluded from public/llms.txt
noindex, nofollow meta tag on the page

Auth (two layers — pick one or both)

Layer A — Cloudflare Access (recommended, free for ≤50 users):

In Cloudflare Zero Trust dashboard, protect x9elysium.com/admin/* with an Access Application.
Email-based auth: only fd3687@yahoo.ca and Adhvait's email can pass.
Cloudflare injects an auth header before the request ever hits the page.
Zero code on your end.
This is the cleanest option.

Layer B — Client-side PIN (mirror the journal pattern):

Same approach as /docs/journal — AES-GCM + PBKDF2-SHA-256, 100k iterations.
Ciphertext-only ships in the bundle; PIN entry decrypts in browser.
Less secure (determined attacker with the bundle could brute-force a 4-digit PIN), but zero infra.
Not appropriate alone for a dashboard with real lead emails. Use Cloudflare Access.

Embedding the Tableau dashboard

Tableau Cloud uses the Tableau Embedding API v3. In app/admin/page.tsx:

"use client";
import Script from "next/script";

export default function AdminPage() {
  return (
    <>
      <Script
        type="module"
        src="https://prod-useast-a.online.tableau.com/javascripts/api/tableau.embedding.3.latest.min.js"
      />
      <main className="min-h-screen bg-black p-6">
        <h1 className="text-white text-2xl mb-4">X9Elysium Admin</h1>
        {/* @ts-ignore — custom element from Tableau JS */}
        <tableau-viz
          id="tableauViz"
          src="https://prod-useast-a.online.tableau.com/t/x9elysium/views/X9Elysium-Admin/Leads"
          toolbar="bottom"
          hide-tabs={false}
          width="100%"
          height="900"
          token="<JWT-FROM-CLOUDFLARE-WORKER>"
        />
      </main>
    </>
  );
}

JWT generation: Tableau Cloud uses Connected Apps. A Cloudflare Worker (one tiny endpoint) signs a JWT with the Connected App secret and returns it to the page on load. The Worker is the only thing that ever holds the secret.

Alternative if JWT/Connected App is too much: publish the workbook as "Embed Password Saved", set a Tableau Cloud password on the workbook, and embed via iframe. Less elegant, works fine.

5. Friend's deliverable checklist

Send your friend this folder:

docs/admin-dashboard/
├── tableau-integration-plan.md    ← this file
├── friend-brief.md                ← what to read first
└── sample-data/
    ├── README.md                  ← schema for every CSV
    ├── leads.csv
    ├── seo-rankings.csv
    ├── seo-traffic.csv
    ├── website-analytics.csv
    ├── core-web-vitals.csv
    ├── tasks.csv
    └── campaigns.csv

He builds one Tableau workbook with 4 sheets matching §1. When ready, he publishes to a Tableau Cloud site (you'll provision one), and we wire the embed.

6. Phased rollout

Phase 1 — Design (week 1, friend builds):

Friend builds workbook against sample CSVs.
Iterate on chart choices + layout via screenshots.
Lock the visual design.

Phase 2 — Real data wiring (week 2):

Stand up the Google Sheets workbook with 7 tabs matching the CSV schemas.
Write Apps Scripts: GSC daily pull, GA4/Plausible daily pull, PSI weekly pull.
Web3Forms → Cloudflare Worker → Sheets webhook for new leads.
Friend repoints workbook from CSV to Google Sheets connectors.

Phase 3 — Embed (week 3):

Provision Tableau Cloud site, publish workbook.
Build app/admin/page.tsx with Tableau Embedding API.
Cloudflare Access in front of /admin/*.
Cloudflare Worker for JWT signing.

Phase 4 — Use it (ongoing):

Daily morning check.
Adjust dashboards monthly as questions evolve.

7. Cost summary

Line item	Monthly cost
Tableau Cloud — 1 Creator + 1 Viewer	~US$90
Asana Starter — 2 users (recommended for tasks, see §9)	~US$22
Tally — Free tier (forms)	$0
Cloudflare Access (≤50 users)	$0
Cloudflare Workers (current)	$0
Google Sheets / Apps Script / Drive	$0
Webhooks (Web3Forms → Worker)	$0
Total	~US$112/mo

Free alternative if cost is the blocker: Looker Studio + Google Sheets + Asana free tier (≤10 users) + Tally free covers ~80% of this for $0. Less polished, no real "embed with auth" story for Looker Studio (embeds are public-by-link), but functional.

8. Open decisions for Darsh

Tableau Cloud (paid) vs Looker Studio (free) — pick your hosting before friend starts polishing.
GA4 vs Plausible vs Cloudflare Web Analytics for the analytics feed — Plausible is cleanest, GA4 has more depth, Cloudflare is free + already on infra.
Asana vs Jira for tasks — see §9. Recommendation: Asana.
Tally vs Fillout for the data-entry forms — both free at the volume you'll hit.
Confirm Cloudflare Access is OK with both founders' emails, or want a shared password instead.

9. Asana vs Jira (which one + Claude integration)

Short answer: Asana for the agency. Use Jira only if you're going to live inside Claude/MCP and want first-party support.

Honest comparison

Dimension	Asana	Jira	Winner for X9Elysium
Daily UX for non-engineers	Clean, list/board/timeline views, low cognitive overhead	Powerful but cluttered, designed for software teams	Asana — your clients are e-comm operators, not engineers
Onboarding speed (a new client/contractor)	~10 min	~1 hour	Asana
Free tier	Up to 10 users, unlimited tasks/projects	Up to 10 users, but only Standard plan unlocks reporting	Tie (both work for 2-person team)
Paid tier (your scale)	Starter US$10.99/user/mo	Standard US$7.53/user/mo	Jira (cheaper)
API for syncing to Tableau	Mature REST API, great docs	Mature REST API, even better docs (Atlassian)	Tie
Claude / MCP integration	Community MCP servers exist (e.g. `mcp-server-asana`), no first-party offering as of early 2026	Atlassian Remote MCP Server (official, first-party, polished) — covers Jira + Confluence	Jira — by a meaningful margin
Automations / workflow rules	Asana Rules — visual, easy	Jira Automation — more powerful, steeper curve	Tie (depends on appetite)
Built-in AI	Asana Intelligence (summaries, smart fields)	Atlassian Intelligence (similar)	Tie
Fits a consulting agency vs internal product team	Built for cross-functional / external collaboration	Built for engineering teams shipping software	Asana

Claude / MCP — the deciding factor only if it's actually deciding

Both work fine with Claude — neither is a wall.

Asana via Claude: community MCP servers (mcp-server-asana and friends) cover the basics — list tasks, create tasks, move statuses, add comments. They're solid for the things you'd actually ask Claude to do for you. Maintenance is community-driven, so quality varies.
Jira via Claude: Atlassian shipped an official Remote MCP Server in 2025 that covers Jira + Confluence with full first-party support. It's the most polished MCP integration of any major PM tool right now. If you imagine yourself saying "Claude, summarize all open Northwind tasks and draft an end-of-week update," Jira's path is genuinely smoother today.

My recommendation: Asana

Three reasons:

You'll bring clients into it. Asana is a much easier ask for a Director of Ecommerce at a Shopify Plus brand than Jira. The UX delta matters more than the MCP delta.
You're not engineering-heavy. Jira's strengths (sprints, story points, custom workflows, Bitbucket integration) don't apply. You'd be paying the UX tax for features you won't use.
Claude integration is good enough on Asana. Community MCP servers do what you need. If you go heavily MCP-dependent later, you can migrate — but that's an unlikely future.

Pick Jira instead if: you primarily operate via Claude prompts (not the PM tool's UI), you want first-party MCP guarantees, and clients are unlikely to be added to your project tracker.

Either way — same Tableau pipeline

Both tools have REST APIs. An Apps Script (run nightly) pulls all tasks → writes to the Tasks Sheet → Tableau refreshes. Schema for the Sheet stays exactly as documented in sample-data/README.md → tasks.csv, so you can swap the source later without touching the workbook.