PostHog Onboarding Lead Research & Qualification

Workflow

Run these five workstreams. Steps 1-2 use web search. Steps 3-4 use Vitally MCP. Step 5 combines web research with Vitally/SFDC/Harmonic data from traits. Interleave them efficiently — don't wait for one to finish before starting the next.

Step 1: Company Research & Growth Trajectory

Goal: Understand who this company is and whether they're on a growth path that makes $2K+/month likely.

Web searches to run:

[company name] what does it do
[company name] crunchbase funding employees
[company name] series funding 2024 2025
[company name] engineering team hiring
[company name] product launch growth

Gather:

• One-sentence description of what the company does
• Industry / vertical
• Company size (employees, engineering headcount if available)
• Funding stage and recency (series, amount, date)
• Revenue estimate if available
• HQ location
• Product type — is this a software product company? SaaS? Marketplace? E-commerce?
• Recent news — launches, expansions, partnerships, hiring sprees
• Growth signals — new product lines, international expansion, increasing headcount

Growth trajectory assessment: Based on the research, classify the company's growth trajectory:

Step 2: $2K+ Monthly Spend Likelihood Score

Goal: Score the probability this account will cross $2K/month within 3-6 months.

Read references/scoring-model.md for the full scoring criteria.

Quick summary of the scoring:

• Positive signals: growth trajectory, multi-product usage, engineering team size, funding recency, industry fit, active user count, event volume trends
• Negative signals: flat/declining usage, no engineering growth, non-ICP industry, single product user, low user count

Output a score from 0-100 with a tier:

• 80-100: High confidence — Strong growth signals, multi-product, expanding team. Will likely hit $2K+ organically.
• 60-79: Moderate confidence — Good signals but some uncertainty. May need a nudge (e.g. new product adoption, config optimization).
• 40-59: Low confidence — Mixed signals. Could go either way. Worth engaging but temper expectations.
• 0-39: Unlikely — Flat usage, small team, no growth signals. May stay at current spend.

Step 3: Vitally Usage & Billing Analysis

Goal: Pull hard data on what this account is actually doing in PostHog and where the spend is going.

3a: Find the Account

Search Vitally for the account. Try these in order:

1. search_accounts_advanced with the company name
2. If that fails, ask the TAE for a contact email and use it to look up via user details

3b: Pull Full Account Data

Once you have the account ID, pull:

1. get_account_full (with detailLevel: "full") — Gets everything: MRR, health scores, traits, product usage, billing data, team assignments
2. get_account_health — Detailed health score breakdown

Always extract these three fields for the Quick Links section:

• organization_id (in traits) → Vitally URL: https://posthog.vitally-eu.io/customers/{organization_id}
• sfdc.Id (in traits) → Salesforce URL: https://posthog.my.salesforce.com/{sfdc.Id}
• usage_dashboard_link (in traits) → direct Metabase usage dashboard link

Extract and analyze:

Current billing:

• Current MRR / monthly spend
• Which products are driving the spend (Analytics, Session Replay, Feature Flags, etc.)
• Billing plan (pay-as-you-go vs. annual)

Usage trends (from traits):

• Event volume (current and trend)
• Session replay volume
• Feature flag evaluations
• Active users count and trend
• Products enabled vs. products actively used

Growth indicators:

• Is event volume increasing month over month?
• Are new products being adopted?
• Are new users being added to the org?
• Is there a gap between products enabled and products used? (expansion opportunity)

Trajectory projection: Based on the trends, project where this account lands in 3 months:

• Current monthly spend: $X
• Projected monthly spend (3 months): $X
• Key driver of growth: [product/volume/new users]
• Risk factors: [anything that could stall growth]

3c: Check for Other Accounts on the Same Domain

Large companies sometimes have multiple PostHog orgs. Search for other accounts with similar names to flag consolidation opportunities.

Step 4: Onboarding Team Context

Goal: Get the full history of what the onboarding team discussed, tried, and learned about this customer.

4a: Vitally Notes

Use get_account_notes (with a high limit, e.g. 20) to pull all notes on the account. The onboarding team typically logs:

• Initial outreach attempts and responses
• Call notes from onboarding sessions
• Configuration changes made
• Product recommendations given
• Customer pain points and goals
• Reasons for the handoff to sales

Summarize the onboarding narrative:

• Did the customer engage with the onboarding team? (calls, email replies, or silence)
• What did the onboarding team help with?
• What products/features were discussed?
• What pain points or goals did the customer mention?
• Why is the onboarding team handing off now?
• Any open issues or unfinished business?

4b: Vitally Conversations

Use get_account_conversations to pull email threads and other conversations. These often contain:

• Email exchanges between the onboarding rep and the customer
• Details the customer shared about their use case
• Commitments or follow-ups that were promised

Extract key details:

• Who was the primary contact on the customer side?
• What was the tone of engagement? (enthusiastic, transactional, unresponsive)
• Any specific asks or blockers the customer raised?
• Were there any promises or follow-ups left open?

4c: Granola Meeting Notes (if applicable)

If the TAE or the onboarding team had calls logged in Granola, query for them:

Query: "[company name] onboarding" or "[company name] [contact name]"

This surfaces any call transcripts with additional detail beyond what's in Vitally.

Step 5: Business Growth Trajectory & PostHog Spend Mapping

Goal: Assess how fast the company's business is growing, because that directly predicts how fast their PostHog usage (and spend) will grow. This is the step that catches cases where Vitally billing data lags the real trajectory — a company doubling its customer base will eventually double its event volume, even if the billing hasn't caught up yet.

Web searches to run:

[company name] Inc 5000 revenue growth
[company name] new customers partnerships 2025 2026
[company name] product launches announcements
[company name] hiring engineering jobs

Also cross-reference signals already gathered from Vitally traits:

• sfdc.harmonic_headcount__c vs sfdc.harmonic_headcount_180d__c — headcount delta
• sfdc.harmonic_headcountEngineering__c vs sfdc.harmonic_headcountEngineering_180d__c — eng headcount delta
• sfdc.harmonic_web_traffic__c vs sfdc.harmonic_web_traffic_90d__c / sfdc.harmonic_web_traffic_180d__c — traffic trends
• sfdc.webtraffic_growth__c — percentage growth
• sfdc.lnkd_follower_change__c — LinkedIn momentum
• sfdc.eng_headcount_change__c and sfdc.headcount_eng_growth__c — engineering team growth

Gather and assess these six growth vectors:

5a: Revenue & ARR Growth

• Revenue estimates from funding announcements, Inc. 5000 rankings, press releases
• Any disclosed growth rates (e.g. "doubled ARR", "3x revenue")
• Evidence of pricing power or contract expansion (moving upmarket, bigger deals)

5b: Funding & Investor Signals

• Recent fundraise amount, date, lead investor, and what the funds are earmarked for
• Repeat investors (insiders participating again = confidence)
• Are they raising a growth round (go-to-market) or an R&D round? GTM rounds drive PostHog usage faster.

5c: Product Velocity

• New product launches or major features shipped in the last 12 months
• New product lines = new surfaces to instrument = more PostHog events and replays
• Platform plays (APIs, integrations, clearinghouses, marketplaces) create multiplicative event growth

5d: Customer / Market Expansion

• New customer segments being targeted (enterprise, SMB, new verticals, new geographies)
• Partnership announcements (channel partners, tech partnerships, integrations)
• Each new customer on their platform = more end-user sessions flowing through PostHog

5e: Headcount & Hiring Trajectory

• Total headcount delta (6-month and 12-month view using Harmonic data from Vitally)
• Engineering team growth specifically — more engineers = more features shipped = more things to instrument
• Go-to-market hiring (sales, CS, marketing) = more internal PostHog dashboard users
• Job postings on their careers page — what are they hiring for right now?

5f: Web Traffic & Brand Momentum

• Web traffic trends from Harmonic/SFDC data
• LinkedIn follower growth
• Industry recognition (awards, rankings, conference appearances)
• Content marketing velocity (blog posts, reports, webinars) — indicates go-to-market investment

Map business growth to PostHog spend growth:

After gathering these signals, explicitly connect the company's business growth to expected PostHog usage growth. This is the key insight - it tells the TAE whether the billing trajectory in Vitally is understating the real growth potential.

The mapping logic:

• More customers on their platform → more end-user sessions → more events, replays, and flag evaluations
• New product launches → new surfaces to instrument → broader PostHog product adoption (analytics + replay + flags)
• Engineering team growth → more features shipped → more things tracked → higher event volume
• GTM team growth → more internal PostHog users → more dashboards and insights consumed
• Platform/marketplace plays → multiplicative event growth (each participant generates events)
• Geographic expansion → new user cohorts → step-function increases in volume

Rate the business growth trajectory:

Step 6: Recommend a Play & Draft Outreach

Based on all five research workstreams, recommend one of these plays:

Play: Outreach (Slack Channel + Email)

Recommend when:

• Score >= 60 (High or Moderate confidence)
• 2+ active users in the account
• No one from PostHog is already actively engaged with this account

Draft an email AND a Slack channel welcome message. See Step 7 for drafting guidance.

Play: Light Touch (Email Only, No Slack Channel)

Recommend when:

• Score 40-59 (Low confidence) but the account has interesting signals worth exploring
• Single active user but spend is meaningful (>= $1K/month)
• The account already has a relationship with PostHog (e.g. engaged with onboarding team)

Draft a short email only. No Slack channel.

Play: Skip / Close as Not Qualified

Recommend when:

• Score 0-39 (Unlikely)
• Non-ICP industry with no offsetting signals
• Usage declining with no business growth to suggest recovery
• Account shows churn risk (past due billing, declining spend)
• Customer explicitly declined engagement (said "no thanks" or similar) - unless their spend is high enough that hearing about discounts from a TAE would be genuinely useful to them

When recommending Skip, you MUST provide a DQ Reason - a plain-text summary of why this lead is not qualified. The DQ reason must be 250 characters or fewer so the TAE can copy-paste it directly into the Salesforce disqualification field. Be specific - name the concrete signals, not generic language.

Important: Non-response is NOT a disqualification signal. If the onboarding team reached out and the customer simply didn't reply, that is not evidence they don't want to hear from PostHog - they may have been busy, missed the email, or the timing wasn't right. Do not cite "no engagement" or "didn't respond to onboarding" as a DQ reason. The only engagement-related DQ signal is an explicit opt-out - the customer actively responded to say they're not interested. Even then, if their spend is meaningful and a TAE could offer genuine value (e.g. volume discounts, billing optimization), an explicit opt-out from onboarding doesn't necessarily mean they'd reject a TAE with a different angle.

Good DQ reasons:

• "MRR declining $1,763->$696 forecast. Single product (analytics). 8 employees, 3 engineers. No growth signals despite $9.1M revenue. Group Analytics likely unnecessary."
• "Non-ICP: school bus routing company (founded 1977). Analytics-only, $1,369 flat MRR. 4.78 health. No expansion signals. Govt/education customer base."
• "Past due billing. MRR forecast drops 49% ($2,130->$1,086). 7-person seed startup. Likely cash-constrained despite Khosla backing."
• "Explicitly declined onboarding engagement. $380 MRR, single product, no growth trajectory. No value angle for TAE outreach."

Bad DQ reasons:

• "Not a good fit" (too vague)
• "Low score" (doesn't explain why)
• "Small company" (which signal matters?)
• "No engagement with onboarding" (non-response is not a signal - they may have been busy)
• "Unresponsive" (same problem - silence is not rejection)

Play: Wait and Watch

Recommend when:

• Early-stage company with high variance - could go either way
• Score 40-59 with strong business growth signals but weak current billing
• Check back in 4-6 weeks to reassess

Step 7: Draft the Outreach

When the recommended play is Outreach or Light Touch, draft the email using the guidance below. Read references/writing-style.md before drafting.

Email Format

All draft emails must use this blockquoted format for easy copy-paste:

> **To:** [full email address - e.g. [email protected]]
> **Subject:** [subject line]
>
> [email body]
>
> [TAE's name]

When the play is Outreach (Slack Channel + Email), also draft a Slack channel welcome message in the same blockquoted format after the email.

Outreach Principles

These are the same principles used across all PostHog lead skills. Follow them exactly.

1. Lead with the Slack channel. For multi-user accounts, the shared Slack channel is the most valuable thing you're offering - a direct line to someone technical at PostHog. Open with it. Explain why it's useful for their specific situation based on your research. Don't bury it.

2. Write like a fellow engineer, not a salesperson. These are competent developers. The TAEs are deeply technical. Write the way you'd message a colleague - direct, concise. No praise or cheerleading about their setup. "Your team's ramped up fast!" reads as patronizing to engineers who know exactly what they've done.

3. Start with an observation about them, not about you. The first real sentence should reference something specific from Vitally or your research - which products they're using, what their team is building, a specific usage pattern. Never open with "I'm [name], your PostHog contact."

4. Keep it short. The initial email should be 3-5 sentences. Slack channel + why it's useful + one specific observation or optimization + a question. Save the details for Slack.

5. One question per email. More than one question overwhelms and reduces response rate. Pick the most interesting one.

6. Be direct about billing when relevant. Unlike big fish outreach, onboarding handoff leads already have meaningful spend. It's OK to reference costs - pair every cost mention with an actionable way to reduce it.

7. Only reference what you can verify. Safe: Vitally usage data, public website, confirmed funding rounds. Never cite enrichment data (Clearbit/Harmonic tech stack fields) as fact. If you state something about their setup and it's wrong, you lose credibility.

Writing Style Rules

Follow references/writing-style.md for the full guide. Key rules:

• Get to the point. No "I'm reaching out because...", "I just wanted to...", "Saw you just signed up."
• American English. Oxford comma. Straight quotes.
• Hyphens only (-) for breaks in thought. Never em dashes or en dashes. Em/en dashes read as AI-generated.
• Links in anchor text. Never bare URLs. Never "click here." Always describe the destination: "here's a guide on connecting replay to funnels."
• Capitalize product names. "Session Replay", "Feature Flags", "Product Analytics."
• Abbreviate numbers. 10M, 100B (capital letter, no space).
• No emojis in emails.

Anti-Patterns (Never Do These)

• "I'm [name], your dedicated PostHog contact" - your name is in the signature.
• "Saw you just signed up" / "Just checking in" - they know. This adds nothing.
• "I'm here if you need anything" - too passive. Say something specific or don't say it.
• "Your team's ramped up fast!" / "That's impressive adoption" - patronizing. State facts.
• Feature laundry lists - reference the 1-2 products they're actually using, not all of PostHog.
• Multiple optimization suggestions in one email - save it for Slack.
• Bare URLs - always embed in anchor text.
• Citing enrichment data as fact - "I noticed you're using [competitor]" from an unverified SFDC field.
• Referencing exact event counts - "you sent 4.2M events last month" feels surveillance-y. Reference products and patterns, not precise numbers.
• More than one question per email.

Outreach Hook Guidance by Scenario

Multi-user account, growing spend: Lead with the Slack channel. Ground it in their usage: "Setting up a shared Slack channel for you - with [N] people using [products] across [N] projects, a direct line for technical questions will save your team time." End with a question about what they're building or trying to learn.

Multi-user account, flat or declining spend: Lead with a specific optimization. If you identified an issue (autocapture waste, stale flags, replay oversampling), that's the hook: "Noticed about [X]% of your events are autocapture without associated actions - [link to reducing event volume] covers how to trim that without losing anything useful." The Slack channel is where you help them act on it.

Single user, meaningful spend: Short and direct. No Slack channel. One specific optimization with a link. "Quick heads-up - at your current usage, [specific optimization + link]. Happy to help if questions come up."

Previously engaged with onboarding (had calls, completed onboarding): Reference the onboarding work. "Your team worked with the onboarding team on [what they covered] - wanted to make sure you have a direct line for questions now that you're past the initial setup." This creates continuity rather than feeling like a cold handoff.

Never engaged with onboarding (unresponsive, skipped): Don't reference the onboarding team at all - they clearly didn't want that engagement. Lead with something useful: a specific optimization, a resource relevant to what they're building, or the Slack channel as a low-friction support resource.

Example Outreach Drafts

Example 1: Multi-user, high-growth, engaged with onboarding

To: [email protected] Subject: Slack channel for Acme <> PostHog

Hey Acme team,

Setting up a shared Slack channel for you - with 57 people in the account across analytics, replay, and error tracking, a direct line for questions will be useful as you scale.

One thing that might be worth exploring: Feature Flags for safe rollouts. With your new product line launching as a platform, gradual rollouts and instant kill switches are the kind of thing that keeps deploys boring. Happy to walk through the setup in Slack.

What's the main thing your team is trying to learn from the analytics right now?

[TAE name]

Example 2: Multi-user, flat spend, optimization hook

To: [email protected] Subject: Slack channel for Example Payments <> PostHog

Hey Example Payments team,

Sending over a Slack channel invite - with 42 people using analytics, flags, and replay, it'll be the fastest way to get answers on implementation or billing as things come up.

One quick win: if your team is running A/B tests on checkout flows, Experiments layers directly on top of your existing Feature Flags setup. For a payments product, measuring conversion impact per variant is where the real signal is.

What's the team trying to optimize right now?

[TAE name]

Example 3: Single user, growing spend, never engaged

To: [email protected] Subject: Your PostHog setup

Hey,

Noticed you're using LLM Analytics alongside product analytics - that's a sharp setup for monitoring AI-generated content quality alongside user behavior.

If you're tracking prompt-to-output quality, this guide on LLM observability covers how to connect traces to user outcomes so you can see which prompts actually help students learn vs. just generate text.

Happy to help if questions come up.

[TAE name]

Output Format

Present the research brief as a structured report with clear sections. Use the Visualizer for the growth trajectory chart if the data supports it.

## [Company Name] - Onboarding Lead Research Brief

### Quick Links
- **Vitally:** [https://posthog.vitally-eu.io/customers/{organization_id}](https://posthog.vitally-eu.io/customers/{organization_id})
- **Salesforce:** [https://posthog.my.salesforce.com/{sfdc.Id}](https://posthog.my.salesforce.com/{sfdc.Id})
- **Usage Dashboard:** [{usage_dashboard_link}]({usage_dashboard_link}) *(or "Not available" if trait is blank)*

### Company Overview
- **What they do:** [one sentence]
- **Industry:** [vertical]
- **Size:** [employees] ([engineering count] engineers)
- **Funding:** [stage], [amount] ([date])
- **HQ:** [location]
- **Growth trajectory:** [Accelerating / Steady / Flat / Early stage]

### $2K+ Spend Likelihood
**Score: [X]/100 - [High / Moderate / Low / Unlikely]**

Key factors:
- [Factor]: [+/- impact]
- [Factor]: [+/- impact]
- [Factor]: [+/- impact]

Projection: Currently at $[X]/month - projected $[X]/month in 3 months

### Vitally Usage & Billing
- **Current MRR:** $[X]
- **Products in use:** [list]
- **Products NOT in use:** [list - these are expansion opportunities]
- **Event volume:** [current] ([trend])
- **Active users:** [count] ([trend])
- **Health score:** [X]/10

**Billing trajectory:**
[Chart or description of the trend - is spend growing, flat, or declining?]

**Expansion opportunities:**
- [Product they're not using that fits their use case]
- [Volume tier they're approaching]
- [Config optimization that could increase value]

### Onboarding Team Context
**Engagement level:** [Engaged / Partially engaged / Unengaged]
**Onboarding rep:** [Name]

**What happened:**
[Narrative summary of the onboarding engagement - what was discussed, what was done, what's left open]

**Key quotes or insights from the customer:**
- [Any notable things the customer said or asked about]

**Open items / unfinished business:**
- [Anything the onboarding team started but didn't finish]
- [Any promises made to the customer]

### Business Growth Trajectory
**Rating:** [Strong accelerating / Steady upward / Flat/stable / Contracting]

**Key growth signals:**
- **Revenue:** [what's known about ARR growth, Inc. 5000 ranking, etc.]
- **Funding:** [recent round, amount, investors, what it's earmarked for]
- **Product velocity:** [new launches, features, platform plays in last 12 months]
- **Customer expansion:** [new segments, partnerships, geographic expansion]
- **Headcount:** [total delta, eng delta, GTM hiring signals]
- **Web/brand momentum:** [traffic trends, LinkedIn growth, awards/recognition]

**How this maps to PostHog spend:**
[Explicit analysis connecting each business growth vector to expected PostHog usage growth. e.g. "More customers on their platform - more end-user sessions - event volume will grow faster than current billing suggests." This is the most actionable part - it tells the TAE whether the Vitally billing trajectory understates or accurately reflects the real growth potential.]

**Score adjustment:** [If the business growth signals are strong enough to warrant adjusting the $2K+ likelihood score from Step 2, note the adjustment and reasoning here. e.g. "Business growth trajectory bumps the score from 72 to ~80 - billing data lags customer growth."]

### Recommended Play
**[Outreach / Light Touch / Skip / Wait and Watch]**

[Reasoning for the recommendation - 1-2 sentences.]

[If Skip: include DQ Reason below]

**DQ Reason (for Salesforce):**
`[250 characters or fewer - specific, concrete, copy-pasteable]`

### Draft Outreach

[If the play is Outreach or Light Touch, include the draft email in blockquoted format. If Outreach with Slack channel, also include the Slack welcome message.]

> **To:** [full email address]
> **Subject:** [subject line]
>
> [email body]
>
> [TAE name]

[If multi-user Outreach, also include:]

**Slack channel welcome message:**

> [Slack message body]
>
> [TAE name]

**Suggested angle:** [One sentence explaining why this hook was chosen and what it's designed to accomplish.]

When Processing Multiple Leads

After completing all individual research briefs, output a Summary & Priority Stack Rank table at the top of the combined output (or as a final summary section). Rank accounts by score descending.

## Summary & Priority Stack Rank

| Rank | Account | Score | Tier | MRR | Trend | Action |
|------|---------|-------|------|-----|-------|--------|
| 1 | [Company] | [X]/100 | High confidence | $[X] | Growing | Outreach |
| 2 | [Company] | [X]/100 | Moderate confidence | $[X] | Flat | Light Touch |
| 3 | [Company] | [X]/100 | Unlikely | $[X] | Declining | Skip |

Column guidance:

• Score: The final $2K+ likelihood score (0-100) with the adjusted figure if business trajectory warranted an adjustment
• Tier: High confidence (80-100) / Moderate confidence (60-79) / Low confidence (40-59) / Unlikely (0-39)
• MRR: Current MRR from Vitally billing data
• Trend: Growing / Flat / Declining - based on billing trajectory, not just business signals
• Action: Outreach / Light Touch / Skip / Wait and Watch

Reference Files

• references/scoring-model.md - Full $2K+ spend likelihood scoring model with weights and criteria
• references/writing-style.md - PostHog writing style guidelines for email drafting

Critical Reminders

1. Research first, then draft. Complete all five research workstreams before recommending a play or drafting outreach. The research informs the outreach hook - without it, the email will be generic.
2. Vitally data is the source of truth for usage and billing. Don't guess at usage - pull it.
3. Read every onboarding note. The onboarding team's context is gold. Skipping it means the TAE walks into the conversation blind.
4. Project the trajectory. The TAE doesn't just need to know where the account is - they need to know where it's going.
5. Flag discrepancies. If the web research and Vitally data tell different stories (e.g. company seems to be growing but PostHog usage is flat), call that out explicitly. This is especially important in Step 5 - if the business is accelerating but PostHog billing is flat, the billing is lagging and there's upside.
6. If Vitally data is thin, say so. Don't invent data. If there are only 2 notes and no conversations, report that - it's useful context in itself (means the customer may not have engaged much).
7. Business growth is the leading indicator; billing is the lagging one. The Step 5 business growth trajectory analysis is what differentiates a good research brief from just reading Vitally numbers. A company that just raised $43M and doubled ARR is going to grow their PostHog spend - even if the current billing doesn't show it yet. Conversely, a company with declining headcount and no recent funding may be at peak spend already. Always connect the business trajectory to the expected PostHog spend trajectory.
8. Use Harmonic data from Vitally traits. The SFDC/Harmonic fields in the account traits contain headcount deltas, web traffic trends, LinkedIn growth, and funding details. Cross-reference these with your web research - they often tell a richer story than the web alone.
9. Industry context matters for the growth mapping. How a company grows affects how PostHog usage grows. A platform/marketplace company gets multiplicative event growth per customer. A SaaS company gets linear growth per customer. An infrastructure company may generate heavy backend events but few replays. Map the business model to the right PostHog products.
10. DQ reasons must be copy-pasteable. When recommending Skip, the DQ reason goes directly into Salesforce. Keep it under 250 characters, make it specific, and include the key data points that justify the disqualification.
11. Draft outreach follows the writing style guide. Read references/writing-style.md. Hyphens only (never em/en dashes). No bare URLs. No feature laundry lists. One question per email. Lead with the Slack channel for multi-user accounts.
12. Use full email addresses. When listing contacts or drafting outreach, always use the full email address (e.g. [email protected]) for easy copy-paste.
13. Validate all URLs before presenting the draft. Fetch every link in the email to confirm it resolves. If broken, search for the correct page or remove the link.