LeadGrow × HeyDigital

Campaign Update

Jun 8–15, 2026
22k
Leads
6
Campaigns
3.3k
Sent
76
Replies
2.3%
Reply Rate

Live Campaigns

CampaignLeadsSentRepliesStatus
Product (913) 3,758 2,123 55 2.6% reply
Founder (911) 7,419 733 15 2.0% reply
PostHog Signal (929) 767 39 2 Just launched
BrandHijack (905) 168 349 4 68% complete
Marketing (912) 5,113 6 0 Ramping
Revenue (914) 1,737 9 0 Ramping

What Shipped This Week

New Campaign

PostHog Signal Campaign (929)

749 leads using PostHog analytics or hiring for PostHog roles. Every email includes 3 AI-generated campaign ideas specific to the prospect's company.

Signal: TechSight detection + LinkedIn hiring data

Companies: 714 PostHog-using companies identified

Ideas: 713/714 generated, 4.63/5 avg quality score

Sequence: 3 E1 A/B/C variants + E2 thought leadership angle

QA: 79 company names corrected, all emails MV-validated, 7 security gateway domains removed

Sender: Ajay (15 accounts)

Completed

Wave 2 Segments Live

Four buyer-role segments launched alongside Founder. Each framed to that audience's worldview, same 4-variant split test.

55
Product Replies (2.6%)
15
Founder Replies (2.0%)
Completed

Data Quality Pass

  • 79 company names corrected in Supabase (slugs, wrong casing, trailing suffixes, YC tags)
  • 108 emails Million Verifier validated (105 blitz + 3 NULL)
  • 7 security gateway domains removed (Proofpoint, Mimecast, Barracuda)
  • 1 invalid email caught and removed (siddhant@alltius.ai)
New Campaign

Lead Magnet Campaigns

Two new lead magnet offers launched as dedicated sequences within existing campaigns:

  • Thought Leadership Ads - "We built 800k in pipeline from thought leadership ads alone." DFY offer to map out a founder content strategy.
  • Pipeline Lead Magnet - Direct pipeline-building offer for marketing and revenue leaders.

Testing which lead magnet generates highest-quality replies to inform future campaign CTAs.

In Progress

AI Agent Contact Discovery

Running Serper-powered AI agents to find missing contact data on newer SaaS companies. No existing database (AI Ark, Blitz, Apollo) has employee coverage for these companies.

+30%
LinkedIn Coverage Gained
TBD
Email Coverage

LinkedIn profiles found first, email waterfall runs after. Filling a gap no data provider covers - early-stage SaaS with <50 employees that haven't been indexed yet.


PostHog Ideas Engine Deep Dive

Each prospect gets 3 personalized campaign ideas generated by AI, specific to their company, industry, and competitors.

Real Example: Actively AI (alissa@actively.ai)

Contact: Alissa Lydon, VP of Marketing • Company: AI-powered revenue ops platform

Real Example: Antithesis (randall.jackson@antithesis.com)

Contact: Randall Jackson, CRO • Company: Autonomous testing platform

Pipeline Numbers

714
Companies with Ideas
2,142
Unique Ideas Generated

Quality eval: 4.63/5 avg across 15-sample LLM judge. 15/15 pass.

Email Verification

  • 767 leads exported (after gateway filter)
  • 100% Million Verifier validated
  • 0 catch-all in primary sequence
  • 749 addressable in Bison

Copy Updates

Updated Today

PostHog Campaign Copy (929)

All 4 sequence steps pushed to Bison with final spintax. Changes from last draft:

  • Killed "quick one/short note" opener on Variant C - salesy trigger phrase removed
  • E2 completely rewritten - was "bumping this" bump, now 800k thought leadership pipeline angle (genuinely different offer)
  • E2 threading fixed - subject was "Re: Re:", now threads properly under whichever E1 variant the lead received
  • All copy uses spintax - thousands of unique combinations per send

CTA differentiation across variants:

EmailCTAOffer
E1-APostHog breakdownSend case study mechanics
E1-BFunnel teardownFree audit, no strings
E1-CCuriosity pullWalk through the playbook
E2Thought leadership strategy800k pipeline proof, DFY on the house
Live - Winning

Product Campaign Copy (913) - 2.6% Reply Rate

Best-performing copy across all segments. PLG-specific framing.

55 replies from 2,123 sends. Key angle: "most agencies run your PLG funnel on an ecommerce template" - calls out the generic approach competitors use.

Live

Competitor Name-Drop

{COMPETITOR_1} personalization variable in templates.

Coverage: 23,588 / 25,560 (92.3%). The 7.7% gap auto-fills with "your top competitor" at lead level.

Standards

Copy Rules

  • 6 subject variants per step - lowercase, 2-4 words, internal-email feel
  • Hybrid spintax: word-level swaps + phrase-level for clauses/CTAs
  • No em dashes. No "free" - replaced with {on the house|on us|fully complimentary}
  • All emails MV-validated. Security gateways (Proofpoint/Mimecast/Barracuda) removed pre-send

Proof Points Validated & Sourced

$44K
Toggl CPL (from $93K)
+316%
Checkly MQLs
50%
Rosie Trial Rate (from 30%)
+480%
Checkly PQLs
36.1%
Avg Waste (43 audits)
200+
B2B SaaS Served

Blockers & Next Steps

PostHog campaign Launched Jun 15. 749 leads, 3 A/B/C variants + thought leadership E2. All MV-validated, security gateways removed, company names QA'd. Already sending.
Copy refresh for 912/914 Marketing (912) and Revenue (914) are ramping slowly (6 and 9 sends). Product (913) is outperforming at 2.6% - analyzing what's working there to apply to the other segments.
BrandHijack Wave 2 Wave 1 at 68% complete (349 sends, 4 replies). ~400+ domains from the full brand hijack audit still need contacts and campaign setup for Wave 2.
OUTLOOK-PARK campaigns (917-920) 4 shadow campaigns (Founder/Marketing/Product/Revenue on Outlook domains) staged but still in draft. 8,394 leads loaded. Awaiting domain warmup completion.

TEST 1 3-ideas variant build, idea-clean pass, ICP prune · campaigns 911-914 · Jun 16

✓ Worked
3-ideas A/B variant built on all 4 campaignsNew order-2 variant (steps 4120/4121/4122/4123) added to 911/912/913/914 via the Bison API create-step — it DOES create real A/B variants (the CLI's "can't create variants via API" note was stale). All four shipped active:false and have never sent. Copy locked, subject spintax validated.
Idea-clean LLM pass — 23,468 rows, 0.2% fallbackRaw internal-note ideas rewritten to prospect-facing idea_1/2/3_clean across 23,468 companies. Only ~0.2% fell back to the approved generic trio — the rest carry company-specific angles.
first_name 422 sync fixcreate-or-update/multiple validates first_name even in patch mode. Fix: echo each lead's existing name back in the patch payload. IDEA_1/2/3 + COMPANY_TYPE now sync cleanly onto leads.
Non-ICP prune complete — 0 off-ICP leads remainKeep-set = the 23,390 ICP-qualified companies (icp_qualified = true), not the full 30,891-row table. 5,871 leads whose domain was absent or flagged icp_qualified=false were removed across 911-914 (911: 2,546 · 912: 1,680 · 913: 1,070 · 914: 575). Re-query confirms 0 off-ICP leads remain. Every removal logged to removed-leads.csv; full per-lead baseline saved for rollback.
✗ Didn't work / had to fix
37% of loaded leads were off-ICPBefore pruning, ~37% of 911-914 leads had a website domain absent from heydigital_companies — they were getting generic fallback ideas instead of real personalization, diluting the variant. Root cause: lists carried leads outside the qualified ICP set. Now removed.
Bison leads API caps at 4,500 per pullThe campaign-leads endpoint stops at ~4,500 rows, so 911 (largest) needed several remove+re-pull passes to surface and clear the tail. Converged to 0 non-ICP.
update-step can't create stepsIt is edit-only — 422s without an id. Variant creation must go through create-step + a renumber round-trip + toggle-variant (which requires the campaign paused).
— Inconclusive
Send performance: no data yetThe 3-ideas variants have never been sent (active:false, campaigns paused for this test). Open / reply / positive-reply rates vs the base step are unknown until the variants are activated. No verdict on the 3-ideas copy yet.
Spam / spintax audit: clean structure, mild base tokensSpintax parses valid and the new variants are clean. The pre-existing base steps carry mild spam-adjacent tokens ("ad", "for you") — flagged, not blocking. Real-world deliverability unmeasured until a live send.