Ship.com Google Ads — Live Dashboard

1st ever paying (Jun 8), churned Jul 8 (~1 month). ⚠️ DATA INTEGRITY: his HubSpot contact record is no longer findable (deleted or merged after churn). Churned-contact purges/merges shrink historical paying counts and break CPA traceability over time — confirm with whoever admins HubSpot whether this is manual, a merge, or an automated process.

Paying Jul 3, downgraded to Starter Subscriber tier after 1 day. 4 labels.

Paying Jul 7, still active. 3 labels. CORRECTION: earlier mis-attributed to "cheap shipping software" — actual keyword per utm_content is "shipping software reviews".

Paying Jul 14, active. 6 labels. GCLID-attributed (no utm_campaign — only caught by the union filter).

Paying Jul 25, active. 1 label. First paying of the post-LP era (dedicated LP went live Jul 20) — encouraging but n=1, not yet causal evidence for the LP.

The individual-intent keyword we kept wanting to move away from is the biggest paying producer — but with shaky retention (1 churn, 1 downgrade, 1 too-new). The bottom-funnel keyword ("shipping software reviews") produced only 1, but it's the cleanest retained. Volume-of-paying vs quality-of-paying tension, with names attached.

The headline experiment. Swapped the final URL from generic /sign-up to a purpose-built, intent-matched landing page (CTAs forward gclid+utm to /sign-up — tracking verified live). Measuring over ~2 weeks whether it lifts trial→paid above the ~1.6% baseline. This is the one un-fired ads-side lever on the $100-CPA problem.

Best-CPA keyword (~$629, produced Melissa — active). Bid raised +15% Jul 17 to unstarve it; still 84% rank-lost. Watching for a 2nd paying to validate the comparison intent.

Manual CPC $3.50. Addon discovers Mon, LLM-approves Tue. "ebay label printing" cut Jul 20 (fair-test failed).

The assets serving were generic account-level ones (Customer Invoicing, Email Marketing…). Added intent-relevant sitelinks + callouts + a structured snippet per campaign to lift expected CTR and Ad Rank. Efficiency lever, not a trial→paid lever.

Paused the 2 non-converting Comparison keywords; MVP budget $26→$34; Comparison $38→$27; bid on "shipping software reviews" +15%→$4.03; routed 3 keywords to intent-matched pages. Cut waste, concentrated on what works.

~80% of signups lack GCLID, forcing email-hash matching. Backend team owns. Improves attribution + Smart Bidding signal.

Replace placeholder values ($25/$50/$150) with real MRR to compute an honest CPA target.

4 lifecycle conversion actions created. Decided NOT to reactivate PMax (historical CAC $3,780/paid) or bid branded (3,500+ organic clicks/90d). 16 brand negatives standard.

$1,200 / 91 trials / 8 activations / 1 confirmed paying. Finding: individual-intent keywords drive trial volume but trial→paid only 1.1%. Trial→activation 8.8% (low).

Finding: Smart Bidding starves new keywords with no history (cold-start trap). 5 of 6 new keywords got 0 impressions. Led to Manual CPC approach.

Findings: Manual CPC fixed cold-start. Shopify (vertical) and Software (generic) killed — genuine negative signal on Shopify, off-intent traffic on Software. Comparison (bottom-funnel) is the best signal after "print shipping label" itself. Expanded into Competitor Alternatives + Keyword Lab on Jul 6.

Competitor Alternatives killed: low search volume + rank-lost 62% + 0/15 click→trial, NOT a budget problem (spent $6 of $14/day). Bare brand terms ("shipstation" 60k/mo @ up to $33 CPC, "stamps com" @ up to $137) are navigational — even worse economics. Conclusion: the cheap seller-intent keyword space is a rare gap (Comparison's 1 paying came from "shipping software reviews" — a one-off, not a replicable pattern); competitor terms and expensive software terms don't close to $100 CPA. The lever is churn/product, not more keywords.

Checked JotForm and Pink Zebra integrations: their specific shipping terms ("jotform shipping", "pink zebra shipping labels") have ~0 search volume — nobody searches "[brand] shipping integration" because they don't know it exists yet. Rule: paid search HARVESTS existing demand; it can't CREATE awareness of a brand-new niche integration. For new-integration launches the channel is push (email / in-app to the user base that can already use it), not pull (search). Cheap bounded tests parked in the Keyword Lab (jotform shipstation integration, pink zebra consultant) but low expectations.

Blended CPA ~$753 (÷4 acquired) / ~$1,004 (÷3 traceable) — 7-10x off goal and drifting up. Verdict: the binding constraint is trial→paid (~1.6%) — product/activation, not keywords or bids (every intent tested leaks at the same step). Response: Option A consolidation + fire the one un-fired ads lever, a dedicated intent-matched LP. Shifted from "expand keywords" to "hold + measure the LP".

Dedicated /print-shipping-label LP live (built + tracking-verified). Intent-relevant ad assets added to MVP + Comparison for Ad Rank/CTR. Keyword ceiling reached (no new cheap seller-intent seam), so the bet is the LP: does an intent-matched page move trial→paid? 2-week read; if flat, constraint is confirmed product and ads scale to floor.

HubSpot lifecycle transitions → Google Ads (GCLID + email-hash)

08:00 Santiago — metrics, changes, keyword candidates

Search terms >$15 + 0 conv → LLM eval → add. Cap 3/day.

Keywords >$50 + 0 conv → LLM eval → pause. Cap 1/day.

5-step pipeline → candidates surfaced in digest

Green-tier candidates → push to live as PHRASE. Cap 3/week.