# Larry Kim

Larry Kim works on CustomersAI, which helps Shopify and Klaviyo brands identify website visitors and use identity data and AI to improve email audience targeting, deliverability, and retention marketing.

- Profile: https://b2binfluencer.marketing/creators/larry-kim/
- LinkedIn: https://www.linkedin.com/in/larrykim
- LinkedIn followers: 94,736
- Topics: artificial-intelligence, ecommerce, content-marketing, ai-agents, ecommerce-marketing, retention-marketing
- Based on 19 dated posts, latest 2026-10-09

## Summary

Larry Kim is most useful for email deliverability diagnostics and identity data advice for Shopify and Klaviyo retention marketing.

- Best for: A B2B SaaS email deliverability launch or advisory and co-marketing engagement.
- Not for: Consumer short-form video campaigns on Instagram or TikTok with broad shopper reach.
- Check first: Confirm audience demographics and recent results for comparable deliverability or AI-retention content.

### Takeaways

- A Klaviyo score of 70% rated Good while 41.3% of test emails landed in spam.
- All Gmail and Google Workspace test emails landed in spam in that test.
- CustomersAI found visitor-identification data was wrong most of the time, 5% to 30% accurate.

## Main ideas

- **Identity data quality is the dividing line in retention marketing**: Kim argues that most third-party visitor-identification data is not accurate enough to use for email, so CustomersAI stopped reselling it and built its own extended first-party identity layer. They base that argument on a head-to-head test: take a known first-party session, such as someone who actually made a purchase, and compare it with the identity the provider claims for that visitor. Kim says many providers were right only 5% to 30% of the time, meaning most of the emails they surface are wrong. Because AI learns from that signal, they argue, a wrong identity cannot be fixed downstream by a better model. Their conclusion: the retention platform with the best identity signal wins.
- **Email deliverability scores can hide the real problem**: Kim argues that an ESP deliverability score is a proxy built from downstream engagement signals such as opens, clicks, bounces, complaints and unsubscribes, so it does not answer the most important question: did the email reach the inbox? When mail lands in spam, people never see it, so they never open, click, complain or unsubscribe. That can make bad placement look quieter than it really is while positive engagement is inflated. They say Apple Mail Privacy Protection, bots, proxies and security scanners heavily inflate opens; clicks are cleaner but still imperfect.
- **AI makes the coordination layer inside companies unnecessary**: Kim argues that most product manager work involves carrying context between customer success, sales, engineering and design, and that AI has made that translation layer obsolete. Once requirements are expressed as a single source of truth through event modeling—a standardized way of stating inputs and outputs—AI can generate the interface, prototype, code stubs, documentation, training material and even marketing content without handoffs. Kim says CustomersAI removed the product management function entirely, with the engineering team leading the change after years of poorly defined requirements documents.
- **Using AI now means building agents, not writing prompts**: Kim argues that prompting alone is no longer enough and that team members who still use AI only through basic prompting will be left behind. They want agent systems that reason, access tools, learn skills, retrieve information, make predictions, take actions and complete work autonomously. Kim frames this as a management responsibility, not a tooling preference, and extends it beyond technical teams.
- **Run marketing as an output engine measured on revenue**: Across three hiring posts, Kim sets out the marketing operation they want: content shipped daily across channels, newsletters twice a week, monthly webinars and co-marketing, consistently published case studies, and one idea turned into many pieces of content, with AI used to move faster. They state the priorities directly: output over perfection, campaign thinking rather than asset production, and pipeline and revenue impact rather than vanity metrics. The role they describe owns a content machine end to end, not just a few blog posts.
- **Meet buyers inside the retention-tool ecosystem**: Kim's posts treat the Klaviyo and Shopify partner ecosystem and its events as a distribution channel, not a side activity. They show up where brands, agencies and fellow vendors gather, discuss identity data and deliverability in person, and use co-marketing with platform partners to reach buyers already working inside those tools.

## Work to read

- [A big update on CustomersAI](https://www.linkedin.com/posts/larrykim_a-big-update-on-customersai-four-years-activity-7473389928127725568-9fEI)
- [The 'Good' Deliverability Score Hid a 100% Gmail Spam Problem](https://www.linkedin.com/posts/larrykim_the-good-deliverability-score-hid-a-100-activity-7473031744640692224-54Xq)
- [We eliminated our product management function at CustomersAI](https://www.linkedin.com/posts/larrykim_eventmodeling-activity-7448013527442747392-DcPU)
- [I challenged all my department teams to an internal AI competition](https://www.linkedin.com/posts/larrykim_i-challenged-all-my-department-teams-to-an-activity-7478127329400094720-ih-V)

## Brand relationships

- **CustomersAI** (owned_business)
- **Klaviyo** (disclosed_partnership)
- **Shopify** (disclosed_partnership)
- **Meta** (disclosed_partnership)
- **Google** (disclosed_partnership)
- **Triple Whale, PostPilot and Okendo** (disclosed_partnership)

## Partnerships

The posts do not disclose a public sponsorship policy, rate card, or partnership inquiry route. The partnerships are company-level: CustomersAI posts state it is an official Klaviyo, Shopify, Meta, and Google partner, and the company has co-marketed with Triple Whale, PostPilot, and Okendo on a virtual DTC event. CustomersAI's team also promoted its presence at Klaviyo's K:BOS event. Treat these as documented ecosystem relationships and co-marketing examples, not a published sponsorship menu. Ask CustomersAI directly for sponsorship terms and an inquiry route.

## Alternatives

- Abraham Buchinger: Choose Abraham Buchinger when the deliverability launch is framed as a DTC ecommerce growth decision rather than an inbox-placement test. The two posts argue that better tools do not raise sales without a decision owner; that segmentation (new versus repeat, high-intent versus browsing, price-sensitive versus convenience-driven) fails when everyone gets the same message; and that decisions to pull back on messaging should be driven by performance and deliverability data, not fear of being too aggressive. Buchinger also writes about where buying decisions start, including Google expanding checkout inside Gemini and the platform risk for brands that sell only on Amazon. For a launch aimed at merchants weighing channel and customer-data choices, that framing puts the product inside decisions the buyer already owns and reaches an operator audience through advice rather than a vendor's own test result.
- Corto S.: Choose Corto S. when the launch should show the product working inside Klaviyo through an AI workflow, rather than explain why inbox placement is hard to see. The two posts walk through turning Claude Code and a GPT model into a Klaviyo execution engine: analysing account performance, finding gaps, prioritising opportunities, running separate workflows for flows, campaigns, segmentation, deliverability, and analytics, then planning a monthly campaign calendar, building lifecycle flows, and drafting copy. Each post packages the setup as a guide, forming a co-marketing offer: a documented build demo with a downloadable resource and the creator as the operator who assembled it. For a buyer who values demonstrated tool use over diagnostic evidence, this is the closer match.
- Meilech Biller: Choose Meilech Biller when the launch buyer is an ecommerce agency or operator who wants to see how the tool gets installed and run, rather than a vendor's published test. One post describes building the agency's own systems with Claude: CRM and project management, a client onboarding flow of custom forms and documents, and automatic daily review of CRM, project, messaging, and email activity to decide the next action per lead. It also mentions a deliverability dashboard built for Klaviyo users. The other post describes a completed client build—email capture forms, welcome flow, browse and cart abandonment, and campaigns—held back until seasonal stock and site updates were ready. Together, the posts show a partner's hands-on deployment sequence, often the missing piece when a tool is sold to agencies on behalf of clients.

---
AI-assisted summary of public posts. Audience demographics are not available. Engagement appears only where we checked recent posts. Method: https://b2binfluencer.marketing/methodology/