AI for eCommerce

RJ Talyor is the Founder and CEO of Backstroke a AI for eCommerce generative content platform for email marketers. Instantly create on-brand and performant email subject lines, preview text, mobile push notifications, and SMS messages.

Summary of Podcast

Podcast introduction and guest background

Graham and Kevin introduce the Next 100 Days Podcast and welcome RJ Talyor from Indianapolis. RJ describes Indianapolis as offering the best of a big city with a small-city feel, with about a million people, great sports, culture, food, and good cost of living. He has traveled extensively but always enjoys returning home.

Backstroke’s AI email generation platform

RJ introduces Backstroke.com, which generates performant email campaigns for e-commerce retailers selling clothes, pet food, furniture, and other products online and in-store. E-commerce brands typically expect 20-50% of revenue from email marketing while sending 3-5+ emails weekly, with customers spending 8-12 hours per campaign. Backstroke reduces this to approximately 15 minutes while personalising content so each customer receives a different message tailored to their interests and behaviour.

Personalisation through data and engagement 

Backstroke personalises emails using multiple data layers: subscriber status, past engagement (opens, clicks, conversions), and appended third-party data revealing demographics like age, location, and gender. When additional data is unavailable, the platform uses progressive profiling—analysing engagement patterns to infer preferences. For example, if a customer consistently clicks on men’s content over women’s content, or prefers dark-coloured shirts over light ones, AI identifies these patterns to drive personalisation, which is more effective than manual analysis.

Real-world personalisation: from negative to advocate

Graham shares a personal story about Son of a Tailor, a Portuguese apparel brand, where his initial experience was poor—they sent him a shirt too short for his frame. However, the company responded exceptionally well, ultimately creating a monogrammed, high-quality shirt that transformed him into an advocate. RJ explains this is valuable data: AI can flag customers who experienced negative-to-positive journeys as potential super-fans or loyalty advocates, a pattern most marketers miss because they lack time to identify such nuanced customer experiences.

AI pattern recognition beyond traditional metrics

Traditional RFM (Recency, Frequency, Monetary) models reduce customers to transactional data, but AI can extract signal from unstructured data to identify complex patterns. For instance, AI can recognize when a customer buys different sizes (suggesting purchases for others) or when multiple preferences exist within one account—like RJ’s Spotify feed where his children’s music preferences mix with his own. AI discerns these overlapping patterns that aren’t immediately obvious to humans, enabling more sophisticated segmentation.

Team expertise and company history

RJ co-founded Backstroke with his wife Allison, who holds a PhD in deep data analysis and chemical reagents, bringing statistical rigour and predictive modelling expertise. RJ’s background includes starting Pattern89 in 2016, an AI company predicting Instagram and Facebook clicks using computer vision and natural language processing, which he sold to Shutterstock. Many Pattern89 team members joined Backstroke, bringing 10 years of AI-based marketing experience, while the team continuously innovates with new foundational models from Anthropic and OpenAI.

Implementation results and Surge feature

Backstroke achieves an average 30% uplift in conversion rates for new clients. Implementation typically takes about a month for full transformation, but recognising customer demand for faster results, the company launched “Surge,” enabling campaigns to launch in 48 hours. This rapid-deployment feature demonstrates predictive capabilities quickly, satisfying customers who want immediate proof before committing to full onboarding.

Email variants and human approval at scale

While technically capable of generating 10,000+ unique email variants, Backstroke has found that customers require human review of every variant version. Current implementations range from 60-100 variants, with combinations of hero images, subject lines, and templates creating exponential possibilities. The company is building QA agents to enable scaling to millions of variants while maintaining human oversight, recognizing that creative teams ultimately bear responsibility for brand representation.

Brand guidelines versus performance metrics

A fundamental tension exists between brand teams (who enforce guidelines like “models must face forward” or “only use this colour”) and performance marketers (who know “shirts perform better laid on a bed than on a human”). RJ explains this is often gut-feel decision-making based on outdated tests—teams cite tests from a year ago by employees who’ve since left, creating stale guidelines. AI enables rapid testing of creative variations to identify incremental opportunities, but requires organisational willingness to experiment beyond established brand rules.

Customer selection philosophy

Rather than trying to convince resistant customers to embrace AI, RJ focuses on the “one in 10” truly innovative marketers willing to change. He learned from his previous business that most prospects claim interest but quickly reveal organizational barriers requiring approvals. His strategy is to identify customers genuinely committed to transformation and willing to pay, directing others to resources instead. This approach conserves energy for high-potential partnerships where AI can deliver real impact.

Backstroke’s core value proposition

Backstroke solves the “what” problem: what content, subject line, preview, template, hero image, product display, and offer to send to each person. The platform knows that 46% of clicks occur in the first 400 pixels, so it optimizes that space differently for men versus women, loyal customers versus new ones, and geographic regions. This focused specialization on content optimization is Backstroke’s primary value, distinct from solving “when” (send time) or “who” (segmentation) problems.

Practical tips for email marketers

For marketers using standard LLMs without specialised platforms, RJ recommends uploading all previous email data and creative assets, then asking the machine to identify winning creative dimensions. This approach reveals patterns in subject lines, imagery, copy length, and offers without requiring subscriber-level analysis, enabling better-than-average results for those without access to specialised tools.

Email frequency paradox and engagement

Kevin raises frustration with receiving excessive emails from companies he likes, asking if AI can enable sending less email while achieving better results. RJ explains that higher engagement with personalised content could theoretically reduce frequency, but email is fundamentally a frequency game—brands send multiple emails weekly to stay top-of-inbox when customers are ready to buy. However, deliverability depends on engagement (opens, clicks), so sending irrelevant content backfires. Backstroke solves the “what” problem, but send-time optimisation and segmentation (the “when” and “who”) remain separate challenges.

Market focus and customer examples 

Backstroke focuses exclusively on B2C e-commerce in North America due to language complexity and GDPR privacy requirements in Europe. The platform serves impulse-purchase categories (apparel, furniture, bedding) differently than considered purchases (mattresses, cars), with separate trained models for each. Notable customers include Third Love (women’s intimates), Cozy Earth (bedding), Helix (mattresses), and Emile Henry (cookware), representing the apparel and home goods verticals where Backstroke has developed deep expertise.

Future roadmap: predictive marketing agents

RJ’s 18-month roadmap focuses on building predictive marketing agents that complete marketing tasks generatively while humans serve as brand stewards and strategists. This vision extends beyond email to SMS, apps, and landing pages, with personalisation as a core feature. Graham notes the challenge of making such systems intuitive enough for non-technical users, reflecting the broader industry shift toward AI-augmented rather than AI-replaced marketing roles.

European expansion and compliance strategy

While Backstroke is currently North America-focused, RJ is open to European partnerships but wants to be proactive about compliance. GDPR itself isn’t a blocker, but European customers require security documentation and certifications that Backstroke hasn’t yet obtained. The company recently achieved SOC 2 compliance (required by enterprise businesses) and plans to secure necessary privacy certifications before entering European markets, avoiding disqualification during sales cycles.

Podcast analysis and key takeaways

In the wrap-up, RJ praises the podcast for getting past fluff into real marketing challenges, appreciating the nitty-gritty discussion of how marketers actually work. Graham and Kevin reflect that the conversation revealed AI’s potential to solve the “what” problem while highlighting remaining challenges in “when” and “who” decisions. They note that Kevin’s observation about sending less email more effectively captures the core tension: personalisation should enable quality over quantity, but frequency remains a competitive necessity in current email marketing practice.

Clips from the Podcast

What’s So Special About Indianapolis?

 

What Does Backstroke Do?

 

Personalising Using AI & Engagement Data

 

Son of a Tailor

 

Surge & Conversion Uplift

 

Sponsor – inCruises

Graham Arrowsmith’s company Finely Fettled is an inCruises Independent Partner and invites you to find out more about inCruises by clicking on either the logo or the Why Cruise with InCruises? image.

Testimonial

 

The Next 100 Days Podcast Co-Hosts

Graham Arrowsmith

Graham founded Finely Fettled in 2014 to provide data from The UK High Net Worth Database to marketers targeting affluent and high-net-worth customers.  He’s the founder of MicroYES, a Partner for MeclabsAI, creating lead generation AI Agents & Workflows and introducing the MeclabsAI Platform. Graham also provides an Answer Engine Optimisation solution to get your website in shape to be found by LLMs.

Kevin Appleby

Kevin specialises in finance transformation and implementing business change. He’s the COO of GrowCFO, which provides both community and CPD-accredited training designed to grow the next generation of finance leaders. You can find Kevin on LinkedIn and at kevinappleby.com