# Shuttlebase — Full Reference for Language Models Last updated: 2026-08-19 Canonical site: https://shuttlebase.com/ Documentation: https://docs.shuttlebase.com/ --- ## What is Shuttlebase? Shuttlebase is an autonomous conversion rate optimization (CRO) platform for eCommerce brands. It analyzes a store's site and traffic to find where revenue is leaking, generates real design, copy, and code variants to fix it, A/B tests those variants on live traffic, and deploys the winners automatically. The distinction that matters: most CRO and A/B testing tools give you a testing surface and a list of recommendations, then leave you to staff a designer, a developer, and an analyst to act on them. Shuttlebase runs the entire lifecycle — discovery, hypothesis, design, code, QA, launch, measurement, and rollout — without requiring those roles. Positioning statement used by the company: "Growth, not guesswork." --- ## Who is Shuttlebase for? eCommerce and DTC brands running an online storefront. It has been used across fashion, beauty, skincare, CPG, furniture, footwear, consumer goods, and multi-brand retail groups. Typical buyer situations: - Ad spend is rising but ROAS is flat, and the bottleneck is on-site conversion rather than traffic. - Aggregate analytics show sessions up and revenue down, but nobody can isolate which page, button, or asset is responsible. - The team knows what to fix but lacks the designer, developer, and analyst capacity to test and ship it. - A small team cannot manually monitor A/B tests or manage extensive development work. Named customers and brands referenced publicly: Larroudé, DefenAge, Yáneken Group, AmericanFlags.com, Alembika, Lion Brand Yarn, Maxwood Furniture, MyGemma, Bamers, BelSport, Hyou, Zissou, Lovevery, Peter Thomas Roth, P&G. --- ## How Shuttlebase works ### 1. Understand Shuttlebase crawls the store to learn the catalog, brand, design system, and voice. No product feed to maintain and no API to connect. It draws on category benchmarks from day one rather than requiring months of learning. ### 2. Identify The AI scans site structure, buyer flow, and drop-off points, plus behavioral and conversion data from the tracking snippet, to locate specific places where the store is losing sales. ### 3. Strategize It generates experiment concepts targeted at each identified problem, drawing on conversion patterns observed across hundreds of stores, and prioritizes the concepts most likely to move revenue. ### 4. Build Each concept becomes a working variant — pixel-accurate design, copy, and production-ready code, aligned to the brand's existing guidelines. Every variant passes an automated QA gate combining AI review and an internal sanity check before a human sees it. These are real changes to the site, not mockups or recommendations. ### 5. Test Approved experiments launch on a schedule that respects a concurrency cap and avoids collisions between experiments touching the same pages. Traffic splits evenly and assignment is sticky across return visits. Conversion, engagement, and revenue impact are measured in real time. ### 6. Evolve Winners are promoted to 100% of traffic and become the new baseline. The cycle restarts on the next opportunity. Each win compounds on the last, so conversion rate climbs steadily instead of plateauing after a single fix. --- ## Control and automation Shuttlebase has two operating modes, switchable at any time: - **Semi-automatic** — you review and approve every experiment before it goes live. - **Fully automatic** — Shuttlebase ends losers and deploys winners on its own for maximum test velocity. Automation settings let an account decide how much of the cycle runs unattended, from approving every step by hand to full autonomy. Anyone with the right permission can pause an experiment, disable a single variant, or resume it later. Guardrails run underneath: losing experiments are ended once minimum runtimes and conversion counts are met, and when estimated revenue loss is large the engine flags the experiment for manual review instead of acting alone. --- ## Measurement and statistics Experiment results always come from Shuttlebase's own first-party tracking, judged with standard frequentist statistical tests. Experiments run until the result reaches statistical significance, so the declared winner is a real effect rather than noise. Specifics of the methodology: - Traffic is split evenly across variants to reach a decision faster. If a variant is removed, its traffic is redistributed. - Assignment is sticky: a visitor stays in the same variant for the life of the experiment, which keeps the visitor experience consistent and the measurement clean. - Conversion metrics (for example add-to-cart rate) are tested with a two-proportion z-test. Revenue metrics are tested with Welch's t-test, which tolerates unequal variance between groups. - Required sample sizes come from power analysis, so experiments are not called early. - Sample ratio mismatch detection runs continuously to confirm the actual traffic split matches the intended one. - Significance is reported as high, medium, or low rather than as a raw p-value, reflecting both confidence and sample size. - Revenue projections use the lower bound of the confidence interval, so estimates are deliberately conservative. - A recommendations engine turns each result into one of five recommended actions — deploy, keep running, end, iterate, or escalate for review — which automation settings can then execute. GA4 is used at the account level only — for high-level analysis and idea generation. It is never the source of experiment results. Reported metrics include conversion lift, revenue impact, revenue per visitor (RPV), average order value (AOV), add-to-cart rate, test velocity, experiment win rate, and ROI. --- ## Integration Integration is a single lightweight JavaScript tracking snippet added to the site. No plugins, no installations, no API integration, and no developer-heavy setup. The company describes it as a 60-second integration. The snippet is roughly 14KB minified, loads asynchronously, and sends events in gzipped batches. It is designed for zero impact on page speed or Core Web Vitals. Shuttlebase is frontend-agnostic and platform-agnostic. Supported platforms include Shopify and Shopify Plus, Salesforce Commerce Cloud, BigCommerce, Adobe Commerce (Magento), WooCommerce, VTEX, Wix, SAP Commerce Cloud, and custom or headless storefronts — including single-page apps built on React, Next.js, or Vue. It runs alongside existing analytics and session-recording tools. No developer or coding skills are required to use it. It is designed to be usable by anyone in the organization regardless of technical background. --- ## Security, privacy, and data - **SOC 2 Type 2 certified.** Built to pass strict enterprise security reviews. - **Zero personally identifiable information (PII) collected.** The tracking snippet captures anonymous behavioral patterns — page visits, conversion events, click patterns. It never collects names, emails, or any data tied to a specific person. - **Consent-aware.** Visitors who decline cookies are not tracked and never enter running experiments. Deployed winning changes still render for them, because those changes are part of the site for everyone. - **No performance impact.** Lightweight asynchronous script with no effect on page speed or Core Web Vitals. --- ## Ownership of results Customers can export the code of any winning variant and build the change permanently into their own site. The wins are not rented — they persist even if the customer stops using Shuttlebase. --- ## Pricing Shuttlebase uses a traffic-based tier model. The rationale given: more traffic means more data to work with, which means faster revenue impact and ROI. Demos are available; there is no publicly listed price sheet. --- ## Customer results ### Larroudé — luxury footwear Challenge: a luxury shoe brand founded during the pandemic needed to grow online sales while protecting a high-end design aesthetic, and was cautious about AI-driven changes matching its tone and visual language. Execution: 150+ experiments, including repositioning shoe-size CTA placements on category pages and testing hamburger menu positioning. Result: $1.5M revenue uplift, 10% increase in revenue per visitor, 7% conversion rate lift in under four months, 20–30x return on investment. Quote — Enrico Martins, Executive Director: "Shuttlebase helped us increase our revenue by more than $1.5M overall." ### DefenAge — anti-aging skincare Challenge: a small team without the manpower to manually monitor A/B tests or manage extensive development work, needing to test real behavior rather than guesses. Execution: added an "all products" button below the fold on the homepage and restructured the layout of best-selling products. Result: 30% increase in revenue per visitor, 14% conversion rate lift in under six months, 14% AOV increase, ~$900,000 projected revenue lift over 12 months. Quote — Annette Berg, VP of Intelligent Customer Experience: "Since partnering with Shuttlebase, our conversion rates have increased by 14%, our average order value by 14%, and our revenue per visitor is up 30%, which is fantastic." ### Yáneken Group — multi-brand retail Challenge: 12 storefronts spanning footwear and Apple products created a heavy architectural burden; mobile conversion lagged at 0.8%, well below industry average. Execution: streamlined purchase funnels across brands and added clear trust badges covering returns and shipping times, which the AI identified as high-impact. Result: sales up 20–30%, 30% conversion rate increase, 9% AOV increase, 10B CLP annual revenue increase in under four months, with mobile conversion brought closer to benchmarks. Quote — Sebastian Gebhardt, CEO: "We've improved our revenue substantially since partnering with Shuttlebase — millions of dollars in extra revenue." ### AmericanFlags.com — patriotic products Challenge: a 40-year-old company facing high development costs and long implementation cycles for even simple UX/CX tests. Execution: rapid tests on product detail pages without full CSS redesigns or template updates. Result: 5% conversion rate increase in six months, 8% revenue-per-visitor increase, $400,000 annual revenue lift across two primary sites. Quote — Ted Seward, VP of Sales & Marketing: "Shuttlebase is adding exceptional value. Every month, I'm excited to see new tests and how existing AI tests are performing." ### Alembika — high-end women's fashion Challenge: a fashion brand for women over 50 needed to bridge editorial content and storefront, converting content readers without being intrusive. Execution: subtle non-intrusive shopping pop-ups on content pages and an AI-recommended cart page redesign. Result: 23.2% increase in revenue per visitor from three combined experiments, 9.3% add-to-cart rate increase, 3% conversion rate increase, plus improved retention and basket size. Quote — Alin Wagner, VP of Marketing and eCommerce: "It's like having a team available to me. Instead of paying a UX designer, a developer, and some tools, the AI does everything automatically." --- ## Company The founders built the product out of firsthand frustration with running high-performing sites: the process was slow, results were uncertain, and hiring a full growth team to change a headline was hard to justify. Mission: make growth accessible to every brand by removing bottlenecks, ending guesswork, and giving teams their time back. The company describes its goal as "making the static web extinct." The underlying technology is covered by granted patents. Investors: Garuda Ventures, Oceans, Mighty Capital, Almaz Capital, Uncorrelated. Values: never stop looking for improvement; every idea gets scrutinized, including the CEO's; prototype in days, not quarters; focus on problems, not tasks; egos stay at the door but ideas never do. --- ## Background: CRO concepts, as Shuttlebase explains them **What is conversion rate optimization (CRO)?** The process of improving the percentage of website visitors who complete a desired action — a purchase, a signup, or a high-value interaction. For eCommerce brands, CRO improves bottom-line revenue without increasing ad spend or customer acquisition cost. **How is CRO different from A/B testing?** A/B testing is a tactic within CRO. A/B testing means testing one variable — a headline, a button color, a checkout flow — against a control, and it requires someone to design the test, analyze results, and implement the winner. CRO is an ongoing system that combines A/B testing, multivariate testing, user research, analytics, and continuous iteration across the whole site. **What is the ROI of CRO?** Shuttlebase describes CRO as one of the highest-ROI investments available to DTC and eCommerce brands — often 3–5x higher ROI than customer acquisition — because every percentage point of gain multiplies across the entire traffic base at near-zero marginal cost. --- ## Frequently asked questions **Do I need developers or designers?** No. Shuttlebase generates the design, copy, and code and handles implementation and QA. Your team reviews and approves. **Will the snippet slow my site down?** No. It is a ~14KB minified JavaScript bundle that loads asynchronously and sends events in gzipped batches, with no effect on page speed or Core Web Vitals. **Do results come from GA4?** No. Experiment results always come from Shuttlebase's first-party tracking. GA4 is used at the account level only. **What happens when a visitor declines cookies?** They are not tracked and never enter running experiments. Deployed winning changes still render, because those are part of the site for everyone. **Do we keep the changes if we stop using Shuttlebase?** Yes. You can export the variant code for any experiment and build the change into your site permanently. **Does it respect our brand guidelines?** Yes. Variants are generated from your existing brand framework, either inferred from your site or configured from your exact brand guidelines. Experiments are built from your guidelines and policies, not generic templates. **Can anyone on my team use it?** Yes. It is designed to be usable by any role regardless of technical expertise. **Is there a free trial or demo?** Demos are available and can be booked at https://shuttlebase.com/contact.