TOFU Acquisition Strategy
Want to back the stream? The «داعم» membership on YouTube is EGP 60 a month: badges, emoji, and I answer your questions first on stream. Become a supporter
The gist
The stream documents wide-ranging practical work on Burnish, a GEO application for Shopify merchants intended to measure store and product visibility in AI answers and recommend catalog improvements. The session focused on rebuilding and simplifying onboarding, polishing the scan page, report-wait experience, and mobile UX, and reviewing plan selection, Shopify billing, and data-processing consent. In parallel, it developed a Meta/Facebook acquisition funnel built around a free report and signals such as scan started, scan completed, and lead, with a recurring view that the mid-funnel and user journey should be improved before ads run. The work also covered clean-state testing and troubleshooting 500 errors, scan failures, deployments, Turnstile, tracking events, Stripe redirection, and Supabase, Cloudflare, Railway, and GitHub integrations; several causes and outcomes remained unresolved, and end-to-end verification was not shown as complete. Performance testing found that reducing sampling depth did not deliver the expected speedup in one attempt, leading to discussion of progressive results and a clearer waiting experience. Audience questions and broader discussions covered home-hosted databases, building Shopify apps through vibe coding, learning programming with AI, GEO versus SEO and AEO, dropshippers as a possible audience, and personal comparisons of Codex, Claude, and agent workflows. The indexed topics also cover separate discussions of Shopify-app market economics and saturation, D2C and Egyptian market strategy, cash flow and AI-tool spending, internet and OBS troubleshooting, digital-access constraints, digital-nomad visas, and asynchronous communication. The supported conclusion is that clear value, low-friction onboarding, reliable funnel behavior, and sound measurement should precede advertising scale; final product readiness, several bug causes, and the best landing-page and consent structure remained open.
183 moments from the stream
- 6:00:55
The One Outcome Onboarding Should Make Clear
The discussion clarifies that this is the one-time onboarding experience before a merchant reaches the Burnish dashboard. Its central outcome should be showing merchants what they are currently missing and how Burnish will help address that problem.
- 6:02:06
Presenting the Problem and Solution While a Scan Runs
The discussion proposes making a lengthy scan feel less static by showing the problem, a potential solution, and visible signs that the report is progressing. The aim is to help merchants wait, while still emailing the report if they leave.
- 6:04:43
Prompting AI Tools with the Desired Outcome
The speaker says he can specify design and advertising details, but for backend work he describes the desired final result and leaves the implementation approach to the tool. He also describes using Claude to suggest ways to present the problem, solution, and bridge during onboarding.
- 6:07:17
Opinions on Google Releases and Coding Models
The discussion offers a personal view that Google often releases products quietly and mentions unverified leaks about new models. The speaker considers Google relatively behind in coding models while still strong across other areas and products.
- 6:09:39
What Is GEO, and Why Does Burnish Start with Shopify?
GEO is explained as generative engine optimization: similar to SEO, but focused on product discovery through AI platforms. Burnish aims to assess Shopify stores’ AI visibility and suggest or apply improvements, using Shopify as a narrower initial market before any possible expansion into a standalone platform. The speaker dates GEO’s beginnings to 2023 and says its trend accelerated in June 2025.
- 6:12:01
Using a Complete Report as Value, Not a Paid Product
Responding to a suggestion to hide part of the solution behind payment, the speaker says the report is meant to demonstrate real value rather than serve as the product being sold. The intended core value is for Burnish to analyze store issues and then make improvements aimed at stronger visibility on AI platforms.
- 6:14:07
Risks of Putting a Long Scan Inside Onboarding
The concept is considered useful for a website-scan funnel but risky inside Shopify app onboarding because it may require a long wait and could also create a quick-uninstall risk. The discussion suggests another way to display the problem, solution, and bridge.
- 6:16:27
Daily Inventory Optimization Updates
The speaker explains that the application's agent checks a merchant's inventory every 24 hours and applies optimization when it finds new items—near-real-time updating rather than second-by-second processing.
- 6:18:20
Choosing Between Claude Code and Codex by Task and Cost
The speaker personally prefers Claude Code for coding while regarding Codex as strong in general intelligence based on benchmarks he cites. He argues that the choice should balance capability, usage cost, and available limits, favoring a tool that offers more working capacity when the budget is constrained.
- 6:23:27
How GEO and AEO Differ from SEO for Stores
The speaker argues that a merchant or product appears in AI answers through a system of signals and data rather than conversational memory alone. He cites llms.txt files and FAQs among practices that overlap with SEO while noting additional GEO/AEO requirements. He also discusses search trends he says emerged around June 2025 and rose in September 2025, then predicts another September surge, although the later year stated in the discussion is inconsistent.
- 6:28:22
Simple, Relatable Onboarding over Gimmicks
While reviewing an interface, the speaker notes that the glyph does not appear to respond to the changes. He then calls for onboarding that is simple, relatable, and not gimmicky, while still showing more creativity.
- 6:32:23
Making Onboarding Questions Optional
A brief discussion about not requiring users to answer an onboarding question, followed by a request to revisit the full onboarding sequence with that quieter approach.
- 6:34:20
Differing Views on Chinese AI Models
A discussion of Kimi K2’s context window and the direction of Chinese AI models, including a question about whether DeepSeek leads that direction and a counterpoint that other Chinese models do not all depend on it. Some version names are unclear in the captions, and the prediction about a model becoming a leader within a year was the speaker’s opinion.
- 6:38:14
Preparing Merchant Catalogs for AI Visibility
In response to questions about new stores and large catalogs, the speaker suggests prioritizing the most important products to protect service economics and cautiously compares the situation with the waiting period new domains may face in SEO. He stresses that quick visibility on AI platforms cannot be guaranteed; the defensible promise is better preparation and visibility measurement, while platform behavior toward new merchants remains uncertain.
- 6:41:20
Reducing Context-Window Impact in Sequential API Tasks
The speaker explains that a workflow using sequential API requests to process products may not require a very large context window, and that conversation compaction can let the task continue. In this case, he considers the economics of how many products are processed more important than context capacity.
- 6:42:56
Keeping users engaged during long report generation
A proposal to inspect the report-generation interface and animate its elements progressively during a three-to-five-minute wait, while telling users they can leave and return and will receive an email notification.
- 6:45:16
An opinion on data collection in China and the United States
The speaker expressed the opinion that entities in China and the United States collect user information in different ways; some of the following details are unclear in the captions.
- 6:46:32
Blank trial page report
A brief testing observation that clicking the trial button in a shared link opened a blank page; the captions do not provide enough context to identify the product or cause.
- 6:46:59
Debugging and handoffs between Claude and Codex
A workflow is described in which Claude inspected the application through developer tools and button interactions, then passed step-by-step prompts to Codex for implementation; the speaker attributed faster progress in an earlier stream to this approach.
- 6:48:09
Mapping Shopify app onboarding and billing
A review of the user journey from the welcome screen to the dashboard, excluding app-store installation and OAuth consent as Shopify-native steps, while emphasizing that pricing-tier selection and Shopify billing activation must be counted before dashboard access.
Got a question about something in this stream?
Ask in the group