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
- 6:51:42
An opinion comparing Codex and DeepSeek
In a brief discussion, the speaker expressed a preference for Codex in several respects and asked someone who uses both tools to compare it with DeepSeek. This remains a personal opinion, with no benchmark testing presented.
- 6:53:55
Parallel Catalog Analysis During Onboarding
After the user continues, a background scan was described as reading the catalog, inferring categories, generating prompts, and performing multilevel fan-out in parallel.
- 6:54:10
Exploring Multiple AI Models Through NVIDIA NIM
NVIDIA NIM was suggested as a way to try multiple models. The speaker qualified that earlier exploration was not deep and described a practical two-agent setup with an orchestrator and a specialist.
- 6:56:28
Open Question About Handling Dark Patterns
An audience question asked how the model would handle dark patterns in a product, but the speaker requested clarification and no answer appears on this page.
- 6:57:51
Balancing Development Speed With V1 Quality
The speaker wanted to move quickly without scattering the work, prioritizing a solid first version that could compete before attempting to surpass competitors in a later version. Expectations about future results remained conditional and uncertain.
- 6:59:41
Requirements for Refactoring the Onboarding Experience
A precise handover to Codex was planned for a full onboarding refactor because the current flow no longer reflected the latest decisions. Requirements included a seven-day free trial, monthly billing by default with an annual option, a substantial design improvement, and reconsideration of where AI-engine-related DPA consent is requested before users continue. The necessity and placement of that consent step remained unresolved here.
- 7:05:52
Should an Extra Consent Page Appear at Product Launch?
The speaker considers postponing an additional consent page because the product is just getting started and, in his observation, competitors do not show one. He clarifies that this opinion concerns only that page—not removing the privacy policy, terms of service, or Shopify’s installation consent screen—and explicitly invites pushback on his reasoning.
- 7:07:34
Specialization Within Digital Marketing and Freelancing
The speaker explains that SEO and GEO sit within the broader digital-marketing field, while his own specialization is paid advertising. He sees value in freelancing and recommends it to some young people, while stressing that its suitability depends on the person and project.
- 7:10:03
Structuring the Onboarding Flow as a Codex Milestone
The speaker proposes a prompt-based handoff for the complete onboarding flow: collect the funnel steps in a dedicated command-center note, divide the large effort into natural tasks, and let Codex execute the resulting milestone.
- 7:13:24
A GEO App for Shopify Merchants’ AI Visibility
Answering what he is building, the speaker describes a GEO application intended to help Shopify merchants increase their visibility in AI systems. He also clarifies that the /scan page is separate from the current work and has its own pipeline and funnel.
- 7:16:14
Carefully Refactoring the Onboarding Experience
The speaker separates the task from other concerns and makes a full onboarding-system refactor the priority, with careful design attention before proceeding with milestone preparation.
- 7:19:27
When Milestones Are Not the Right Workflow
The speaker argues that some funnel-activation work requires direct collaboration and is not well suited to being handed off as an isolated milestone.
- 7:19:46
Keep Marketing Fluff Out of Product Descriptions
The speaker criticizes urgency, scarcity, and social-proof claims inside product descriptions, saying they conflict with SEO and AI-oriented best practices. He recommends using the description for precise product facts, materials, and characteristics while keeping promotional messaging elsewhere.
- 7:21:54
Using Prompts Based on Task Size
The speaker explains that non-small tasks should come with a prompt and says he prefers this for most tasks, while small tasks may not need one.
- 7:37:26
Delegating Research on a Recurring Software Issue
When an issue previously fixed during the session returns, the speaker proposes assigning a sub-agent to investigate how it was resolved before. He also mentions preparing ten prompts for Codex and estimates the related work at roughly an hour.
- 7:43:42
Prioritizing the Burnish Funnel and User Onboarding
The speaker identifies two important priorities: completing the Burnish funnel needed before running ads, and working in parallel on the application onboarding process. He expects finishing both to produce substantial progress.
- 7:44:51
Form Progress Despite Minor Bugs
The speaker says work on the form is still underway and progressing, although a few minor bugs remain.
- 7:45:07
Optimizing Ads for the Middle of the Acquisition Funnel
An explanation of an acquisition funnel that begins by collecting a URL and business contact information, then moves people into lead and qualified-lead stages. The stated approach is to optimize Facebook campaigns for middle-of-funnel leads rather than final purchases that may occur incidentally.
- 7:48:26
Designing a Self-Executing and Reviewing Codex Workflow
A proposed workflow for folding findings into a prompt that asks Codex to execute an entire milestone with sub-agents, assign review items to an auditor, and move passing work to done. It also calls for investigating an auto-mode command and producing a suitable prompt; the captions do not establish that the workflow was executed or succeeded.
- 7:52:32
An Observation on Speaking Maghrebi and Egyptian Dialects
The speaker offers a personal view that a mix of Arabic, Amazigh, and other linguistic influences helps some Moroccans and Algerians learn dialects, and that speaking Egyptian Arabic is easier for them than speaking North African dialects is for Egyptians. It is presented as an opinion, not a demonstrated linguistic finding.
Got a question about something in this stream?
Ask in the group