Turning complex problems into practical digital products.

Product vision & strategy

MyNextReel started from a specific customer problem: people have access to more streaming content than ever but still spend too much time deciding what to watch.

I defined the proposition around solving that decision problem rather than trying to recreate a streaming platform. Recommendations are designed around the individual rather than popularity, paid placement, sponsorship or other commercial influence.

Australia is the initial market so streaming-provider availability, customer behaviour and the commercial model can be validated in one market before expansion. The initial proposition is deliberately simple: personalised recommendations across the services a customer already uses, supported by a low-cost subscription.

The wider vision is significantly larger, including international expansion and eventually independent-film streaming. I deliberately separated that long-term ambition from what actually needs to be proven first.

• Defined the core customer problem and proposition
• Commercial independence established as a product principle
• Australia selected as the initial validation market
• Simple subscription model rather than multiple product tiers
• Personal fit kept distinct from commercial promotion
• Long-term ambition separated from MVP scope

Market opportunity & evidence base

Before moving further into the product design, I wanted to test whether the opportunity was supported by actual market behaviour rather than relying on the idea alone. I sourced current Australian streaming and population data, structured it for analysis and built a Tableau dashboard to explore adoption, paid streaming usage and the growing use of multiple services.

The analysis showed that streaming is already mainstream in Australia, with a large paid audience and significant use of multiple platforms. That matters for MyNextReel because the problem is not access to content. It is helping people navigate an increasingly fragmented set of services and make a decision about what to watch.

I used the dashboard as part of the evidence base for the business case and product direction. It gave me a clearer way to test assumptions, identify the scale of the opportunity and communicate the findings through a reporting view rather than presenting individual statistics in isolation.

The work followed the same approach I use for data and reporting problems more broadly: start with the business questions, identify and validate the information needed, structure the data into a usable form, build the reporting output and then interpret what the results mean for the decision being made.

• Defined the business questions the market analysis needed to answer
• Used real Australian streaming and population data rather than invented market assumptions
• Structured source data into a reporting-ready dataset
• Built and reviewed the analysis in Tableau
• Compared adoption, paid usage, multi-service behaviour and provider usage
• Translated the results into business-case and product insights
• Used reporting to support decisions rather than treating the dashboard as an end in itself

Customer journey

I mapped the customer journey as a continuous learning loop rather than treating onboarding and recommendation as separate one-off interactions.

The journey begins with discovery, account creation and preference collection before asking the customer to rate enough movies to create an initial recommendation profile. Inputs include streaming services, language preferences, subtitle comfort, genres and other preference signals.

The journey then continues through recommendations, choosing where to watch, saves and post-watch feedback. Those later actions are important because the product should become more useful as it gains real evidence about the customer.

I also separated streaming availability from personal matching. A movie can be a strong personal match while still being unavailable on the customer's selected services. Keeping those concepts separate protects the integrity of the recommendation logic.

• Discovery through to post-watch feedback mapped end to end
• Progressive collection of customer preference information
• Initial ratings used to create the first recommendation profile
• Favourite, Like, Meh, Dislike and Haven't Seen treated as distinct inputs
• Availability separated from personal relevance
• Feedback designed to improve later recommendations

MVP prioritisation

One of the key product decisions was deciding what not to build. The wider vision includes machine learning, international markets, independent streaming and significantly more sophisticated engagement features, but those capabilities do not need to exist to test the core proposition.

I therefore prioritised the MVP around proving whether customer information can be converted into useful personalised movie recommendations and whether customers find enough value in those recommendations to continue using the product.

The initial matching approach is intentionally explainable and rules based. Building a machine-learning model before real customer behaviour exists would add technical complexity without providing evidence that the central proposition works.

The roadmap separates what is required now, what can improve recommendation quality and retention next and what only becomes relevant if the product proves it can scale.

• Core Australian recommendation proposition first
• Account, onboarding and provider selection
• Initial movie ratings and rules-based matching
• Personalised recommendations and availability
• Basic saving and feedback loop
• Advanced learning delayed until real customer behaviour exists
• International markets and streaming positioned as later strategic capabilities

Requirements & delivery

I translated the product decisions into a structured backlog rather than moving directly from interface ideas into development.

Requirements cover customer behaviour, recommendation logic, availability, data inputs, edge cases and the expected outcome of each interaction. Acceptance criteria are used to verify work before it is considered complete.

A recommendation story, for example, needs more than a screen showing movies. It needs rules defining who the recommendation is for, what data can be used, minimum evidence requirements, Australian availability, what the customer can do with a result and how that action is captured.

The backlog has also been structured into releases so new ideas do not silently expand completed scope or displace work required for the MVP.

• Epics, stories and acceptance criteria
• Customer behaviour and edge-case requirements
• Recommendation and availability business rules
• Data and integration requirements
• Release and fix-version planning
• Verification before completed work is committed
• Later changes treated as new backlog work

Matching engine

The matching engine is the central capability behind MyNextReel. It combines customer evidence with movie data to rank suitable viewing options.

My role has been to define how that recommendation process should behave as a product. This includes the evidence that can influence a score, how positive and negative preferences are treated, how conflicting evidence is handled and how availability interacts with recommendation delivery.

The engine has progressed from a simple proof of concept into an advanced rules-based ranking pipeline covering candidate selection, filtering, evidence, confidence and diagnostic validation.

The current approach establishes an explainable baseline. Later machine learning can then be measured against a functioning system rather than being introduced simply because it appears more sophisticated.

• Genre, keyword, language and preference evidence
• Positive, negative and conflicting signals
• Confidence and missing-evidence handling
• Commercial influence excluded from personal scoring
• Streaming availability kept separate from matching
• Automated validation across recommendation scenarios
• Rules-based baseline for later real-user learning

UI & UX direction

I developed the visual and interaction direction through customer journeys, annotated desktop and mobile wireframes and evolving interface concepts.

The experience uses video-store-inspired colours, typography, shelf layouts, ticket shapes, textures and handwritten elements to move the product towards the familiarity of an independent video shop.

The nostalgic direction is deliberately balanced with simple forms, clear navigation and readable content. The visual identity should make the product distinctive without creating additional effort for the customer.

• Annotated desktop and mobile wireframes
• Video-store-inspired product direction
• Progressive onboarding interactions
• Shelf-based recommendation concepts
• Consistent typography and visual components
• Usability prioritised over decorative complexity

Strategic roadmap

The five-year roadmap separates validation from scale. Australia is the starting market rather than the final market and early investment is focused on proving recommendation value, paid demand and retention before materially increasing operating complexity.

Later phases introduce broader markets, independent-film streaming and the organisational capability required to support them. This includes content acquisition, rights management, localisation, data infrastructure and a larger specialist team.

The strategic principle remains the same throughout the roadmap: recommendation scoring remains independent even if MyNextReel eventually owns or licenses content.

• Year 1: validate the proposition in Australia
• Year 2: establish retention and a sustainable Australian business
• Year 3: begin transformation and independent streaming capability
• Year 4: expand across additional markets
• Year 5: prepare for broader global scale
• Organisation and investment increase only as evidence justifies them

My approach

I treated the work as both a customer facing redesign and an internal process improvement project rather than copying the existing Wix pages into new technology.

I reviewed the content library, existing article structure, search visibility, affiliate model and the steps required to create and maintain posts. This separated the parts of Wix that were creating genuine friction from the content and URLs that still had value.

The largest authoring improvement came from replacing repeated manual page construction with a structured template and database backed content process. Common article elements could be entered once through a consistent form and rendered automatically into the finished page.

This allowed the new solution to address both sides of the problem: make the experience clearer for the reader while making it substantially easier for the creator to maintain and expand.

• Mapped the existing publishing process and its friction points
• Reviewed existing content, URLs and customer pathways
• Defined customer, creator, commercial and technical requirements
• Designed reusable article and product structures
• Introduced template based content creation
• Planned redirects, indexing, analytics and post launch checks

Business case & performance targets

Before committing the time to rebuild the site, I wanted to understand whether the change could be justified beyond greater technical control and a less frustrating publishing process. I used the available Search Console, Amazon Affiliate and platform cost data to create a measurable current-state baseline and a set of future-state targets.

Q1 represents the monthly average from January to March 2026 and Q2 represents April to June 2026. Organic search data comes from Google Search Console, affiliate activity comes from Amazon Associates and platform costs use the actual Wix subscription and known replacement-site operating costs.

The Wix trajectory carries forward the proportional movement observed between Q1 and Q2 rather than assuming an arbitrary decline. The self-built figures are targets rather than claimed results and model what I want the replacement platform to achieve by changing publishing capacity, technical control and cost.

The purpose of the model is not to claim the redesign guarantees a specific traffic result. It creates a measurable business case and a baseline that can later be compared with actual post-launch performance.

• Q1 and Q2 use complete-quarter actual data
• Wix trajectory extends observed Q1 to Q2 movement
• Self-built Q3 and Q4 figures are targets
• Organic visits linked to impressions and CTR
• Affiliate targets linked to qualified traffic growth
• Publishing capacity increases from three to six then eight updates per month
• Wix cost approximately $41.67 per month
• Replacement platform approximately $5 per month after initial domain cost

Delivery evidence

The redesign was delivered as a custom web application using Next.js, JavaScript, HTML, CSS, database backed administration and Railway hosting.

AI-assisted development supported implementation while I retained ownership of requirements, solution design, validation, prioritisation and deployment decisions.

Validation covered both the customer-facing experience and the operational requirements needed to protect existing search traffic and measure the new platform.

• Redirect testing for previous Wix URLs
• Search Console and sitemap configuration
• Mobile and desktop validation
• Product image standardisation and creation
• Affiliate link checking
• Security checks for exposed files and routes
• Production monitoring and indexing reviews

Outcome

The project transformed Music Thought House from a platform dependent content site into a more focused and controllable digital product with a publishing process designed around how the site is actually maintained.

One of the most significant operational improvements has been reducing the actual construction of a post from roughly four to six hours to under one hour. The template system removes much of the repetitive layout, image and affiliate-link administration that previously had to be completed manually for every article.

That time saving changes the economics of creating content as well as the experience of doing it. More time can now be spent on research, writing and image creation while the platform handles more of the repetitive production work.

• Post construction reduced from four to six hours to under one hour
• Reusable templates replace repeated manual page construction
• Greater control over performance, SEO and deployment
• Higher potential publishing capacity
• First-party visibility of article and affiliate performance
• Lower ongoing platform cost
• Defined targets against which the rebuild can be measured

Project background

I had accumulated a large collection of interview questions, prepared answers and examples from previous roles. The information was useful but the document format was difficult to navigate during a live video interview.

I needed a solution that could remain open on one half of a laptop screen while Microsoft Teams occupied the other half. It had to make a large amount of information easy to scan and preserve the original wording of my notes.

Rather than introducing an unnecessary platform or database, I defined a tightly focused browser solution that could be created and used immediately.

• Fit comfortably within half a laptop screen
• Keep the original wording unchanged
• Make every topic quickly accessible
• Require no account, installation or setup
• Remain simple enough to create and improve quickly

MVP solution

The first version was built as a self-contained HTML application that could be opened directly in a browser.

A compact topic cluster acted as the main navigation. Selecting a topic moved directly to the relevant notes while return controls allowed the user to move back quickly.

Search provided another route to information when an unexpected question was asked.

• Clickable topic navigation
• Smooth movement between questions and notes
• Search across headings, tags and content
• Back-to-top controls throughout the page
• Sticky navigation
• Responsive half-screen browser layout
• No external hosting dependency