Building a consumer product from 0 → 1 as Founding Product Designer at getCredible

Consumer apps trap users in engagement loops. For a B2B data engine, time spent is the wrong metric. The product is engineered to extract maximum information with minimal cognitive friction.

The onboarding flow plunges users immediately into polls populated by their own contacts. Within their very first session, users generate high-quality referral data—delivering core value even if they never return.

Initial metrics: 260 users voted talents, creating 10.1k matches and generating 33k performance insights.

This architecture creates a one-touch active data network. The value compounds at activation, not retention. Every single referral thickens the network, sharpens vertical leaderboards, and fuels viral growth loops through shareable cards and direct messaging integrations.

Outcome: scalable, viral, high-quality data generation engine.
The result is a self-reinforcing data engine. It scales with the viral dynamics of a consumer social network but yields B2B-grade intelligence, successfully bridging the hardest gap in product: consumer distribution scaled with enterprise monetisation.

The interface is relentlessly distilled down to one-click poll completions. By stripping away cognitive overhead and emotional barriers, a single poll is answered in under 4 seconds. The onboarding completely bypasses profile creation or complex tutorials. Users contribute instantly.

This ruthless simplicity drives absurd activation rates: two out of three downloads finish their first set of 20 matches immediately, generating performance data for hundreds of individuals in under four minutes. By optimising exclusively for conversion throughput, the system maximises data liquidity per user.

Every poll dynamically samples five peers from the user’s contact network. A proprietary algorithm sequences them to maximise information gain, balancing network diversity against signal uncertainty. A single tap dictates relative strength.

Under the hood, an Elo-like model instantly recalibrates the performance scores of both the voted and compared peers. To refine the upper percentiles, "crown polls" periodically force users to rank their absolute top connections, cementing reliability for the highest-tier talent.

High-quality data requires shared context. The system algorithmically infers dominant verticals (e.g., engineers, founders, designers) by parsing contact metadata and voting histories.

Polls are strictly confined within these domains. Users compare peers they genuinely know in contexts they actually understand. These domain-specific signals form the architectural backbone of getCredible’s high-resolution talent graph.

The funnel operates on a single mandate: maximise initial data generation. From the first screen, users are routed directly into preloaded polls via a one-tap interface.

Every micro-interaction—address book permissions, poll generation, the first vote—has been rigorously A/B tested to shred friction and build emotional momentum. The result is an onboarding funnel that converts first-time visitors into prolific data contributors before they even fully grasp the platform's long-term utility.

The data validates the design: 66% of landing users complete their first poll sequence.

On average, a retained session yields 28 standard polls and 10 crown polls. Vote times hover at 3-4 seconds, proving cognitive load is practically nonexistent. Drop-off analysis reveals that once a user starts a set, they finish it. The product extracts volume rapidly without relying on legacy engagement hooks like streaks or notifications.

getCredible is designed to be a passive data network. The overwhelming majority of enterprise value is extracted during short, sporadic sessions. Even dormant users continue to subsidise the network as their historical signals serve as baselines for new voters.

Activation—not retention—is the core growth engine.

Every interaction feeds segmented leaderboards—by vertical, company, and geography. These instantly visible social benchmarks generate curiosity and ego-validation.

Seeing peers ranked triggers immediate sharing and retaliatory voting. Shareable voting cards act as potent viral assets, heavily leaning into the STEPPS framework (Social Currency, Emotion, Public visibility). By tying professional identity to game mechanics, sharing becomes intrinsically rewarding.

Virality is community-led. Deep integrations with WhatsApp and iMessage turn the user’s phonebook into a frictionless distribution network. An invite is just a tap away.

Every new participant instantly becomes a poll target for existing users, tightly closing the viral loop. The k-factor within tight professional communities (VCs, engineers) rivals early Facebook. Shared social proof—e.g., “You’re ranked in the top 5% of London Engineers”—acts as a high-conversion trigger, driving zero-CAC reactivation and exponential growth.

Standard platforms map *who* you know. The performance graph maps *how you are perceived* by the people you know. Interactions map relational intensity and cross-signals across contexts.

Profiles auto-enrich via Google and third-party data layers. This dataset updates autonomously, creating a self-healing foundation for B2B talent discovery and monetisation.

getCredible isn’t a hiring platform; it’s a living data engine. While legacy networks rely on self-reported, stagnant CVs, getCredible harvests behavioural and relational signals that are invisible to the public web.

By automatically fusing peer interactions with contact graph metadata, the platform creates an un-gameable source of truth.

Profiles are algorithmically rendered via peer consensus. Vertical tagging and leaderboard placements happen autonomously, making the platform instantly monetisable for recruiters without demanding manual inputs from the talent.

Each click compounds the network's accuracy. As density increases, the models train faster, expanding a verified professional graph.

This proprietary dataset forms an impenetrable moat, powering models that predict performance and cultural fit with mathematical precision far beyond legacy platforms.

The core ranking algorithm fuses PageRank with Bradley-Terry principles. It weights votes dynamically based on voter credibility, producing an Elo-like score for 1-vs-many comparisons. Winning a vote from a high-status user shifts your score more than a standard vote.

Modifiers include job-title proximity (via LLM analysis), penalty dampeners for invited collusions, and global reliability scoring based on historical voting patterns.

The polling engine balances three proprietary layers: anti-spam, contact prioritisation, and objective optimisation. The anti-spam filter aggressively purges roughly 16% of the contact graph before polling, utilizing a local LLM (Phi-3.5-mini-instruct) to flag low-quality nodes.

The prioritisation schema ranks viable contacts based on metadata frequency and relevance. Finally, the optimisation layer dynamically toggles poll targets between tight industry peers and broader networks, balancing data integrity with user engagement. This compounding loop guarantees both an addictive user experience and high-fidelity data.

Traditional hiring is plagued by distortion. Human recruiters are bound by cognitive biases; current AI tools merely amplify the flawed data they are trained on. Over 40% of candidates admit to lying on applications, and remote identity fraud is skyrocketing. A single evaluator—whether human or algorithmic—is statistically unreliable.

getCredible approaches talent discovery as a statistical probability problem. Instead of trusting a single heavily curated CV, it aggregates hundreds of independent peer signals. Individual bias is massive; collective bias mathematically cancels out. It’s the Airbnb or Amazon rating model, applied to professional competency.

With users providing up to 200 comparative signals in under 10 minutes, the dataset converges on the truth exceptionally fast.

By weighting signals against voter credibility, the system aggressively filters noise rather than amplifying it. Like Google's PageRank, the precision and trustworthiness of the network improve automatically at scale. Platforms that successfully harness compounding data effects redefine their industries—and the human capital market is next.

At an initial milestone of 50,000 users—generating insights on ~200 peers each—getCredible will map the performance of the entire London tech workforce (600k+ professionals).

This instantly arms the $4B London tech headhunting market with access to the first nationwide, verified performance ranking.

Growth to date is 100% organic. The forthcoming phase integrates strategic brand positioning, targeted community partnerships, and PR to accelerate conversion at the top of the funnel.

By engineering weekly referral pulses, users will be nudged to invite peers to unlock their accumulated "wallet" value, transforming latent network equity into predictable viral surges.

Thanks to severe unit economic advantages, getCredible can bid up to 3x the competitor CPA while sustaining a 90% contribution margin. By incentivising invites directly, getCredible becomes a self-sustaining viral engine, funded by its own structural efficiency.