Skip to main content

← Back to projects

NerdWallet · Case study

Credit card review experience

A rebuilt review template that made editorial judgment, rating confidence, product facts, and apply intent feel like one NerdWallet review system.

Role
Product Designer
Company
NerdWallet
Team
1 designer, 1 product manager, 1 engineer

Scope

  • Competitive Intelligence
  • Concept exploration
  • Redesign
  • Conversion

Labels

  • Shipped
  • Representative reconstruction

Summary

Executive summary

Problem

The credit card review template was the company's largest single source of revenue and one of its largest sources of organic traffic. The existing layout had drifted into a hybrid of editorial article and marketing landing page that served neither purpose well.

Approach

Rebuilt the template as a structured editorial surface with explicit zones for review content, comparison data, and monetization — each with its own rules, governance, and treatment.

Outcome

Shipped to the full card catalog without an organic traffic regression. Conversion lifted on the cards that adopted the new template ahead of an industry-wide rate cycle.

Guided Flow

Recreate the real review-page rhythm

A representative prototype of the review surface: headline and byline establish trust, the Bottom Line anchors the rating, and the product card carries the award, offer facts, rates, fees, and apply path.

Representative prototype rebuilt for portfolio review. No customer data or proprietary production screens shown.

Problem

Diagnosis

The template was simultaneously failing three audiences: readers who wanted a credible review, the editorial team who wanted to ship reviews without negotiating every layout, and the partnerships team who needed apply-now placements that did not feel grafted on.

  • Editorial team retros

    Writers reported that the layout pushed them toward marketing-style headlines they did not want to write, because the structural constraints did not give the review body enough room to breathe.

  • Onsite analytics

    Reader scroll depth dropped sharply at the boundary between the review body and the comparison module — the page felt like it ended there even though more content followed.

  • External SEO audit

    The template's structural data was inconsistent across cards in the catalog, hurting eligibility for review rich results.

Constraints

What was fixed

  • Editorial governance is sacred

    Apply-now placements and rate disclosures cannot interrupt or precede the reviewer's analysis. Editorial leadership held veto on any pattern that confused the two.

  • One template, hundreds of cards

    The template had to work for every card in the catalog without per-card overrides — the editorial team did not have capacity to maintain layout exceptions.

  • Rate data must be live, not baked in

    APR ranges, fees, and promotional offers change frequently. The template had to read from the rate service, not from copy embedded in the article body.

  • Lighthouse and Core Web Vitals as a launch gate

    Mobile LCP could not regress against the existing template at the 75th percentile. The visual rebuild had to net out faster, not slower.

Principles

Design principles

  1. 01

    Editorial first, monetization explicit

    Apply-now placements are clearly labelled and have a dedicated zone — they do not pose as editorial commentary.

  2. 02

    One canonical structure, not a system of overrides

    Every card review has the same modules in the same order. The system absorbs variation through content, not through layout exceptions.

  3. 03

    Rate data is a citizen, not an ornament

    Live rate values render with the same care as editorial copy — properly typeset, disclosed inline, and never abbreviated in a way that misleads.

Exploration

Concept Exploration

Rather than iterate on one layout, I explored four fundamentally different product strategies. Each concept optimized for a different hypothesis about how people evaluate financial products before deciding to apply.

Concept 01

Editorial First

Trust before conversion

Overview

A review page that withholds the apply path until the reader has seen enough editorial proof to believe the recommendation.

Hypothesis

People arrive from search looking for trusted advice, not marketing.

Success Metrics

  • Scroll depth
  • Reader engagement
  • Organic traffic
  • Returning visitors

Concept 02

Decision Dashboard

Optimize for fast decisions

Overview

A compressed product surface for readers who already understand credit card basics and need a fast yes/no read.

Hypothesis

Many visitors already know they want a card; help them decide in seconds.

Success Metrics

  • CTR
  • Apply clicks
  • Bounce reduction

Concept 03

Personalized Recommendation

The right card depends on the reader

Overview

A front-end decision assistant that reframes the review around fit: who should apply, who should hesitate, and why.

Hypothesis

Instead of reviewing a product, recommend whether this product fits this person.

Success Metrics

  • Quiz completion
  • Recommendation confidence
  • Qualified apply clicks
  • Downstream approval quality

Concept 04

Comparison First

Comparison drives confidence

Overview

A review that assumes readers are choosing between cards and makes the competitive set part of the primary interaction.

Hypothesis

Most users are deciding between multiple cards.

Success Metrics

  • Comparison engagement
  • Return visits
  • Apply confidence
  • Multi-card exploration

Design review artifact

Comparing the Concepts

Each direction optimized a different product bet, so the critique centered on fit and operating cost.

ConceptPrimary UserBusiness GoalStrengthsWeaknessesReason Not Selected
Editorial FirstSearch visitor validating whether NerdWallet is credibleProtect SEO trust and deepen review engagementBest expression of editorial independence; strongest fit for organic review intent.Under-serves decisive readers and makes monetization feel like a late reveal.Too pure for the business surface. The template needed trust first, but not conversion last.
Decision DashboardHigh-intent applicant comparing offer factsIncrease apply-path clarity and reduce pogo-stickingFastest read; strongest at making rates, rewards, fees, and CTA discoverable.Compresses the review into a product sheet and weakens the writer's judgment.Useful as a module, not as the whole page. It solved speed by sacrificing voice.
Personalized RecommendationUnsure reader who needs a fit check before applyingImprove qualified intent and long-term recommendation trustMost reader-centered; explicitly names best-fit use cases and warnings.High interaction cost and governance complexity for a catalog-wide template.Strategically promising, but better suited to a dedicated recommendation flow than a review template.
Comparison FirstShopper deciding between several similar cardsIncrease confidence through transparent alternativesMakes the competitive frame explicit and helps readers understand tradeoffs.Pulls attention away from the reviewed card before the review earns trust.Comparison needed to support the review, not become the opening argument.
Shipped template — five labelled regions describe how the review, rate data, and apply-now placement coexist without overlapping each other's jobs.
  1. NerdWallet shell — nav, category context, and review breadcrumbs
  2. Editorial hero — headline, summary, byline, and disclosure context
  3. Bottom Line — rating and concise editorial judgment
  4. Product card — award band, card art, CTA, and offer facts
  5. Page outline — persistent review navigation
  6. Rates and fees drawers — detailed partner information below the facts grid
Before: a generic portfolio diagram with separated rating, facts, and CTA blocks. After: a NerdWallet-like review page with the same headline, Bottom Line, award card, offer facts, and page-outline rhythm as the live review templates.

Outcome

What shipped

The new template shipped across the full card catalog over six weeks, gated by Core Web Vitals at the 75th percentile and an editorial sign-off per card. Editorial reported the new layout felt easier to write into, not harder.

Mobile performance net-improved against the legacy template — required, since CWV at p75 was a launch gate. Conversion lifted on cards on the new template during the migration window, but the result the team felt most was editorial throughput: the template did one thing the old one did not — it told the writer where the editorial content stops and where the comparison begins, so writers stopped negotiating that line every week. Exact figures are held back here because this case study is a reconstruction; the directional trends were real but I do not want to publish numbers I cannot pull from the original instrumentation.

Reflection

Looking back

The strongest lesson from this project was that the most contentious zone — the apply-now placement — got easier to design once the editorial zone was unambiguous. We had spent months trying to soften the apply module before realizing it only felt aggressive because the editorial zone was bleeding into it. Once they were properly separated, the apply module did not need to be soft, just clear.

Available

Want to talk about a problem in this shape?

Best fit for platform UX, growth systems, internal tools, and teams that need a senior IC who can shape the problem and ship the details. I respond to most inbound within a day.