Unbiased insights to help you decide if solar is right for your house.

An end-to-end PM case study: problem framing, sourced UK market and regulatory research, three personas, synthetic user interviews, an ICE-prioritized roadmap, and a working interactive viability-check prototype.

First marketUnited Kingdom
DeliverablesBrief · Personas · Interviews · Roadmap · Working calculator
Introduction to SunnySideUp case study
Prototype

A working, coded viability-check calculator: postcode and segment (rooftop or plug-in) selection, self-reported consumption/tariff/occupancy inputs, and a green/amber/red result with the assumptions behind it shown openly. Built as a desktop web tool rather than a mobile app, so it's shown here on a laptop rather than in a phone frame.

The calculator is a full working tool. Open it full-screen for the real experience on your phone.

The short version
SunnySideUp gives UK homeowners, renters, and leaseholders clear, independent solar viability checks to help them take confident climate action.
Problem & positioning

Market signal

  • The CMA's 2022 investigation into the solar sector specifically flagged "inaccurate and unrealistic headline price information" (CMA case page).
  • A Which? survey found more than a third of solar-panel owners had been cold-called with false claims, including bogus "government-required" checks (Which?, Apr. 2024).
  • Every UK solar comparison platform identified in research (CompareMySolar, Uswitch, Solar Selections, Easy PV, Renewable Energy Hub, Solar Planet, Solar PV Quotes Compare) runs a lead-generation or referral-fee model.
  • BS 7671 Amendment 4 (in force 15 April 2026) legalized electrician-installed plug-in solar; true DIY self-install becomes legal 27 August 2026 under SI 2026/848 (IET/BSI, Apr. 2026; SI 2026/848).
  • 2025 was UK solar's record installation year, 31-37% growth, so rooftop demand, and the sales pressure that comes with it, is already at an all-time high (Solar Power Portal; pv magazine, Nov. 2025).
  • UK surveys consistently show a gap between climate concern and felt ability to act: one 2025 survey found 93% of respondents felt some degree of eco-anxiety, many describing frustration at caring about climate change while having no way to act on it (Reading Agency, 2025).

Positioning

SunnySideUp doesn't sell, install, or take referral fees from installers. It's an independent viability check used at two moments: before any sales conversation, to set a baseline, or after receiving a quote, to get an unbiased second opinion. The closest real analog is Energy Saving Trust, a government/private-funded impartial-advice body, but even it defers the final personalized number back to the installer being evaluated. SunnySideUp gives that specific, personalized read directly.

Market opportunity

No single figure captures the plug-in category's size yet since it's too new to measure, but two forward-looking signals and one comparable-market analog suggest real scale.

UK Solar Roadmap
45-47GW

Government target for installed solar capacity by 2030, up from ~19GW in May 2025, roughly 2.5x growth in five years, with rooftop prioritized.1

No-roof-access households
~11M

English renters (~8.8M, ONS) plus owner-occupied leaseholders (~2.6M, MHCLG), the plug-in category's addressable population.2

Balkonkraftwerke (Germany)
270K → 1M+

Germany's comparable balcony-solar market, 2023 to mid-2025, growth accelerating after renters gained an explicit legal right to install in 2024.3

  1. UK government 2025 Solar Roadmap. See grounding-research.md §UK solar capacity trajectory.
  2. ONS renting-household data combined with MHCLG leaseholder statistics. See grounding-research.md §No-roof-access population size.
  3. Germany's Balkonkraftwerke market, the closest comparable-market signal for a category too new to measure directly in the UK. See grounding-research.md §Balkonkraftwerke growth in Germany.

Competitive landscape

CompanyModelGap SunnySideUp targets
CompareMySolar, Solar Selections, Easy PVLead-generation / referral-feePaid per lead, incentivized to maximize quotes, not clarity
Uswitch solar comparisonReferral-fee comparison siteSame conflict of interest, general-comparison brand extension
Renewable Energy Hub, Solar Planet, Solar PV Quotes CompareLead-generationSame referral-fee model across the board
Energy Saving TrustGovernment/private-funded impartial adviceClosest structurally-independent analog, but its own calculator uses generic assumptions and defers the final number back to an installer

No independently-funded platform identified in research gives a direct, personalized viability read without routing the user back to an installer. That gap is SunnySideUp's bet.

Research
Graham Whitfield
58, Solihull · The Sales-Call Avoider
"I don't want a salesman doing my sums for me before I've even worked out if this makes sense."
JTBD

An independent read on his specific financial case before any installer contacts him.

SunnySideUp feature

Pre-conversation viability check

Denise Okafor
47, Reading · The Quote Skeptic
"Two companies, two numbers four thousand pounds apart, and neither one explains why."
JTBD

An independent check on whether a quote's price and system size are reasonable before signing.

SunnySideUp feature

Post-quote second-opinion check

Aisha Rahman
29, Manchester (renter) · The Plug-In Fence-Sitter
"Everyone online says I can just plug it in. No two of them agree on whether that's actually still true."
JTBD

A clear, unbiased read on both legality and value for a plug-and-play setup, calibrated to her own skills.

SunnySideUp feature

Plug-in viability + legal-status check, with an honest "still unresolved" flag

What synthetic interviews surfaced

All three personas were interviewed with the same guide (AI-generated, in-character, grounded strictly in each persona's brief and real UK research, explicitly flagged as synthetic). Cross-persona themes directly shaped the roadmap:

01

Trust comes from independence plus honesty about limits

All three trust conditions center on non-referral independence, and, for Aisha, a willingness to say "this part is genuinely unclear."

02

A "red" result is trusted and acted on

All three would accept an unfavorable result and change their behavior, specifically because of the non-referral positioning.

03

Sharing is peer-to-peer

All three said they'd tell someone specific, a neighbour, a sister, a renters' Facebook group, rather than post publicly.

04

Aisha's disappointment is capped, and no product design fixes that

Her ceiling is structurally lower than the rooftop segment's for as long as the tenancy-consent question stays legally unresolved, independent of what SunnySideUp itself does well.

Prioritized roadmap

NOW MVP

Rooftop viability calculator (core engine)
ICE 7.7 · Graham & Denise JTBDs
Self-reported data input flow
ICE 7.7 · consumption, tariff, occupancy, roof/socket, financing
Plug-in viability + legal-status calculator
ICE 7.0 · Aisha's JTBD, explicit "still unresolved" flag
Assumptions / "show your work" transparency panel
ICE 8.3 · highest score, direct trust-theme match
Honest "red" result with substantive reasoning
ICE 8.0 · a stated reason behind every result

NEXT Post-validation

Basic quote-comparison (manual entry)
ICE 8.0 · Denise's JTBD, gated on MVP validation not feasibility
Shareable result / referral summary
ICE 6.7 · organic word-of-mouth
Quote document upload/parsing
ICE 5.3 · removes Denise's manual-entry friction
Battery/storage add-on calculator
ICE 5.3 · thinnest-sourced figures in the research

LATER Vision

Real data integration (smart-meter / bank-linked)
No UK equivalent mechanism researched yet
Scotland/Wales-specific guidance
Distinct regimes, explicitly not researched in depth
Institutional/DNO/local-authority portal integrations
A distribution motion, not an engineering roadmap item
Further legal research on tenancy-agreement consent
A research task, tracked separately in Open Questions

A prioritization call worth showing

The plug-in calculator's Confidence score (6) is deliberately the lowest in Now. Aisha's segment is the newly-legal, most differentiated part of the pitch, but grounding-research.md itself flags plug-in cost and generation figures as its weakest-sourced section, and that caveat ships inside the tool's own output to real users, carried through to launch rather than dropped along the way.

Limitations

Industry commentary on AI-assisted prototypes in 2025-2026 converges on one point: a polished demo and a shippable product are different things, and it's easy to mistake one for the other. A few of the gaps commonly named by product practitioners, applied honestly to SunnySideUp:

A working demo isn't 10% of the way to done

  • The calculator genuinely live-fetches three real sources: a postcode's region and coordinates from postcodes.io, solar irradiance history from Open-Meteo, and current electricity prices from Octopus Energy's public API, each tagged with its own sourcing tier in the code.
  • What's still missing is everything around that calculation: no real backend, no account or data persistence, and no quote-upload parsing (that stays a self-reported, manual-entry comparison for now, and is already scoped as a Next-horizon, post-validation roadmap item rather than an MVP gap).
  • The underlying payback and generation figures are wide, industry-consensus ranges corroborated across multiple sources; no single authoritative UK payback dataset exists to draw a tighter number from.
Source: Tony Fadell, on Lenny's Podcast ("if you're going to build a real company, it can't be throwaway")

Automated design still needs real users

  • AI-generated screens can miss subtle hierarchy, spacing, and comprehension issues that only surface with real usability testing.
  • SunnySideUp's personas and interviews are synthetic and internally consistent, but were never tested against an actual homeowner, renter, or leaseholder clicking through the flow.
Source: Judd Antin (Airbnb, Meta), on Lenny's Podcast, on evaluative/usability testing as one of the research tools AI hasn't replaced

Some edge cases are handled, most aren't

  • AI-built prototypes typically skip error states and unusual inputs entirely, often by design: Lenny's own template prompt for AI-generated prototypes explicitly scopes them to "happy path only for now." This one is a partial exception: an unrecognized postcode, a failed lookup, or an unexpected API error each get a specific message and a graceful fallback to an England-calibrated default, rather than a silent break.
  • What's still unhandled: an unusually shaped roof, conflicting self-reported values (e.g. a stated consumption figure that doesn't match the household-size estimate), and any quote-upload edge cases, since that feature doesn't exist yet.
Source: "Build your personal AI copilot", Lenny's Newsletter

Prototype code isn't production code

  • AI-generated prototypes are typically isolated from any real codebase and would need to be re-implemented by an engineering team to actually ship.
  • The prototype is static HTML/CSS/JS with no data layer, no security review, and no compliance pass, relevant since a real version would handle a user's address, energy usage, and quote details.
Source: Zevi Arnovitz (Meta), on Lenny's Podcast ("the big difference between just vibe coding and going along with the vibes and really building serious apps")
Full documentation

For anyone interested in documents on product development, platform and technical integration, growth, or product analytics, get in touch.