Methodology
The 51-case stress test, published in full
On 22 July 2026 we ran the live screening engine — the same model and DSIT rubric that powers this site — over 51 scenarios written to represent the cases a UK accountant actually sees, including nine deliberate traps built to fool a tool that rubber-stamps anything that sounds modern. Every case, and the band the engine returned, is below.
How the test was built
The 51 cases split deliberately: 15 genuine qualifiers dressed in ordinary industrial clothing (a sauce maker, a concrete firm, a games studio), 17 plausible-but-vague borderline cases where the right answer is a settling question rather than a verdict, and 19 non-qualifiers — including nine traps engineered to sound like innovation while failing a specific DSIT test — plus one sole-trader eligibility edge case.
Expected bands were fixed before the run, by applying the DSIT Guidelines the way a competent R&D adviser would. The engine screened each case blind — plain-English description in, cited verdict out — with no per-case tuning.
Result: all 51 landed in the expected band, each verdict citing the specific DSIT paragraphs it relied on (5.1 citations per case on average), with a drafted four-section HMRC narrative on every qualifier and the failing test named on every rejection.
The traps it refused to swallow
Each of these sounds like innovation. HMRC would reject all of them — and so did the engine, citing the exact test that fails.
- “First time we’d done it.”A fabricator’s first aluminium-welding job — company upskilling, not a field-level advance (para 6, 8).
- Commercial risk in technical costume.A Stripe pricing engine whose “hard part” was maximising revenue without churn — a commercial uncertainty, not a technological one (para 28).
- Modern-sounding, actually routine. A help-docs chatbot built from a standard RAG tutorial; a CV-screener that just fine-tunes an open model (para 8, 12).
- Wrong entity entirely.A sole trader’s genuinely novel staircase joint — no limited company, so no route to claim, full stop.
- Effort mistaken for uncertainty.A year-long monolith→microservices AWS migration — “massive and difficult” describes scale, not a technological unknown (para 13).
The full run
All 51 cases, grouped by the band the engine returned. Trap cases are annotated.
| # | Sector | Project | Verdict |
|---|---|---|---|
| Qualifies · full narrative drafted (15) | |||
| 01 | Food & bev | Zero-added-sugar sauce, shelf life held | Likely |
| 02 | Manufacturing | 40%-lighter titanium aerospace bracket | Likely |
| 03 | Software / OR | Sub-2s streaming re-route, 3k drops / 200 vans | Likely |
| 04 | Biotech | Lateral-flow assay at 10× lower detection limit | Likely |
| 05 | Construction | 60%-cement-replacement low-carbon concrete | Likely |
| 06 | Agri-tech | Closed-loop vertical strawberry control | Likely |
| 07 | Fintech | Sub-10ms fraud scoring at 40k txns/sec | Likely |
| 08 | Chemicals | Low-temp-cure anti-microbial powder coating | Likely |
| 09 | Hardware / IoT | Adaptive BMS for second-life EV cells | Likely |
| 10 | Games | Real-time global illumination in 4ms mobile budget | Likely |
| 11 | Textiles | Wash-durable breathable flame-retardant finish | Likely |
| 12 | Software / ML | Clinical imaging model in 512MB, no GPU | Likely |
| 13 | Automotive | Rare-earth-free traction motor | Likely |
| 14 | Food & bev | Off-flavour-free shelf-stable alcohol-free beer | Likely |
| 15 | Software / infra | <50ms serialisable geo-distributed commits | Likely |
| Borderline · one settling question returned (17) | |||
| 16 | Software / AI | Review summariser + reply drafts | Possibly |
| 17 | Ecommerce | Rebuilt recommendation system for conversion | Possibly |
| 18 | Manufacturing | Biodegradable film on existing line | Possibly |
| 19 | Healthtech | Reconciling messy NHS data sources | Possibly |
| 20 | Construction | Faster modular housing system | Possibly |
| 21 | Legaltech | Contract clause-extraction pipeline | Possibly |
| 22 | Robotics | Mixed-item tote-picking reliability | Possibly |
| 23 | Energy | Smart solar / grid battery controller | Possibly |
| 24 | Food & bev | Less-crumbly gluten-free bread | Possibly |
| 25 | Software / infra | Higher-concurrency video pipeline | Possibly |
| 26 | Agri-tech | Drone-imagery crop-stress detection | Possibly |
| 27 | Hardware | Better-ANC smaller wireless earbud | Possibly |
| 28 | Fintech | Alt-data thin-file credit scoring | Possibly |
| 29 | Materials | Warp-free printing in a new polymer | Possibly |
| 30 | Cybersecurity | Lower-false-positive intrusion detection | Possibly |
| 47 | Medtech | Arrhythmia from noisy single-lead wearablehidden qualifier | Possibly |
| 50 | Cosmetics | Peptide serum, novel delivery systemscience, no stated uncertainty | Possibly |
| Does not qualify · failing test cited (19) | |||
| 31 | Agency | Shopify e-commerce site buildroutine web dev | Unlikely |
| 32 | Marketing | A/B pricing & messaging campaigncommercial | Unlikely |
| 33 | Integrator | Salesforce↔ERP via documented APIsconfig | Unlikely |
| 34 | Design | Drinks brand identity + bottleaesthetic | Unlikely |
| 35 | Software / AI | RAG-tutorial help-docs chatbotmodern but routine | Unlikely |
| 36 | Prof services | Zapier onboarding automationprocess / admin | Unlikely |
| 37 | Restaurant | New seasonal menu developmentroutine recipe | Unlikely |
| 38 | Construction | 60 low-energy homes, standard methodsknown methods | Unlikely |
| 39 | SaaS | Monolith→microservices AWS migrationmigration | Unlikely |
| 40 | Ecommerce | Subscription-box model iterationcommercial uncertainty | Unlikely |
| 41 | Manufacturing | New sofa range + second linecosmetic / scaling | Unlikely |
| 42 | HR tech | CV-screener, fine-tune wrapperno advance | Unlikely |
| 43 | Health | Physio programme + no-code appno tech uncertainty | Unlikely |
| 44 | Logistics | Off-the-shelf WMS + conveyor rolloutbuy & deploy | Unlikely |
| 45 | Software house | Six months bug-fixing & optimisationmaintenance | Unlikely |
| 46 | Renewables | Rooftop solar array designed to codeknown engineering | Unlikely |
| 48 | Fabrication | First-ever aluminium welding jobfirst-time-for-us | Unlikely |
| 49 | SaaS | Usage-based pricing engine on Stripecommercial-in-disguise | Unlikely |
| — | Retail | Sole-trader curved-staircase jointnot a limited company | Unlikely |
What this test is — and isn’t
It’s an internal benchmark: we wrote the scenarios and fixed the expected bands ourselves, against the published DSIT Guidelines. It is not a blind third-party audit, and scenario descriptions are cleaner than a real client’s first email. What it demonstrates is the thing that’s hardest to fake — consistent, correctly-cited banding across sectors, with every trap caught.
Think you have a case that would fool it? Run it free — that’s the honest test that matters.
R&D Radar is an informational screening tool. It flags and drafts; a competent professional and your accountant confirm and file.