Why BuyerLens requires a RoastIQ result
BuyerLens never runs without a RoastIQ result. This is a deliberate architectural decision, not a limitation. Together the two form one pre-spend creative intelligence workflow: RoastIQ grades the ad before budget is committed, BuyerLens explains the resistance behind the grade.
Generic synthetic interviews, where you describe a product and ask AI personas what they think, produce generic responses. The personas have no specific creative to react to, no diagnostic data to ground their objections, and no evidence timeline to reference.
BuyerLens starts differently. It starts with evidence.
What context flows from RoastIQ
When a BuyerLens study opens from a RoastIQ result, the following context is passed automatically:
- The 5 KPI scores and the verdict. The persona knows the creative scored 71 with a Sharpen verdict
- The weak KPI and its score. The persona knows Sell Proposition scored 53
- The evidence timeline. The persona knows the product pour arrives at 18 seconds
- The platform and category. The persona knows this is an Instagram Reels ad for FMCG in the UK
This context shapes every interview question and every response.
How context changes interview quality
Without context, a generic question like "What do you think of this ad?" produces a generic answer: "It looks nice but I would not buy it."
With RoastIQ context, the same persona answers differently, because the interview script itself is different. The question becomes: "The product benefit does not appear until 18 seconds. As an FMCG buyer on Instagram Reels, at what point did you consider scrolling past?"
The response is specific: "By second 5. The visuals are strong but there is no personal hook before the halfway point. I have already decided whether I am interested by then."
The difference is not in the AI model. It is in the context.
The study design flow
BuyerLens uses three setup steps, each informed by RoastIQ:
- Audience. Pre-filled from the RoastIQ category, platform, and region. Editable
- Participants. The user chooses 1 to 15 synthetic buyers, drawn from a directory of 36 buyer personas. Personality traits (OCEAN) shape response patterns
- Goal. Pre-filled from the weak KPI evidence. Can be customized or inferred from RoastIQ
The "Infer from RoastIQ" button on both Audience and Goal auto-fills from the diagnostic data. The user can accept, edit, or replace these defaults.
Source enrichment
For sharper interviews, users can upload source documents: creative briefs, past research, customer reviews, landing pages. These documents are retrieved at response time, grounding the synthetic buyer in the specific brand context.
Without sources, BuyerLens gives directional objections. With sources, the objections reference the brand's actual positioning, competitive context, and audience expectations.
What this is, and what it is not
BuyerLens is scenario simulation: structured synthetic buyer interviews, not a consumer panel. The personas are synthetic, and their reactions are not survey responses from real buyers.
What the RoastIQ context buys you is not realism, it is relevance. A grounded synthetic objection points at a specific, checkable thing in your creative, which is something you can act on or dismiss on the evidence. An ungrounded one just produces plausible-sounding text. Either way, the output is a directional hypothesis to verify, not evidence about real consumers.
The RoastIQ scores anchoring each study are model predictions, validated against public engagement and click-intent outcomes, not ROAS or sales (held-out Spearman ρ +0.30 to +0.32, May 2026): see the methodology.
That is why BuyerLens exists as the second layer after RoastIQ rather than as a standalone tool.
For a tool-agnostic version of this workflow, see synthetic users for ad testing.