Roadmap · Updated May 2026
What's in the offer today. What we're evaluating next. What we're deliberately not building.
A working research lab evolves its deliverables as the engines move and as the first cohort of Residency partners tell us what they actually use. This page is the honest version of that evolution — not a feature-promise marketing page. If something here is "evaluated" rather than "shipping," it means we haven't decided whether to build it and we want your input before we do.
Shipping today
Three deliverables. Each one earns its place in the engagement.
Every Residency partner gets these. They are the minimum coherent research engagement — below this set, the work doesn't compound; above it, complexity outpaces our ability to deliver well.
Your domain's citation report
Which of your pages were cited, on which engines, for which queries, with what persistence rate. Per-engine breakdown. Updated weekly. Your raw data is yours.
Prioritized recommendations queue
The specific pages and articles most worth updating right now, ranked by expected citation impact. Each recommendation cites the underlying research finding and shows the data behind the call.
Competitive citation surveillance
Weekly tracking of five named competitors' citation share in your category. When their share moves — up or down — you get the specific pages driving it. No quarterly waiting. Engine velocity matched.
Evaluating · not committed
Two deliverables under consideration. We want feedback before we ship either.
These are real candidates for the next addition to the offer. We've thought through the value, the failure modes, and the operational cost. We have not committed to building them. If you're a current or prospective partner and one of these would change your buying decision, tell us — that's the input we're missing.
Quarterly longitudinal write-ups
Two deep reports per six-month cycle covering not your domain specifically, but your category. What happened across the AI search surface in this quarter that affects how your buyers find sources like yours. Which named competitors gained or lost citation share at the category level. How the engines themselves drifted in behaviour.
Why we're considering it: the cadence matches the CMO's board-update rhythm rather than the content team's publication cadence — two distinct buyer reads of value from the same engagement. It also becomes the structured research narrative that feeds The Half-Life Newsletter, case studies, and our published methodology.
Why we haven't shipped it: if you're acting on the weekly recommendations queue, a quarterly retrospective risks being a curiosity rather than a thing you act on. It might be the wrong unit of analysis for an engine surface that changes weekly.
Engine update intelligence briefs
Irregular short reports — published whenever Claude, ChatGPT, or Perplexity ship a detectable behaviour change across our broader corpus. What shifted. What kinds of content gained or lost. What we'd do about it. No fixed cadence by design: when nothing changes, no brief; when an engine ships a major update, partners hear within days.
Why we're considering it: it's the deliverable that most directly matches the velocity of the engines themselves. Today, when an engine ships a new behaviour, no one tells you — you discover it via citation drops weeks later. Engine update briefs close that loop.
Why we haven't shipped it: irregular deliverables are operationally harder to scope contractually. "You'll receive briefs when relevant" is honest but doesn't fit easily into a standard SLA. We're working through whether the value justifies the contractual complexity.
Deliberately not building
One thing we've decided against shipping, and why.
Rigorous-looking experimentation reports
We considered — and rejected — a deliverable that would run formal A/B-style experiments on a client's pages and publish significance-tested results. A peer in the AEO space could probably ship something that looks like this within a month.
Why we're not shipping it: the math doesn't work. Rigorous experimentation requires variable isolation (one change per page), pre-registered hypotheses, and enough time for stable measurement — typically four to six weeks per test. Real content teams ship multiple changes per page per week. The engines themselves change underneath the experiment. Any "experimentation report" we published would either (a) measure pseudo-experiments confounded by rapid implementation and engine drift, or (b) constrain client teams to a measurement-friendly cadence that's operationally untenable.
Publishing pseudo-rigorous reports that mostly say "results inconclusive due to confounded design" would be worse than no report. The brand promise is honest research — rigour-theatre violates that. We are open to revisiting this if a partner has a specific, low-velocity content surface where formal experimentation is genuinely possible.
How to influence this
If you'd value something above over something we're already shipping, that's the input we're missing.
The Residency cohort is small on purpose. Your trade-off preferences directly shape which of the evaluated deliverables we commit to next. The fastest way to influence it: an email naming what you'd buy, what you'd skip, and why. Two sentences is enough.
— Dan Russell, founder of Quoted