2027 → 2035

Future Demand Radar 2035

Which productivity problems will people pay to solve in 2028, 2030, 2032 and 2035? This radar scores twelve markets with keyword data, Google Trends, labour-market evidence and explicit expert assumptions. Nothing here is a prediction presented as fact.

Build nowValidate in 12 monthsMonitorDeprioritize

How to read the radar

  • Rings are the market-entry classes: Build now (score ≥ 55), Validate in 12 months (48–54), Monitor (40–47), Deprioritize (< 40). The thresholds were calibrated on the first model run (November 2026), because eight components averaged on a 0–1 scale naturally cluster scores between 35 and 70; they are stored with every run. A category closer to the centre is more attractive.
  • Dot size shows current US search demand for the category's keyword set (log scale).
  • The bar through each dot is the 10th–90th percentile of the Monte Carlo score distribution. Long bars mean we do not know much yet.
  • Scenarios change the demand-growth assumptions: A is conservative (AI adoption gradual), B baseline (copilots and agents common in knowledge work), C accelerated (autonomous systems reorganise work). Scenario B is our planning case.
  • The year slider changes only the dot size, which shows the modelled demand multiplier for that year. A dot's distance from the centre is always the scenario's opportunity score, the same number that sets its ring and colour.

The score

Future Opportunity Score = 20 % expected demand growth + 10 % current demand + 15 % commercial purchase intent + 15 % monetization potential + 10 % SEO ranking feasibility + 10 % product differentiation potential + 10 % recurring-revenue potential + 10 % long-term durability, minus risk penalties for forecast uncertainty, platform dependency, regulatory constraints, commoditization, AI substitution and saturation.

Data-derived components (current demand, half of growth, half of purchase intent, SEO feasibility) come from DataForSEO keyword volumes, three-year growth rates, keyword difficulty, cost-per-click and intent classification, plus Google Trends twelve-month ratios. Expert components (monetization, differentiation, recurring potential, durability, the other half of growth and intent) are assumptions grounded in the evidence base, listed per category under Key assumptions. Each run samples every component 4 000 times with wider noise for weaker evidence to produce the bands.

What the radar is not

It does not forecast revenue. It does not tell us an article will rank. It is a prioritisation device. Every run is logged so the error can be published later; re-scoring and the comparison with observed demand are done by hand for now, not on a fixed schedule. Alternative weightings (growth-heavy, cash-now, defensive) are stored with each run; when they disagree with the baseline, the category is flagged for discussion rather than built.

Scenario assumptions

A — Conservative. Generative AI stays an assistant feature inside existing suites. Search demand for AI-agent topics grows but stays a fraction of classic productivity demand; attention and burnout topics keep growing with remote and hybrid work.

B — Baseline. AI copilots and delegated agents become routine for knowledge workers by 2028–2030. New demand forms around orchestrating agents, verifying their output, personal knowledge systems that feed them, and protecting attention from an even noisier environment.

C — Accelerated. Highly capable agents absorb much routine knowledge work. Individual productivity becomes a question of judgement, orchestration and recovery; software-comparison demand shifts from "best app" to "which agent/platform", and the price of attention rises.

See the evidence base and the method in full in the methodology.