How we do the research.
Civico Method v1: a startup-led technology and use-case radar, built on published foresight and technology-intelligence methods. Every score can be recomputed from the files you can download here.
The short version
- Scan. Each week we collect sourced deals and signals (policy, tenders, partnerships, launches, market data) for Southeast Asia and India. Every row links to a source we opened.
- Define the space. A list of technologies and use cases, each with keywords, is matched to that evidence. We review the matches by hand.
- Score. Attractiveness is computed from the evidence. Addressability, maturity and time horizon are analyst judgements, written down with a reason.
- Test. Each ranking is recomputed under three different attractiveness measures. If the position moves a lot, we say so.
- Decide. The result is a call per item: Act, Watch or Ignore. It is a hypothesis to test with other information, not an instruction.
Current dataset: 28 deals and 20 signals between 25 Aug 2026 and 10 Oct 2026 (6.6 weeks), 25 technologies and 25 use cases. Momentum and rankings are indicative until we hold at least eight weeks, and become more reliable each week.
Why startups, not patents, are the primary signal
The patent-analysis literature we studied was built for technology fields where firms patent heavily, such as machine tools, chemicals and nutrition. Software, data and service-led urban technology is different: a lot of it reaches the market without patents, so patent counts under-read it. Mogee's own study notes that not all inventions are patented, and that patents should confirm or question hypotheses formed from other information. In Asia's built world, funded startups, tenders and policy moves are where technology becomes visible first, so they are our primary evidence.
Patents stay in the method as a secondary layer for hardware-heavy technologies (batteries, power electronics, materials). That layer is planned for a later version and is not part of any score today.
What we took from each source
| Source | Idea | How Civico uses it |
|---|---|---|
| Rohrbeck Heger, Foresight Radars: Radar Framework & Methodology (2020) | A radar has segments, a radius for time horizon, and content. Opportunities are assessed on attractiveness and addressability, giving Act, Watch or Ignore. | Segments are urban technology areas. Rings are 0–3, 3–6 and 6–10 years to relevance. The portfolio view plots attractiveness against addressability. |
| Ernst, Patent portfolios for strategic R&D planning (1998) | Technology attractiveness from growth in activity, including recent change against earlier change. Objective measures validated by expert judgement. Rankings are sensitive to the measure, so test several. | Momentum compares recent weekly activity with earlier activity. Rankings are recomputed under three measures and unstable ones are flagged. |
| Fabry, Ernst, Langholz and Köster, Patent portfolio analysis… (2006) | Strength is activity multiplied by quality. Keep the dataset in a workable size by splitting a broad area into technology fields. | Attractiveness blends evidence count with capital behind it, not counts alone. Technologies are split into focused items instead of one broad field. |
| Mogee, Using patent data for technology analysis and planning (1991) | Life-cycle stage read from activity and concentration among players. Validate results against expert opinion. | We report capital concentration for each item. Analyst maturity ratings are reviewed against the evidence each run. |
| CambridgeIP, IP Landscape: Wave & Tidal Energy (2011, sample report) | Define the technology space with iterative searches, then clean false positives by expert check. Show top players, geography and the newest activity. | Keyword sets are tuned and the match list for each item is checked by hand. Heatmap shows geography. Timeline shows the newest activity. |
The scores
Attractiveness (0–100, computed)
The balanced measure is 35% disclosed capital (log scale, all deal types), 30% evidence count (deals plus signals), 15% breadth (number of countries) and 20% momentum. Each part is scaled against the highest item in the set. Two alternative measures are used only for the stability test: capital alone, and activity (count and momentum) alone.
Momentum
Evidence in the latest four weeks per week, divided by earlier evidence per week, with a small smoothing term so one item never divides by zero. Above 1 means rising. We show the raw recent and earlier counts next to it.
Addressability (1–5, analyst judgement)
How realistic it is for a Civico client (an asset owner, agency, corporate or software vendor) to do something about the item within its time horizon. Each rating carries a one-sentence reason and a review date.
Maturity and time horizon (analyst judgement)
Maturity runs from Concept, Pilot, Early commercial, Scaling to Mainstream. Time horizon is 0–3, 3–6 or 6–10 years to relevance for the region. These adapt the maturity and market-readiness scales in the Rohrbeck Heger framework to a startup-led view.
The call
| Call | Rule |
|---|---|
| Act | Attractiveness 50 or more, and addressability 3 or more. |
| Ignore | Attractiveness under 25 and addressability under 3, or no evidence at all. |
| Watch | Everything else, including attractive items that are hard to address. |
Confidence is Higher with five or more pieces of evidence, Medium with three or four, and Low below that. Low-confidence items are drawn faded on the radar.
Core tier and the adoption score (Method v1.1)
Mature technologies, such as BIM, common data environments, e-procurement and smart metering, do not show up in startup funding, so scoring them on capital would be wrong. They sit in an inner Core ring and get an adoption score from 0 to 100 instead: 30% geographic breadth (countries with evidence, out of seven), 25% depth (datapoints, capped at eight), 25% institutional share (mandates, tenders, standards and national rollouts), and 20% recency (evidence from the last 24 months). A score of 65 or more with an analyst maturity of 4 or higher is called Established; 45 to 64 is Spreading; below 45 is Patchy. Core items get no Act, Watch or Ignore call, and they do not change any emerging score. Adoption evidence lives in adoption.csv, every row with a source link.
Landscape: coverage and attack (first pass)
The Landscape page maps what incumbent software vendors sell against what Asian startups build. We fixed a taxonomy of 79 capabilities along the project lifecycle (plan, design, procure, build, operate, renew, plus cross-cutting). Each incumbent product, taken from the vendor's own pages, is mapped to one primary capability. Each Asian startup is mapped to one or two. A capability is then placed by how many global vendors sell in it (none, 1 to 2, 3 to 5, 6 or more) and how many Asian startups were found (none, 1 to 2, 3 or more). White space means few or no incumbents and no startups found; Startup frontier means startups with thin incumbent cover; Incumbent-held and Contested mean incumbents are present, without or with startups.
- The startup side is a first pass. A white-space cell means we found no Asian startup yet, not that none exists. Each weekly run adds startups, and the page says how many are in.
- Incumbent coverage is a floor. Some vendor sites block automated reading, so a vendor can be under-counted. Products named only inside a parent product page share its link.
- Mapping is judgement. A product fits one primary capability even when it spans several. The taxonomy will change as products fail to fit; changes are logged.
- AI flags record only what the vendor page we read says. A blank means not seen, not no.
What this method cannot tell you
- Coverage is incomplete. We use public announcements and media. Some Southeast Asian deals sit behind paywalls or are never announced.
- Short history. With a few weeks of data, momentum swings on a single item. Treat early rankings as a starting point.
- Keyword matching is blunt. An item can pick up a loosely related story. We review match lists and tighten keywords when we find one.
- Capital is not proof of value. A large round or acquisition raises attractiveness, but says nothing about whether the technology works.
- Judgements are ours. Addressability, maturity and horizon are analyst calls. We publish the reasons so you can disagree with them.
Update rhythm and corrections
The database is updated weekly and the radar recomputes with it. A weekly Urban Tech Pulse reports what moved, a monthly deep-dive covers one theme in depth, and a quarterly report assembles the full method, data and findings. If we got something wrong, tell us through the network page, with a source, and we will correct it in the next run.
Data you can check
- deals.csv: every deal, with source link
- signals.csv: policy, tender, partnership, launch and market-data signals
- radar.csv: the items, keywords and analyst judgements
- adoption.csv: adoption evidence for the Core tier
- taxonomy.csv, products.csv, startups.csv: the landscape database