The Listening Score — what cities sound like when they're heard

Every city says it listens to its citizens. Almost none can prove it.^[4]

That gap — between the press release and the lived experience — is what The Listening Score tries to close. It's one of the nine dimensions inside The Pulse, Cittopia's composite urban indicator, and it's the only one that isn't built on top of an existing Eurostat or city open-data feed. We built it ourselves because nothing comparable existed.

This piece walks through what it actually measures, why we landed on five components instead of one, what it gets wrong in v0.1, and why we're publishing the formula on a public page^[3] in the first place.

The question the score is trying to answer

Imagine you're a citizen of Warsaw or İstanbul or Sofia. You file a comment on a public consultation about a new tram line. What happens next?

Three things, usually. In the worst case, nothing — the comment vanishes into a PDF nobody reads. In the middle case, you get an automated thank-you email and the city publishes a summary report six months later that may or may not mention the issue you raised. In the best case, the project plan visibly changes, the change is attributed to citizen input, and you're notified that your specific concern was either addressed or — and this matters — explicitly rejected with a reason.

Most participation metrics in the literature measure inputs^[5][1]: number of submissions, hours of public meetings, sign-ups to a platform. Those are useful as supply-side indicators — they tell you whether the channels exist. They tell you nothing about whether the channels are plumbed to anything. The Listening Score is an attempt to measure plumbing.

The formula — five components, each capped at 20

Each of the five components produces a 0–20 score. The Listening Score is the sum: 0–100.

The formula deliberately weights response, loop-closure, and decision trace equal to reach and diversity. A city with brilliant turnout but zero loop-closure caps at 60. A city with modest turnout but perfect loop-closure can outscore it.

That's not an accident. It's the central methodological claim of the score: listening is a back-end discipline, not a front-end one. The hard part isn't the form. The hard part is the response that follows the form.

A city with brilliant turnout but zero loop-closure caps at 60. A city with modest turnout but perfect loop-closure can outscore it.

— Listening Score design note, Cittopia methodology v0.1

What good looks like

In the methodology white paper we anchor the bands with rough references:

• 80–100 — Hearing-organised. The city has named a directorate, published a participation playbook, and routinely cites citizen input in formal decisions. The clearest examples we've found in Europe are in Helsinki, Madrid (during the Decide Madrid era), and parts of Brussels^[6][2].
• 60–79 — Hearing-capable. The infrastructure exists, several channels work, but loop-closure is patchy and decision-trace is rare. Most well-resourced European capitals sit here.
• 40–59 — Hearing-aspirational. The city talks about participation, runs occasional flagship consultations, but day-to-day listening isn't institutionalised.
• 20–39 — Hearing-symbolic. Participation is performative — a once-a-year survey, a single public meeting per project.
• 0–19 — Not listening. No reliable channels, no published follow-ups, no traceable decisions citing input.

These bands aren't grades. They're an attempt to make the score readable to a non-specialist in 30 seconds. A city official who scores 47 should be able to tell, immediately, what gap closes the most ground for the next score band.

What v0.1 gets wrong (and we know it)

The score is in version 0.1 for a reason. Five honest limitations:

1. Small sample bias. A city that runs three deep consultations with rich loop-closure can outscore a city that runs three hundred shallow ones. That's the right direction philosophically — but it means small cities with active mayors can game the score upward against larger cities with thinner-per-capita engagement.

2. Digital-channel bias. Most of the data we can collect at scale is digital. We use a coarse adjustment for analogue participation (in-person meetings, town halls, paper submissions) but it's an estimate. Cities with strong analogue cultures (parts of Italy, rural Poland) score lower than they probably deserve.

3. Language asymmetry. We currently parse follow-up documents in English, Polish, Turkish, German, French, Italian, Spanish, and Dutch. Cities publishing primarily in less-represented EU languages get a manual review fallback, which is slower and less consistent.

4. "Citizen" is ambiguous. Are non-citizen residents included? Tourists? People who work in the city but live elsewhere? Different consultations define this differently. We follow each consultation's own definition. The aggregate score therefore mixes apples and oranges in a way we explicitly disclose per city.

5. Loop-closure is the hardest component to verify. "Did the city actually use the input?" requires reading the follow-up document and judging intent. We do this with a hybrid pipeline (rule-based extraction plus human review on edge cases), but a city that publishes elaborate-looking but vacuous follow-ups can score above its true position.

Why this exists at all

Eurostat publishes a "trust in local authorities" survey. The OECD publishes Better Life Index participation data. SDSN tracks SDG 16 indicators. All of them are useful and we use several as input feeds.

None of them tell you whether the city responds. They tell you whether citizens feel listened to, which is a different (and worthwhile) measurement. The Listening Score is the back-end-of-house equivalent: not what citizens feel, but what the city demonstrably does with what citizens say.

The two correlate strongly in cities we have full data for. They diverge — usually — in two patterns:

• Trust > Listening. The city has a strong reputation for participation but the loop-closure data is thin. Often a sign of legacy reputation from a prior administration.
• Listening > Trust. The city does the back-end work but doesn't communicate it well. Often a sign of an opportunity — better PR could shift the trust survey upward inside 18 months without changing any underlying behaviour.

Either pattern is useful information for a mayor's office or a civic-tech NGO. The first says "your reputation will erode if you don't fix the plumbing." The second says "your plumbing is good, your storytelling is broken."

Where it goes next

v0.2 of the methodology, due Q3 2026, will:

• Add weighting by consultation type (a budget vote weighs more than a public space naming poll)
• Add a manual override channel so cities can flag follow-up documents we missed
• Triangulate Loop-closure scoring against a structured external panel (academics + civic-tech practitioners) on 10 randomly-sampled cities per quarter
• Publish per-component time-series for each city, so a 12-month listening trajectory becomes visible
• Translate the formula and bands into Polish and Turkish (Phase 1 of our internationalisation plan)

We're also open about which cities currently score where, and why — that publication is part of the September data-study release on the 30 European capitals comparison. If you're a researcher, a journalist, or a city official who wants to interrogate the formula before it goes wide: methodology@cittopia.com.

The point of publishing this now

Every metric is also a political choice. By publishing The Listening Score formula on a public page — with its components, its bands, its bias disclosures, and its limitations — we're inviting critique earlier than is comfortable.

That's the price of an open methodology. The alternative is to ship the score, claim it works, and then defend the claim. We'd rather ship the score, name its weaknesses, and let the people who'd notice find them.

If you spot one, the inbox is open.