Public Sector
Social intelligence for government and not-for-profit
How a contested issue is being argued, by the people arguing it.
A poll tells you how many people agree with a statement. We read the conversation first, to find the statements people make themselves, so the poll can ask about them.
We read the public conversation around an election, a policy or a contested issue, and map the narratives, frames and divides that organise it. Our public studies of the French election, migration narratives across the EU and the Hungarian election were made this way.
Where we start
Questions we are brought
- Which narratives are carrying this debate, and where do they come from?
- Who is in the middle, and what language loses them?
- How does the same issue read from one country and language to the next?
- Is disinformation moving, and through which sources?
- What should the survey ask, before it is fielded?
Human analysts read the material. The Reading Machine holds the whole conversation at once, so nothing is left out because it did not look interesting. Interpretation stays with the researcher. How the work is done
From the Atlas of the Unexpected
What came up when nobody was asking
Cases
9 entries
Public sphere, Europe
The Hungarian Election, 2026
See the study: The Hungarian Election, 2026Expected: the two camps differ in what they want.
Found: Tisza supporters wrote from personal conviction far more often than Fidesz supporters, a gap of 12.4 points.
The question: does a campaign message built on personal agency reach the other camp at all?
A polarised electorate is not divided by what people want. It is divided by two opposing psychological architectures — the Employee, who locates safety inside the system, and the Entrepreneur, who locates it in personal agency. Identity comes first. Politics follows.
Public sphere, Europe
Amsterdam's Ad Ban: a meat debate in disguise
Expected: a ban on meat and fossil-fuel advertising would be argued as a free-speech question.
Found: people argued almost only about meat, and their side followed their diet.
The question: does dietary identity decide how a food policy is received, more than politics does?
On May 1, Amsterdam introduced an ad ban on meat and fossil-fuel products. We explored the reactions of social media users across various channels. The language was English.
What we found is that the discourse predominantly focuses on the meat part of the ban, and to a significantly lesser extent on the fossil-fuel part. The fossil-fuel half is largely waved through as common sense. The meat half is where the conversation concentrates.
By and large, opinions fall into two opposing sides — and which side people land on tends to track whether they're an omnivore or follow a meat-free lifestyle. In this discourse, diet predicts stance more reliably than politics.
Supporters celebrate the policy. They expect that reducing the visibility of meat will normalise vegetarian and vegan options over time, underpinned by a belief that less advertising means meat stops being the default. Other supportive arguments include retribution for the environmental damage attributed to the meat industry, and a smaller group who back the ban because they think fewer ads would have a calming, aesthetic effect on the quality of city life. The stance is anchored in a self-identity as environmentally conscious and forward-looking.
Opponents read the ban as an attack on a normal, everyday way of life — elitist, performative, and a step toward broader censorship. A recurring move is whataboutism: gambling ads are permitted, but meat is where the line is drawn? The register here is largely ridicule and sarcasm rather than evidence, while the supportive side tends toward research, precedent, and moral framing. As a result, the two camps rarely engage directly.
Beneath the meat divide sit two deeper questions that also separate the camps: whether advertising actually changes behaviour, and whether a collective authority should decide what may be promoted. The fossil-fuel half of the ban functions as a useful point of comparison — the same policy, the same city, and the same censorship objection available, yet only meat generates sustained debate. In this discourse, the stronger driver of reaction is dietary identity rather than free-speech principle.
Public sphere, NATO member states
NATO perception dashboard
- 30
- member states
- 14
- months
With the McCain Institute at Arizona State University we built and kept up to date an interactive dashboard of how NATO is perceived in public conversation, and how each of its member states is targeted by misinformation. The project was part-funded by the US Department of State.
The dashboard held country-by-country data, so that awareness of the alliance and of the threats aimed at it could be compared across all of its members.
See it at the McCain InstituteCountries
- Belgium
- Albania
- Bulgaria
- Canada
- Croatia
- Czech Republic
- Denmark
- Estonia
- France
- Germany
- Greece
- Hungary
- Iceland
- Italy
- Latvia
- Lithuania
- Luxembourg
- Montenegro
- Netherlands
- North Macedonia
- Norway
- Poland
- Portugal
- Romania
- Slovakia
- Slovenia
- Spain
- Turkey
- United Kingdom
- United States
Public sector, client undisclosed
Migrants in transit, Turkey and Tunisia
- 2
- transit countries
- 5
- countries of origin
- 24
- months of conversation
We read the public social media conversation of people in transit towards Europe, and of the communities that host them, in their own languages and with native analysts. The study followed five groups from their countries of origin through two transit countries.
The reading came before the survey research, so that the questions could build on what people had already said without being asked.
Countries
- Turkey
- Tunisia
- Syria
- Iraq
- Afghanistan
- Nigeria
- Somalia
Public sphere, South Caucasus
Public participation, Azerbaijan
- 24
- months of conversation
- 3
- languages
- 10
- issues
For a USAID programme in Azerbaijan we read two years of public conversation in Azeri, Russian and English, with a team of local analysts. The reading covered ten social and economic issues, and looked at what helps or hinders people in taking part in public life, among them women, young people and people with disabilities.
It fed the analysis the programme was designed on. The report itself is not public.
Public sphere, Central Asia
Civil society, Kazakhstan
- 24
- months of conversation
- 2
- languages
We read two years of public conversation in Kazakhstan, in Russian and Kazakh, about civil society: what people take the term to mean, and how they discuss the organisations that work under it.
The reading was made to be used together with survey research, in support of NGOs’ part in public debate.

Public sphere, Europe
Poland’s school meal reform: listening first, then asking
- 10,000+
- posts read
- 1,017
- Poles asked
- 52 / 18
- support / oppose
Read the case study: Poland’s school meal reform: listening first, then askingOnline, the opposition to plant-based school meals sounded like one voice. Read closely it was five objections, and under all of them something nobody states as a position: a shared shame about the food we give our children. Asked directly, only 14% call the food poor. Asked about the objections, the majority agrees with every one, and backs the change three to one.

Public sphere, Europe
The French Election, 2017
- 8M
- shared links
- 1,000+
- media sources
- 17
- clusters
See the study: The French Election, 2017Around the 2017 election, French social media had virtually no common ground. People shared links from one side of the media map or the other, and almost never both. Fake news travelled because it answered an emotional need, not because it was believed.
Public sphere, 28 EU member states
Migration Narratives in Europe
- 28
- EU member states
- 5
- narrative frames
Across 28 EU countries, migration is argued through the same few frames: security, identity, economy and demographics, humanitarianism, and distrust of the political establishment. What changes from country to country is their weight.
See the study: Migration Narratives in EuropeCountries
- Germany
- Austria
- Belgium
- Bulgaria
- Croatia
- Cyprus
- Czech Republic
- Denmark
- Estonia
- Finland
- France
- Greece
- Hungary
- Ireland
- Italy
- Latvia
- Lithuania
- Luxembourg
- Malta
- Netherlands
- Poland
- Portugal
- Romania
- Slovakia
- Slovenia
- Spain
- Sweden
- United Kingdom
How to work with us
- Agenda
- What are people talking about right now?
- 48 hoursto first learnings
- Frames
- Which stories do people tell about it, and where do they disagree?
- 1 weekto first learnings
- Publics
- Who are these people, and where do we stand with them?
- 2 weeksto first learnings
- Common Sense
- What do people take for granted about this category?
- 2 weeksto first learnings
- Public Mood
- What changed since last month?
- Every monthto first learnings
Among those we have worked with: Open Society Foundations, Friedrich Ebert Stiftung, Ofcom, Pew Research Center, Parliament of Victoria, McCain Institute / Arizona State University, Greenpeace.