Crisis research measures what a crisis does to an organisation's stakeholders and what the organisation's responses do in return. The field has two established bodies of work. Crisis communication research (Coombs, 2007; Coombs & Holladay, 2008) asks how much responsibility stakeholders attribute to the firm and which response matches that attribution. Reputation research (Rhee & Haunschild, 2006; Pfarrer et al., 2008; Zavyalova et al., 2012; Bundy & Pfarrer, 2015) asks how a firm's standing absorbs a shock and whether corrective action restores it. Both have mostly worked from data collected after the fact, in waves, which is why the timing of a crisis, the order in which judgments move and the moment a response takes effect, has been hard to study.

Measured as it happens

StakeholderWatch measures a crisis as it happens. The client's daily survey continues through the crisis without interruption, so the deterioration and the recovery are observed day by day rather than reconstructed. Structural break tests estimate and confirm the day each evaluation shifted: the judgment of the specific problem, the firm's character, its capability, and the intention to keep buying or working. The structural equation model estimates which signal moved each judgment and how much: the negative information itself, the firm's admission or accountability action, the verified fix, the campaign that followed. Peer brands measured on the same panel show how much of a fall belongs to the sector and how much to the firm.

What a combined score hides

Because the judgments are measured separately, the dashboard shows what a combined reputation score hides. An accountability action that is read as sincere can lower the judgment of the problem at the same time as it begins to repair character, because the cost of the action is also read as a measure of how serious the problem was. A technical fix can restore capability and the intention to buy while leaving character where it fell. A firm that sees these movements separately, and on the day they occur, can judge which of its responses is working and on which audience.

The research behind it

This is the subject of StakeholderWatch's research programme with VU Amsterdam, which studies how signals during a crisis are routed across stakeholders' judgments, how reputational wounds form, go dormant and reopen, and when the dynamics observed at one firm transfer to another. The founder's doctoral dissertation on reputation dynamics during organisational crises is scheduled for defence at VU Amsterdam in 2027.

What the client receives during a crisis is the daily dashboard with the break dates, the coefficients for each response, the peer comparison and the raw records, and, when the crisis is over, a dated account of what moved, when, and after what.

Our approach

Real-time, which means daily. Every study collects a fresh sample every day of the year at the same rate. A change appears in the data on the day it happens, and the estimation that runs every evening carries it onto the dashboard the next morning. Daily collection is the condition for causal estimation: a break can be dated to a day only if there is data for every day.

Real people, verified. We work with several panel bureaus selected against our verification standard, run our own detection on every complete, and remove failures before they reach the model. No large language models, no synthetic respondents, no synthetic sampling.

One world model. Every service is embedded in one behavioural model that holds the economy, the news, the issues of the day and the contact channels at once, so the service's own effects can be isolated from everything else that moves people.

Causal intelligence for every service. The structural equation model is re-estimated every evening and shows the causal effect of more than fifty factors, with an effect shown only when it reaches statistical significance.

Dedicated, never omnibus. Every study is designed for one client and fielded for that client alone. Respondents are not shared between clients.

Raw data, always. Every respondent record is the client's own and can be downloaded at any moment, with the codebook and weights. The coefficients on the dashboard can be reproduced from the raw data.