Data Study and Linkable Asset Campaign, German Fieldwork

A German business desk treats a survey as national at n=2,000 with quotas on age, gender and Bundesland. That is plus or minus 2.2 points at 95% confidence, and it is the point below which an editor starts asking questions you cannot answer. Most studies sold as link assets are n=500, unquoted, fielded over a weekend, and they get picked up by aggregators and by nobody who matters. The sampling: Fieldwork through Bilendi and respondi, Appinio or YouGov Deutschland depending on the quota and the deadline. We write the questionnaire, they field it, we work from the raw file. Screening questions come before the interesting ones so the base for every chart can be stated honestly. Screen-out and non-response rates go into the methodology, not into a footnote we hope nobody opens. Where a question is behavioural rather than attitudinal we say so, because people are unreliable about what they returned last month and reliable about what they bought. Where the data comes from when it is not a survey: Your own order table, aggregated past the point of re-identification, with a DPA signed before anything is exported. Public price observation collected on a fixed schedule against a fixed basket. Destatis and Eurostat series for the denominator. Nothing scraped from behind a competitor login, nothing bought from a broker whose provenance cannot be traced back to consent. What the asset actually is: One number a journalist can put in a headline and defend to their editor, the chart that proves it, and a methodology page at a permanent URL that will still resolve in three years. A CSV of the tabulations, downloadable, licensed so a newsroom can redraw the chart. Chart images at print resolution with the source line set into the file. An embargo pack holding the fact sheet, two pre-cleared quotes and a contact for follow-up questions. Interactive maps and calculators are usually the wrong asset: hard to quote, expensive to maintain, and they rot. What we refuse to publish: An index that happens to place you first. If the analysis ranks brands and you come out on top, we either drop the ranking or exclude you from it and say so in the method. Findings we cannot show the workings for. A second wave that contradicts the first and gets quietly shelved. Waves publish whichever way they go, and the year-two number is what turns a study into a citation instead of a news item. Launch: The study goes out through the same pitching method as our press work, but the asset is built for the nine months after launch week. Studies that work accumulate. The links that arrive in month nine come from people writing about the subject, not about you. Not for you if: You need to approve the finding. We will not run a study whose conclusion is agreed in advance and we will not soften a result because it is awkward for the brand. If your legal team needs sign-off on the interpretation as well as the facts, this will fail and we should not start it. Also not for you if the question is small. A study needs a question the category argues about. How many Germans shop online is not it. How much of the returns rate is size inconsistency rather than buyer's remorse, and which categories are worst, is. Timing: Six to nine weeks from brief to embargo where fieldwork is commissioned, most of it spent in questionnaire design and cleaning. Desk-research indexes built on existing public series run four to five weeks and are usually the right first study for a brand that has never published one.