In short: The reporting cascade in a dealer network, head office, country, region, outlet, is manual work at most manufacturers: one slide deck per level, built from an Excel file, delivered months after the survey. The way out is a different dataset: one for the whole network, in which country, region and outlet are variables, with access profiles per level, a minimum case count and exports that are filters rather than projects. This article describes the pattern step by step.
The problem has four levels
A manufacturer with 400 outlets in eight countries surveys its dealers once a year and end customers after every workshop visit, continuously. Four levels want to see something different from both surveys. Head office wants the country comparison and the drivers. The country organisation wants its regions side by side and its place in the country ranking. The regional manager wants their outlets and the regional average. The outlet wants itself and the comparison with the region, nothing else.
In the Excel world that produces four report types, times 8 countries, times 40 regions, times 400 outlets. Whoever has built that once does not enjoy building it again next year, which is why the outlet gets its report in March, if at all.
The pattern: one dataset, four views
The core of the pattern is that there is only one dataset and no reports. Each level gets a view of the same dataset, defined by a filter. Five building blocks make that possible.
1. The hierarchy enters the dataset through the link
Every outlet gets its own survey link with parameters, for example ?country=at®ion=west&outlet=4711. DataLion stores the three values as separate variables with every response. The questionnaire does not ask for them, and yet the complete hierarchy sits in the dataset. For the end-customer service feedback the same outlet parameter goes on the QR code on the invoice; that places the dealer's view and the customer's view side by side per outlet.
You generate the link list once from the dealer master: a CSV with outlet number, region and country from which the links are built. Maintain it in one place; an outlet that changes region changes it in the list and nowhere else.
2. Access profiles with a fixed filter per level
An access profile in DataLion is a view of the project with a hard-wired filter. The profile "Region West Austria" shows the dashboard pre-filtered to country equals AT and region equals West; the user can filter within that view but not out of it. The filter is enforced server-side: anyone not allowed to see the region never receives its data, not even in an export.
For 40 regions you need 40 profiles; that is an hour of work, once. For 400 outlets you do not need 400 logins, see block 4.
3. One minimum case count that applies everywhere
At outlet level the service feedback gets thin quickly: a small outlet has twelve responses a month, and one of them is recognisable. Set a minimum case count in the project, and cells below it stay empty in every table and every export. The rule then applies to head office just as it does to the outlet, and nobody has to apply it by hand in a report.
For the dealer survey itself the rule applies per region, because there the outlet is usually not captured for anonymity reasons; see the article on dealer satisfaction and the DSI.
4. The report per outlet is an export with a filter
The 400 outlet reports are 400 exports of the same dashboard with the outlet filter applied, as PowerPoint in the brand's layout or as PDF. What the region sends out monthly is thus one click per outlet, not a project. For outlets that should see the report continuously there is a second route: a QR code that opens exactly one report through an access profile, with an expiry date and revocable. The outlet needs no login for that.
5. Waves and time sit in the same dataset
The dealer survey gets a wave field; the service feedback has the timestamp anyway. Both time series run in the same project: the DSI per region across years and the CSI per outlet across weeks. Whether the new bonus scheme changed anything is shown by the line in the following wave; whether an outlet has slipped since the service manager changed is shown by the weekly line the month after.
Eight countries, eight languages, one dataset
You set up the question texts as one survey per language with identical variables and combine the datasets for analysis. The questionnaire controls appear automatically in six languages; the question texts you translate yourself or bring into the project with AI translation. A single multilingual questionnaire is not possible at present; the discipline of using the same template in every country replaces it.
For analysis, AI translation renders the whole project, codebook, charts and reports, as an additional language layer. The country organisation in Poland sees the same dashboard as head office, in Polish, with the same filter profile.
Sales data alongside
Whether dissatisfied dealers sell less becomes a crosstab as soon as the sales figures per outlet sit as a second data source in the same project, joined on the outlet number. The sales data comes in via Excel, via API or as a scheduled import from a database. It stays a separate source next to the survey dataset and is joined on the attribute; people play no part in the join.
Where the pattern snags in practice
- Parameter hygiene. An outlet whose link says "region=west" while the list says "West" lands in a region of its own. Fix the spelling and generate the links from one file.
- The dealer master changes. Outlets are sold, merged, change region. The link list needs an owner and a date on which it is reconciled before every wave.
- Too many profiles. Creating an access profile per outlet means maintaining 400 profiles. Outlets get exports or QR reports; profiles go to regions and countries.
- The minimum case count is bypassed in the report. If the rule only applies in the dashboard and not in the export, the cell with one response ends up on the slide anyway. The rule belongs in the project.
- Two truths. As long as head office keeps computing its metric in Excel, there are two DSI values. The index belongs in the dataset as a computed metric.
Next step
Start with the end-customer service feedback, because it delivers data continuously and needs the outlet parameter anyway: after-sales service feedback template. Build the link list from the dealer master, set the minimum case count and create the access profiles for the regions. The dealer survey joins the same project in the next wave: dealer satisfaction template. The full picture is on the page Dealer surveys and after-sales feedback for sales networks.