The period
A change compares the reports in a period with the reports before it. A period is the latest weekly report, the last four, the last twelve or all of them. Home always reads the latest weekly report, so its figures, score and status line state the same change as the other pages and the API. The Monday email, the Segments table and the manager Portfolio state the same change for the latest weekly report, a segment’s over that segment’s answers. The Period filter on the pages that have one and the API’speriod choose another. A period counts reports, not days, so a paused week stretches it over more calendar weeks.
The figures pool the answers in the period and compare them with the preceding reports. The baseline holds as many reports as the period, and at least four when available. With a nonempty baseline of fewer than four reports, a measured, comparable figure is calibrating. Without earlier reports, there is no change. The response states the baseline’s dates, report count and unpaired cells. Figures in the API describes these fields.
The paired test
Heralded compares the two sides prompt by prompt and engine by engine, using only the cells both sides measured. A cell measured on only one side drops out of the comparison instead of skewing it. For each shared cell it takes the difference between the two sides and weights it the same way as all four figures. It sums the differences and puts a 95% interval around the total. If that interval excludes zero, the change is real. For the Heralded Score, the test runs on Mentioned and Recommended and the result is placed on the calibrated scale. A change can differ from the gap between the two sides’ figures when some measured cells are present in only one report, because the test uses only shared cells. The interval reflects how much the prompts’ own differences scatter. Each prompt’s difference averages its shared engines equally. A figure for one prompt uses its engines as the units. The test uses a Student t interval, which is wider with fewer units. It needs at least three prompts, or three engines for a single prompt. Identical differences have no scatter, so their interval has zero width. This approach assumes the units’ differences are independent. Its calibration is provisional while Heralded checks how answers move together across weeks. Share of voice keeps its own pooled-slot interval.Reading a change
A change has three parts: the difference in points, its range, and whether it is real. When it isn’t real, treat the figure as unchanged, whatever the difference says. At 95%, about one figure in twenty that didn’t really move will still show as real. One real change is a signal. The same move over several reports is a trend.When there is no change
A figure has no change when:- no weekly report comes before the period;
- fewer than three units remain after pairing, so there are too few answers to test the change;
- Heralded changed how it reads between the period and its baseline, whether that is the engines read, a segment’s market, the score version or the way answers are read;
- the brand moved to a new domain. Change returns after two reports on the new domain.
null values and real: false. Figures in the API describes when the change itself is null.
Adding or archiving a prompt, or adding a segment, doesn’t blank the change. Cells with nothing to pair with sit out.