How the KI Score is calculated

This article explains how the KI Score for your market is built. Afterward, you’ll be able to follow the number on the KI Score page and know which factors move it and which don’t.

The formula in one sentence

The KI Score combines the mention rate (two thirds) and position coverage (one third) into a value from 0 to 100:

KI Score = (2/3 × mention rate + 1/3 × position coverage) × 100

It is based on the answers from the market measurement. KI Score asks each AI provider every one of the 10 to 15 questions in the question set twice, on two measurement days three days apart. All brands in a market are measured with the same questions.

Mention rate and position coverage

The mention rate (Erwähnungsrate) is the share of answers that name your brand.

Position coverage (Positionsabdeckung) also takes into account where your brand appears. Each mention gets a position weight of 1 divided by its position: first place counts as 1, second place as 0.5, third place as one third. An answer without a mention counts as 0. The average across all answers is the position coverage.

A mention with no identifiable position counts toward the mention rate but adds nothing to position coverage.

This means the KI Score can never be higher than the mention rate times 100. If your brand appears in half the answers, it can reach 50 at most, even if every mention is in first place.

Weighting by question intent

Not every question counts the same. Each question in the question set has an intent, and answers are weighted accordingly:

Question intent	Weight
Recommendation	5
Comparison	2
Problem	2
Information	1

An answer to a recommendation question carries as much weight as five answers to information questions. The weighting applies equally to the mention rate and position coverage.

A quick example: your brand is named in the answer to a recommendation question but not in the answer to an information question. Unweighted, the mention rate would be 50%. Weighted, it is 5 of 6 units, or about 83%.

Averaging over three measurement cycles

The KI Score page doesn’t just show the result of a single measurement. It shows the average of the scored measurements from the same methodology series in the current measurement cycle and the two before it. With weekly measurement, that’s the current calendar week and the two before it. A methodology series covers measurements with the same question set, the same calculation method, and the same selection of AI providers.

Keep in mind:

• A measured week in which your brand wasn’t named counts as 0.
• A week that wasn’t scored, for example because too many answers were missing, isn’t replaced by an older week. It simply drops out of the average.

With weekly measurement, the info text on the KI Score page reads: “Angezeigt wird der über bis zu drei Wochen geglättete Score derselben Methodik-Serie.” (Shown is the score smoothed over up to three weeks from the same methodology series.) This refers to exactly that window of three measurement cycles, not to the last three measurements.

A special case: monthly measurement

On the Essential plan, your market is measured only once a month, in the first measurement week of the month. A measurement cycle then covers one month, so the value is smoothed over up to three monthly measurements. The info text reads accordingly: “Dein Markt wird monatlich gemessen. Angezeigt wird der über bis zu drei Monatsmessungen geglättete Score derselben Methodik-Serie.” (Your market is measured monthly. Shown is the score smoothed over up to three monthly measurements from the same methodology series.)

There is one exception: if another customer with weekly measurement uses the same market, the market is measured weekly, and your value is smoothed over up to three weeks.

Sample calculation

Say there are ten answers to information questions. Your brand is named in six of them, three times in first place and three times in second place.

1. Mention rate: 6 of 10 = 0.60
2. Average position strength of the mentions: (3 × 1 + 3 × 0.5) / 6 = 0.75
3. Position coverage: 0.60 × 0.75 = 0.45
4. KI Score: (2/3 × 0.60 + 1/3 × 0.45) × 100 = 55

For reference:

Situation	KI Score
Never named	0
First place in every answer	100
Second place in every answer	83.33
Second place in half the answers	41.67

Note: A value of 50 doesn’t mean “average.” It means your brand appears in roughly half of the relevant answers, at the position quality measured. Your standing in the market is shown by your rank in the market.

What doesn’t affect the score

The following factors don’t change the KI Score:

• whether AI answers cite your domain as a source
• the tone of a mention, that is, whether the AI recommends your brand or only mentions it
• the results of KI-Readiness
• your credit balance
• your own monitoring prompts and their results

You’ll find these figures elsewhere in the dashboard. To improve your KI Score, your brand needs to appear more often and higher up in the AI answers for your market.