Four ground rules
Two values, never mixed
The Pause-Score (0–100) says how good the place is for you. Confidence (0–1) says how complete and reliable the data is. Stale data or estimated walks change only confidence, never the score.
Unknown is never positive
An unknown property contributes 0 points and lowers confidence. There is no average that upgrades the unknown.
Hard filters before ranking
Technical suitability and mandatory conditions are fully checked before rating. Whoever fails is never rated.
Nothing invented
No demo data in production, no live availability without source and timestamp.
Filter first, then rate
Every search is evaluated in this order. Only the survivors of stage 6 receive a score.
| Stage | What happens | Excluded if … |
|---|---|---|
| 1 | Fetch charge points from all sources and merge them (≤ 30 m apart, similar power) | – |
| 2 | Check radius (straight line) | outside the search radius |
| 3 | Technical filter: connector, DC, minimum power | wrong connector, no DC, or power too low or unknown |
| 4 | Candidate limit: the 15 nearest by straight line | further away than the 15 nearest (reported) |
| 5 | Load surroundings, route walking paths (otherwise marked as estimate) | surroundings data could not be loaded |
| 6 | Mandatory place (e.g. bakery) within the maximum walking time, optionally open at the planned time | no place of that category, too far, closed or opening hours unknown |
| 7 | Only now: Pause-Score of the rest, sorting, top 3 | – |
The formula
Each of the ten criteria yields a value s between 0 and 1 (unknown: 0). The score is the weighted sum:
Pause-Score = Σ 100 · weight · value
confidence = 0.55 · known + 0.20 · routed + 0.15 · freshness + 0.10 · sourcesBase weights
Without preferences these weights apply (sum 100 %):
How your selection shifts the weights
The base weight is multiplied by a factor and then normalised to 100 %. Niche criteria you did not choose are weakened, not hidden.
| Selection | Effect |
|---|---|
| With children | Child-friendliness × 3 (otherwise × 0.25) |
| With dog | Dog-friendliness × 3 (otherwise × 0.25); green × 1.5 unless “green” is also selected |
| Green / movement | Green × 2.5 |
| Shopping | Shopping × 3 (otherwise × 0.5) |
| Food in focus | Food & drink × 2 |
The ten criteria in detail
Food, shopping, green and opening hours count places up to 15 minutes’ walking time, children up to 12 minutes. Walking time is the real routing duration, otherwise an estimate marked “estimated”.
| Criterion | How the value (0–1) is computed | Unknown if … |
|---|---|---|
| Charging suitability | 60 % power (credit of 0.5 for passing the filter, above that up to 300 kW), 20 % number of sockets, 20 % live-free (only with timestamp); minus 0.4 if out of service | never |
| Reachability | 1 − distance ÷ search radius (straight line) | never |
| Food & drink | 40 % count (3 = full), 30 % variety of kinds (3 = full), 30 % proximity of the nearest place | surroundings data not loaded |
| Child-friendliness | 70 % proximity of the nearest playground, 30 % kids’-area tag | surroundings data not loaded |
| Dog-friendliness | 60 % proximity of the nearest dog meadow, 40 % places with “dogs allowed” (2 = full) | neither a dog tag nor a dog meadow nearby |
| Shopping | 70 % proximity of the nearest shop, 30 % count | surroundings data not loaded |
| Green space & movement | 70 % proximity of the nearest park or dog meadow, 30 % count | surroundings data not loaded |
| Opening hours | Share of relevant places that are open at the planned time (unknowns count in the denominator, never in the numerator) | no evaluable hours |
| User ratings | weighted average ÷ 5, once a source provides ratings | currently always |
| Data quality | 1 if at least two independent sources confirm the station, otherwise 0 | never |
Confidence
It measures only the data situation – never how nice the place is. The score never flows into confidence and vice versa.
| Part | Weight | Calculation |
|---|---|---|
| known | 0.55 | Sum of the weights of all criteria that are known. If something you care about is missing, confidence drops more. |
| routed | 0.20 | Share of the places considered with real walking routing |
| freshness | 0.15 | 1 − data age ÷ 48 hours; halved if stale data is delivered because of a source outage |
| sources | 0.10 | Number of independent sources of the station ÷ 2 (at most 1) |
Ranking
Sorted by Pause-Score (descending); ties are broken by higher confidence, then shorter distance. The first three are the top 3.
Three worked examples
All examples: start Bonn, radius 15 km, DC from 150 kW, Saturday 10:00, status unknown, no user ratings. The numbers are automatically checked against the program code.
Example 1: well-equipped 300 kW station, no preferences
3.0 km from the start, four CCS sockets at 300 kW each, confirmed by Open Charge Map and OSM. Bakery 3 min, café 5 min (dogs allowed), playground 6 min, supermarket 7 min, park 9 min – all routed.
| Criterion | Weight | Value | Points |
|---|---|---|---|
| Charging suitability | 23.8 % | 0.80 | 19.0 |
| Reachability | 14.3 % | 0.80 | 11.4 |
| Food & drink | 21.4 % | 0.71 | 15.1 |
| Child-friendliness | 2.4 % | 0.35 | 0.8 |
| Dog-friendliness | 2.4 % | 0.20 | 0.5 |
| Shopping | 4.8 % | 0.37 | 1.8 |
| Green space & movement | 11.9 % | 0.43 | 5.1 |
| Opening hours | 9.5 % | 1.00 | 9.5 |
| Ratings | 4.8 % | unknown (0) | 0.0 |
| Data quality | 4.8 % | 1.00 | 4.8 |
Example 2: same station, with children and dog
The weights shift towards children and dogs, whose values here are only 0.35 and 0.20. So the score drops to 53.7, while confidence rises slightly to 0.98.
Example 3: patchy data
9.0 km away, exactly 150 kW, one source only, data 24 hours old, surroundings data not loadable. Result: Pause-Score 12.9 and confidence 0.36. Gaps produce a low score and low confidence – and both are shown. With a mandatory bakery this station would not be rated at all.
Known limits
- Reachability uses the straight line to the start, not the driving route.
- Public holidays and complex opening-hours expressions are not evaluated (then “unknown”).
- In very densely mapped areas the result of a surroundings query is cut off at 500 or 600 elements; this is currently not reported separately.