Data Quality

How much of the raw data survives the validity rules, and exactly what is left out. Read this before quoting any figure externally.

What the data says

Generated from the records currently in scope — observation, why it matters, and the decision it points to. Nothing here is written in advance; every figure recomputes on the next upload.

What the evidence can carry

Observedn = 15,697

Observation: 13,409 searches and 2,288 enquiries are in scope. 99% of enquiries state a usable budget (2,262) and 99% of searches carry a plausible move-in date (13,223).
Interpretation: Every price statement in this app rests on 2,262 records, not on 2,288. The move-in date field is the binding constraint: each missing entry removes one observation from the pricing and timing evidence, and the loss is not random — it skews toward enquiries that dropped out of the form early.
Investment implication: Coverage is sufficient to underwrite on; keep the field mandatory and re-check this share at every upload before figures are quoted externally.

Budget completion

100%

Share of leads that stated any budget figure.

Observedn = 2,496

Budget validity

97.5%

Stated budgets inside the plausible range of EUR 200 to EUR 3,000.

Calculatedn = 2,496

Stay length validity

98.7%

Searches with a usable move-in date and stay duration.

Calculatedn = 14,400

Location mapping

99.3%

Share of lead and search records resolved to a portfolio asset.

Observedn = 16,896

Validity by source

Each source is measured against the rules stated in the methodology.

Source	Records	Usable	Share usable	Main exclusion
Term range searches	14,400	14,206	98.7%	Missing or implausible stay dates
Exit-intent leads	2,496	2,434	97.5%	No budget stated or outside range
Configurator	106	88	83%	Unresolvable location or filter type

Known caveats

Structural limits of the evidence base that no amount of filtering removes.

Budgets are self-stated intent at exit, not signed rents; treat them as willingness-to-pay signals.
225 contacts submitted more than once — use the first-submission filter to remove their weight.
5 configurator size options span a wide range; their midpoint is a coarse proxy.
Configurator clicks measure interest in an option, not a completed booking.
Searches and leads come from different funnel stages and should not be divided into one another as a conversion rate.

Unmapped labels

These raw labels did not resolve to a portfolio asset and are excluded from location views.

Raw label	Records
Unknown	121
Frankfurt.	1
Berlin Clay	1
Berlin Oak	1
Berlin Yard	1
Berlin Yard Ii	1
Bochum	1
Bonn	1
Dresden North	1
Dresden South	1
Essen	1
Frankfurt	1
Halle	1
Karlsruhe	1
Kiel	1
Ludwigsburg	1
Munich	1
Pamplona	1
Stuttgart Central	1
Stuttgart Urban	1