Asking rent is not achieved rent: build the right dataset
Compare advertised rental observations with consistent periods, property attributes, concessions and included-cost definitions.
In this article
THE KEY IDEA
Describe what was advertised. Do not turn an asking-rent sample into a claim about signed leases without supporting data.
01Start with a consistent rent definition
A rental listing can show weekly, monthly or another payment period. It may include some utilities, exclude service charges or advertise a concession. Preserve the original amount, currency and period before normalisation. A bare numeric rent field is not enough for comparison.
Define whether the analysis uses headline rent, effective rent under a stated concession, or a broader recurring housing cost. Keep those measures separate. Do not assume that an advertised incentive applies to every tenant or lease term.
02Capture the property and tenancy context
Keep property type, bedroom count, unit identity and location at the precision the source provides. Furnished status, lease duration and included costs can materially affect comparability. Missing details should remain unknown rather than being filled from nearby listings.
An address can contain several rental units. Match at the appropriate level and preserve source listing IDs. A building-level merge can make different apartments look like price changes on one property.
| Context | Comparison risk |
|---|---|
| Rent period | Weekly and monthly values mixed |
| Unit identity | Different apartments treated as one |
| Furnishing and lease term | Different tenancy products compared |
| Included costs | Headline prices with unequal coverage |
| Concessions | Conditional discounts treated as universal |
03Keep listing history separate from market events
Store first seen, last seen and successful check times. A listing disappearing does not prove that a lease was signed. It may have expired, been withdrawn, moved to another URL or fallen outside the current search filters.
Record explicit source status changes when available, and label inferred transitions separately. When collection fails, preserve the last known observation without presenting it as current. These rules prevent gaps in your pipeline from becoming false market signals.
04Compare similar cohorts
A median can rise because the sample contains more large or furnished properties. Define geography, unit type and other relevant cohorts before calculating trends. Show the number of observations and coverage changes with each summary.
When normalising rent periods, state the conversion convention and avoid implying that it changes the contract's actual payment schedule. Retain original values so the consumer can apply a different convention if needed. Review outliers before removing them automatically.
05Deliver the limits with the numbers
Describe the output as advertised rental information from specified sources and dates. Disclose missing areas, failed collection and deduplication choices. Do not label it achieved rent, occupancy or completed leasing activity unless those facts are actually supplied by the sources.
A useful rental dataset supports transparent comparison and further investigation. Its value comes from consistent definitions and history, not from presenting every observed number as a complete account of the rental market.
References & further reading
Put it into practice.
Try a public-page workflow, or scope a maintained dataset with our team.