Could the 8.1% be recalculated after removing unresolved records? That is the fastest test of whether the classification issue is driving the headline.
I’d add a third version using only records with consistent original asking dates. Otherwise the price movement and 111-day figure may rely on different subsets.
There is also a framing issue: “price movement +8.1%” sounds like appreciation. If it is merely a difference between selected asking prices, call it that.
Has anyone challenged the median itself? With a small sample, 111 days may be determined by just one or two central records. Showing the individual durations would help.
A simple ordered list of marketing times would make that obvious. It would also reveal whether there is one cluster or several groups tied to condition.
Do not forget active listings are unfinished observations. Their eventual marketing time is unknown, so combining them with completed or withdrawn records needs careful wording.
This may explain some noise attributed to condition. Older active stock accumulates days, while attractive listings that complete disappear from the active snapshot.
So a current-availability sample is inherently tilted toward slower stock. It can describe competition facing a buyer today, but not cleanly estimate typical selling time.
For a buyer choosing among current listings, missing energy information itself is actionable uncertainty. For measuring price effects, it is simply another incomplete variable.
For a seller, the next step could be obtaining clear estimates for likely improvements before setting price. The legal and technical specifics depend on the property and Spanish jurisdiction, so forum assumptions are not enough.
I’d avoid assuming improvement is always rational. A price adjustment may be cheaper and quicker, especially if buyers would renovate for other reasons anyway.