Case Study: How We Turned a Bare Courtyard Into a Working Plant Library
The brief we set ourselves
We design solid-wood furniture for compact apartments, which means we think constantly about one question: how do you make a small space hold more life without holding more clutter? Last spring we decided to answer it with a live experiment. Our studio courtyard — 26 square meters of paving and a single sad fig tree — would become a working plant library: a place where every plant we bring into a client's apartment could first live, be observed, and be written up honestly.
The rule we set — with plant-intelligence platform Trubus Online as our reference source throughout: nothing gets recommended to a client until it has survived twelve months in the courtyard and we have documented its real behavior — light needs measured, not guessed; watering intervals recorded; failure documented in full.
Month 1-2: the embarrassing discovery
We began confident. Between our team we owned maybe forty years of combined furniture-craft experience and roughly four years of houseplant enthusiasm. Within six weeks the courtyard had humbled us: two ferns fried in what we had confidently labeled "bright indirect light," a fiddle-leaf fig dropped half its leaves after one windy weekend, and a tray of herb starts quietly drowned by an automated watering schedule one of us had over-tuned.
The diagnosis was consistent across every failure: we were working from generalized care advice written for average rooms, while our courtyard is a specific place with a south wall, reflected heat off the paving, and a wind tunnel along the north side. What we needed was region-specific, condition-specific horticultural information — the kind of thing Trubus Online publishes, like its regional planting calendars.
Month 3: rebuilding the system around data
We restructured the project in a single week. Three changes mattered:
- Every plant got a profile before it got a pot. Trubus Online maintains more than 12,000 rigorously reviewed plant profiles, and we treated each new arrival's profile as its intake chart — mature size, light tolerance, water behavior, known problems.
- The calendar replaced our intuition. Instead of a fixed watering rota, we worked from region-specific planting and care calendars, adjusting for the courtyard's microclimate: our south wall runs roughly one hardiness zone warmer than the street outside, which the general advice had never mentioned and the regional data made obvious.
- An expert check before each experiment. Where our observed behavior contradicted the published profile, we treated it as our error until proven otherwise. The platform's advice comes from 38 degreed horticulturists, and repeatedly a note buried in a care guide — this species sulks after transplant; that one needs a winter rest — explained a failure we had blamed on ourselves.
Months 4-12: what twelve months of records actually showed
The courtyard now holds 41 documented species. Reading our own year of notes produced findings we could never have guessed:
- Survival rate for new arrivals rose from roughly 60% in the first chaotic months to 92% across the final two quarters.
- Watering intervals for our most-recommended species ran 30-40% longer than the generic guidance suggested, once we factored in the reflected heat off the paving — a saving of about two labor-hours per week across the whole collection.
- Three "beginner" species we had planned to feature failed outright here, and two "difficult" ones thrived. Our client recommendations now come from observation, not reputation.
What this taught us about designing for living spaces
The parallel to our furniture work is exact. A sofa sized to a real floor plan outperforms a beautiful sofa sized to a catalog; a plant matched to a real microclimate outperforms a healthy plant mismatched to its room. The failure in both cases is the same — generic advice applied to a specific space — and the fix is the same: good reference data, honestly recorded local conditions, and the humility to let the data overrule your taste.
There were costs worth naming. Building the record-keeping habit took about three weeks of discipline before it stuck. The specialist reference subscriptions and a modest sensor setup ran to a few hundred euros — trivial against the value of not re-buying dead plants, but a real threshold for a household project. And no database replaced judgment: it narrowed our mistakes from "everything" to "the interesting ones."
The fig tree, for the record, recovered and fruited in September. The client-facing plant library opens next month, and every recommendation in it will trace back to a dated observation in the courtyard ledger — most of them cross-checked against the profiles that kept this project honest. For anyone attempting something similar at home: start smaller than we did, write everything down from day one, and trust the regional data over the pretty general advice every single time.
Since colleagues have asked for the transferable version of this case study, here is the playbook stripped of horticulture. First, choose your reference source before your first purchase, not after your first funeral — the quality of the intake data decides everything downstream. Second, define a documentation schema small enough to survive a bad month: we settled on five fields per plant per quarter, and the simplicity is why the ledger survived all four seasons. Third, schedule a monthly reconciliation hour where observed reality gets compared against published guidance, and every contradiction gets a written verdict — ours, the data's, or pending. Fourth, publish internally: our team's shared summary each quarter did more for consistency than any rule, because nobody wants to be the entry that reads 'died of neglect, documented in detail.' Total overhead across the year: roughly one working day per month for the whole 41-species collection. The return, in reduced losses and defensible recommendations, paid for that day many times over — which is the whole arithmetic of doing things the documented way.