We are not in the room when the job happens. We are not there when the professional rings the bell, when they explain what they're about to do, when they make the judgment call between a quick fix and a full repair. We vet professionals carefully, we track arrivals, and we structure the post-job review — but the actual work, the human interaction inside a homeowner's kitchen or bathroom, is between those two people.
This is the fundamental tension in any home services marketplace. You can design around it, reduce the variance, create accountability loops — but you cannot eliminate it. Pretending otherwise is one of the most dishonest things a platform in this space can do. What we can do is describe honestly what our trust architecture consists of, and where its limits are.
Before the job: information as the first layer of trust
Trust starts before anyone rings a doorbell. A homeowner who books through OSCAR sees the professional's name, photo, certification status, and review scores before confirming the booking. They see how many jobs that professional has completed through OSCAR, and they see dimension-specific reviews — arrival punctuality, price accuracy, work quality, and cleanliness — not just a single aggregate star rating.
We made the decision early to show separate ratings rather than a combined score because a combined score conceals the dimensions that matter for specific job types. A homeowner booking a plumber for a boiler service cares most about technical competence and whether the professional leaves the boiler room clean. A homeowner booking a locksmith for an emergency lockout cares most about arrival time. Showing one number flattens information that should be visible.
We also show the professional's verified credential status directly on their booking profile. "IMPIC registered" or "Alvará de Instalador verified" is not marketing language — it is a factual status that we check and update annually. If a credential lapses, the status changes on the profile. We don't wait for the professional to notify us.
During the job: tracking and documentation
Once a booking is confirmed, homeowners can see the professional's real-time location from the moment they mark themselves as travelling to the job. This is the same mechanism used in any delivery service — not a novel feature — but it is particularly meaningful in the home services context where uncertainty about arrival time is one of the most corrosive trust problems in the market.
Professionals are required to photograph the work area before starting and after completing every job. The before photo creates a documented baseline state of the space. The after photo confirms the work is done and the area is in acceptable condition. These photos are attached to the job record and accessible to both the homeowner and, in cases of dispute, to our operations team.
We introduced the pre-job photo requirement after noticing a pattern in early disputes: homeowners and professionals had genuinely different recollections of the initial state of a work area. A plumber who arrived to find existing water damage under a sink had no documentation that the staining predated their work. A photo taken when they arrived would have resolved that clearly. The requirement is now mandatory for all job categories.
After the job: the review structure and what it actually measures
Within 24 hours of job completion, both parties receive a review prompt. The professional reviews the homeowner (conduct, access, clear description of the problem). The homeowner reviews the professional across four dimensions:
- Arrival punctuality — did they arrive within the confirmed time window?
- Price accuracy — did the final invoice match the confirmed price?
- Work quality — is the problem resolved?
- Conduct and cleanliness — were they professional, and did they clean up after the work?
We do not aggregate these into a single score for display purposes. We show all four dimensions separately on professional profiles. A professional who consistently delivers excellent technical work but leaves tools out during cleanup will see that reflected specifically in their "Conduct and cleanliness" dimension rather than dragging down an undifferentiated average.
Review prompts expire after 48 hours. We don't chase homeowners for reviews — we prompt once. Review completion rates vary by job category and range between 55% and 75% in our current data. We consider that acceptable. A forced review is not a useful review.
The accountability loop for professionals
Review scores feed into a professional's scheduling priority. A professional with consistently high scores across all four dimensions ranks higher in candidate selection for new bookings, all else being equal. A professional whose "arrival punctuality" score drops below a threshold triggers an internal review of their scheduling patterns.
This is not a punishment system — it is a feedback system. A plumber who is chronically running late on afternoon bookings may simply be scheduled too tightly. The review data surfaces that signal, and we adjust their scheduling buffers accordingly before the pattern becomes a customer problem. In two cases since launch, review data revealed that professionals were accepting bookings on a parallel channel (direct from homeowners they met through OSCAR) and managing that external book alongside their OSCAR schedule without disclosing the conflict. Both professionals were deactivated.
Where the trust architecture fails
We want to be direct about this. Our trust architecture works well for the majority of straightforward jobs. It works less well for three categories of situation:
Long-duration complex jobs. A bathroom retiling that spans two days involves extended access to a home, multiple materials deliveries, and scope complexity that is hard to document fully with before-and-after photos. Our documentation requirement was designed for the 60-90 minute standard service call. We are working on a structured milestone documentation process for multi-day jobs, but it is not live yet.
Latent defects. If a plumber's repair fails two weeks after the job — a fitting that held initially but wasn't properly seated — the review was likely already submitted and closed. Our warranty policy covers workmanship defects reported within 30 days of job completion, and we handle those through our dispute resolution process. But a review system cannot surface problems that haven't appeared yet at review time.
Interpersonal conduct that isn't reported. A homeowner who experienced uncomfortable conduct from a professional during a job may not report it if they feel unsure about the process, feel the conduct was ambiguous, or simply want to move on. Our review structure does not capture what homeowners don't choose to disclose. We have a separate confidential conduct reporting channel for these situations, and we take those reports seriously — but we cannot compel homeowners to use it, and we are honest that some situations go unreported.
Trust on a two-sided marketplace is built incrementally, through consistent small signals, and it is more fragile than it appears. Every job that goes well is a deposit. Every unexplained problem is a withdrawal. We are building the deposit side of that account carefully, and we are not pretending the withdrawal side doesn't exist.
