urban company — rebuilding trust after covid disrupted home services

The moment the product assumptions broke

In February 2020, Urban Company was one of India's cleanest marketplace success stories. The company — formerly UrbanClap — had built a defensible position in home services by professionalising a historically fragmented, unorganised sector. Electricians, plumbers, salon professionals, deep-cleaning crews: Urban Company had trained them, equipped them, rated them, and put them in front of urban consumers via an app that made booking simple and the experience predictable. By late 2019, the platform had processed over one million bookings and expanded to UAE, Singapore, and Australia.

The product's trust architecture rested on a specific assumption: safety meant professional competence. Was the plumber skilled? Would the beautician arrive on time? These were the questions the rating system, the training programme, and the ₹50 lakh liability cover were designed to answer. Users trusted the platform to vet for professional quality, and the platform delivered on that contract well enough to command premium pricing over the street rate.

Then, in late March 2020, India went into lockdown. And the word "safety" meant something entirely different.

The question was no longer "is this a good plumber?" It was "is this person going to enter my home and put my family at risk?" These are different questions. One is answered by a four-star rating and a background check. The other requires a different set of product answers entirely — answers that Urban Company's existing infrastructure wasn't built to provide.

The Decision

Urban Company's initial response was the standard crisis playbook: blog posts, email campaigns, social media announcements about new safety protocols. They published guidelines for providers. They communicated PPE requirements. They told users what the company was doing to keep them safe.

This was the wrong product response — not because the protocols were bad, but because marketing communicates intent while product enables verification. Users who had spent two months watching news coverage of surface transmission and asymptomatic spread were not going to trust a company blog post. They needed the product itself to show them, at the point of booking, that the provider standing at their door had met a concrete safety standard.

The product shift required was moving safety signals from company claims into the booking interface. This meant four concrete changes. Safety badges — visible on provider profiles in the booking flow, not buried in a FAQ — showing completion of certified health and safety training. Health declarations, allowing providers to log their vaccination status and daily health check-ins, surfaced to users before they confirmed a booking. Contactless workflows for payment, scheduling, and feedback, removing physical touch points that weren't essential to the service itself. And flexible cancellation, because users living with pandemic uncertainty needed the ability to reschedule without penalty when their household situation changed.

Each of these was a trust mechanism, not a feature. The distinction matters: features add capability, trust mechanisms remove doubt. Urban Company needed to remove doubt.

The sequencing question was real. Engineering bandwidth was constrained. Urban Company couldn't ship all four trust mechanisms at once. The prioritisation call required understanding which mechanism addressed the biggest conversion blocker.

Safety badges addressed the users who weren't booking at all because they had no way to evaluate provider compliance. Contactless payments reduced friction for users who were already willing to book but anxious about physical interaction. The gap between "not booking" and "booking with some friction" is larger than the gap between "booking with some friction" and "booking frictionlessly." Safety badges first.

The Algorithm Question

Urban Company's service listing algorithm had been optimised for the pre-COVID world: proximity, provider ratings, and price. That weighting accurately reflected what users cared about in 2019. It no longer reflected what users cared about in 2020.

Rethinking the algorithm is a product decision, not a data science decision. The data scientist builds the model the PM defines. And the PM's job in this case was to rewrite the objective function to reflect current user priorities without destroying the utility of the platform for users who had lower safety anxiety.

The key design decision was weighting safety parameters by service risk profile. Deep cleaning and plumbing involve extended in-home presence, close physical proximity, and hard-to-avoid contact. These warrant heavy safety weighting in the ranking model. A mobile beauty service — already contactless in many formats — carries different dynamics. A one-size-fits-all safety parameter produces a ranking model that treats a 20-minute eyebrow appointment with the same anxiety as a 3-hour deep clean. That's bad product design — it distorts relevance and degrades experience for users who don't need that level of reassurance for lower-risk services.

The algorithm redesign also had a supply-side consequence that required active management. Providers who hadn't completed safety certification would rank lower — and lower ranking meant fewer bookings. Urban Company had to decide: hard gate (only certified providers appear) or soft gate (uncertified providers appear lower). A hard gate protects demand-side trust but risks supply-side attrition from providers who haven't yet completed training. A soft gate maintains supply breadth but weakens the trust signal. Urban Company's choice — soft gate with time-limited pressure toward certification — was the right call for an early transition period, with the expectation of tightening over time.

What Worked

The safety badge system did what marketing couldn't: it made provider compliance visible at the moment it mattered — the booking decision. The badge in the profile wasn't decorative; it was a conversion lever. Users who saw a certified provider were significantly more likely to complete a booking than users who didn't. The product change addressed the specific friction point — uncertainty about compliance — rather than layering messaging on top of it.

The contactless payment flow also removed a real friction point for a meaningful segment of anxious users. Cash transactions had historically been common in the home services category — Urban Company's push toward full digital payment wasn't just a safety measure, it was an acceleration of a product direction that was already strategically correct.

The service-specific risk segmentation in the algorithm was the most sophisticated part of the recovery. It preserved recommendation quality for lower-anxiety users while surfacing safety signals prominently for high-anxiety ones. A PM who had applied a single ranking change across all service categories would have degraded the experience for users who weren't primarily motivated by safety concerns — which, even in the middle of a pandemic, was a significant portion of the user base.

What a PM Should Take From This

The Urban Company case teaches a specific kind of product judgment: when an external shock redefines the customer's definition of a core value like safety, trust, or privacy, marketing responses fail and product responses are required. The company's credibility was sufficient to tell users what it was doing. It was not sufficient to make users feel safe. Only the product could do that.

The mechanism matters as much as the intent. A safety badge that lives on a FAQ page doesn't reduce booking anxiety. A safety badge that appears at the point of confirmation does. Product design is fundamentally about where something appears in the user journey and what signal it sends at that exact moment. Moving safety signals upstream in the booking flow was not a cosmetic change — it was the difference between marketing safety and delivering it.

The algorithm redesign illustrates a broader principle: when user priorities shift, the objective function shifts. Product managers who treat the algorithm as engineering infrastructure rather than product expression will find their platform systematically delivering the wrong results as user needs evolve. The PM's job is to keep the algorithm honest about what users currently value — not what they valued when the model was last trained.