Skip to content

From 25% to 96% satisfaction: the performance cycle QuintoAndar runs with Zivo.

A performance cycle in 8 countries and three languages, from evaluation to talent read, on the same platform.

Company context

Key facts

Setor
Proptech · Real estate marketplace
Stage
Series E
Cycle scale
3,000+ people
Geografia
8 countries
Platform languages
Portuguese, English, Spanish
Start of the partnership
April 2025
Entry point
Merit cycle

Latin America's largest real estate ecosystem needed a performance cycle to match.

QuintoAndar is the largest real estate ecosystem in Latin America. Founded in 2013 in São Paulo, the group connects renting, buying and selling of properties and has grown outside Brazil with acquisitions in the region and a technology hub in Portugal.

The partnership with Zivo began with compensation, in April 2025, with the merit cycle as the first front. Today, this partnership also extends to performance, with an experience that operationalizes the cycle end to end and is part of a bigger move by QuintoAndar toward an RH AI-Native. The data from each cycle feeds the next ones, and the AI enters the decision flow.

Before the partnership began, the tool that ran the cycle had 25% satisfaction. The decision was to build with Zivo a tailor-made platform to operationalize the process QuintoAndar already had. Today, the cycle they call Performa runs entirely on it, with more than 3,000 people.

O desafio

The company's most sensitive cycle had no optimized experience for either end: those running it and those being evaluated.

  • Experience

    25% satisfaction with the cycle tool

    The evaluation is run end to end by leadership, who were also the ones registering dissatisfaction in the internal survey, with 25% satisfaction with the previous system and process.

  • Autonomia

    Slow setup

    Adjusting dates, forms, eligibility or criteria was a slow process, which took away the People team's freedom to design the cycle the way it needed.

  • AI outside the flow

    AI assistant in a separate tool

    Leaders already had AI support to write feedback, but in a separate tool that couldn't see the cycle or who was being evaluated. In other words, each evaluation required leaving the flow to interact with the AI.

  • Reaction time

    Adoption always a day behind

    Adoption tracking always arrived with a day of delay, so any move to unblock the cycle started from an outdated snapshot.

The system we built

The cycle runs in five steps, each with a single owner.

  1. People team

    Desenha o ciclo

    The cycle design changes at the moment the People team decides, without going through Zivo. Dates, eligibility, form prompts, Talent Review criteria and even the AI prompts are editable fields on the platform.

  2. Leadership

    Evaluates with the assistant alongside

    The Mid-year Checkpoint for the 3,000+ people runs with the assistant inside the evaluation itself, where it asks before it writes and builds the draft from (i) what the manager answered, (ii) QuintoAndar's leadership principles and (iii) the context about the person being evaluated, like the band they're in. One click takes the text to the editor, where the manager adjusts what goes to the report and, when done, answers whether the assistant helped.

  3. Leadership

    Reads talent

    In the same flow, each manager records potential, readiness, criticality and flight risk for those eligible on the team, ~1,500 people in the cycle, with previous cycles side by side for comparison.

  4. Colaborador

    Recebe o feedback e confirma a conversa

    With feedback released, the person evaluated confirms whether the conversation with leadership has already happened. That confirmation is recorded in the cycle's own flow.

  5. People team

    Acompanha e destrava durante o ciclo

    Adoption by leader and by area updated with every submission, with a talent snapshot and usage of each feature. How the data is from the moment, not from the day before, the People team can lay out the adoption strategy while the cycle is still open.

Technology we built

The platform from the inside, in the four fronts that run the cycle every day.

The leader's cycle

The home screen shows leadership only what depends on them, with pending evaluations, deadline and status per direct report. When each person is opened, role, tenure, band and last evaluation appear next to the feedback field, without having to hunt for data in spreadsheets or other systems.

Review Assistant

On one side of the screen the assistant leads the conversation, on the other the feedback takes shape. What's new in this cycle is the context that goes into that conversation: the assistant sees the band the person is in and helps the manager check whether the feedback is consistent with it. One click takes the draft to the editor, where the manager decides what stays, and in the first cycle 63% of the evaluations were written with the Review Assistant.

Talent Review

Each person's talent read appears side by side with previous cycles, and the matrix crossing talent and performance fills in as leaders evaluate, becoming direct input for succession and retention decisions.

HR Dashboards

Three dashboards for the People team: adoption by leader and by area, status of feedback conversations and usage of each platform feature, including the assistant funnel. All of them update with every submission, without waiting for the end of the day.

Impacto entregue

At the first checkpoint, satisfaction with the cycle tool jumped from 25% to 96%.

25% → 96%
satisfaction with the cycle tool, measured inside the product after the first evaluation.
63%
of the evaluations were written with the Review Assistant.
89%
of the AI drafts were kept by managers in the final text of the evaluation.

The satisfaction and perception indicators for the assistant come from the survey built into the product itself, filled out by managers when they finish the evaluation. Among those who gave an opinion on the Review Assistant, 96.5% said it helped. The usage data, in turn, comes from the platform itself, which measures adoption of each feature.

Leaders are publishing what the AI writes with them, and they say so in their own words in the survey's spontaneous comments.

Each feedback released goes back to the People team with the employee's own answer on whether the conversation took place. All these numbers come from the first mid-year, measured with the cycle still in progress, and serve as a baseline for comparison at year-end.

“Super useful, this AI tool that connects our personal perception to the company's principles and rules.”
Manager in Mexico · spontaneous in-product survey response · free translation from Spanish

What comes next

Performance is a subfunction of a larger cycle.

O merit cycle, the first one Zivo delivered, is the next to run on the same platform, now alongside performance. Next come the experience surveys, so that the result of each cycle becomes context for the next.

The platform was built on top of what previous cycles had already taught, at QuintoAndar and in the other operations where Zivo runs merit and performance.

How satisfied are you with your performance cycle today?

Zivo co-builds and helps operate AI-native HR at Brazil's most innovative companies. In 30 minutes, we learn how your cycle runs today and what would make sense to build on top of it.

Outros casos

More success stories.