The concept of the "Feedback River" was inspired by the Reforge course, 

. Our team has since developed and refined this powerful tool to revolutionize how we handle user feedback. The purpose of the Feedback River is to establish a seamless and efficient process that empowers product managers to stay consistently in tune with user feedback.


The most important part of the feedback river process is ensuring a process of consistently reviewing customer feedback and adding structured insights. This is possible through a number of different tools/processes. This artifact will discuss how we build this process using mainly Slack and Google Sheets, but certainly could be implemented in other ways!


Pro Tip: Aligning these with your user research function can also add great structure to your research and ensure insights from the feedback river are more transferable and interchangeable with your user research function.


The art of the feedback river is to have enough feedback streams that provide a great pulse of the user feedback without it being a deluge. We try have the following rough limits/restrictions:

No more than ~10 streams flowing into the river

Ensuring that we don’t receive repeated feedback from the same customer where possible


I think the constraints of our implementation via BigQuery - > Google Sheets -> Slack were an accidental help when it came to driving adoption of the feedback river within the company. It meant that we had to be thoughtful with introducing streams, naturally a bit limited in the amount of data coming from a stream, and a little bit more control over the format of the content so that it's didn't feel like a lot of noise. All of this makes the river much more consumable. More of a waterpark lazy river than a true river 😀

In fact, when we demoed our feedback river to executives, one executive said he wanted a feedback river for all the client feedback and requested that we help him build a river that funneled in all the feedback in real time. This channel was quickly flooded with feedback, and as quickly as it was formed it was abandoned. It was too noisy for people to pay attention to.

When it came to actually capturing feedback, I did have to lead by example. I was pretty much solely responsible for capturing the feedback initially. However, because feedback capture is very transparent, others joined in soon after — especially once we started producing some of our insight reports.


A few features inspired by the feedback river:


- Favorite Talent - The ability for our clients to tag a talent they worked with as a favorite and quickly rehire them in the future
- Budget Visibility - Visibility to our clients on how the spend of their talent compares to their expected budget
- Consultations - A new product that allows clients to request on-demand experts for quick consultations.

Of course, the feedback river wasn't the sole driver of these new features. Other research went into them as well, but we definitely saw the early signals that they were worth pursuing from the feedback river.

Paul Timmermann

This artifact outlines how we actively use feedback rivers at Toptal to continue to deliver the best experience for our talent and clients.

Our implementation of the Feedback River aims to achieve the following objectives.


We recognize the significance of keeping every member of the product team well-informed about the most critical feedback items from our users. By utilizing the Feedback River, product managers can maintain a real-time pulse on user sentiments, pain points, and desires, fostering a deeper understanding of our audience's needs.


Capturing valuable insights from user feedback is paramount to improving our products effectively. With the Feedback River, we ensure that these insights are not lost in the deluge of information but rather organized and documented for easy reference. This process establishes a robust knowledge base that informs future decision-making processes.


The dynamic nature of our product constantly introduces new workflows and feedback sources. The Feedback River's adaptability empowers us to integrate these changes effortlessly, without relying on complex coding or implementation processes. This flexibility enhances our ability to respond promptly to emerging trends and customer preferences.

Here’s the feedback river process at a high level.

On a regular basis, new feedback items are automatically included in the feedback river through various integrations and workflows. Feedback river members review the feedback to keep a pulse on their users.

A simple thread-based workflow kicks off when a channel member reacts to a feedback item. This opens a form to capture specific insights. Observations are posted back to the channel.

All insights are copied back to the spreadsheet for future analysis and drive future product decisions.

. You’ll use this spreadsheet throughout the following steps to document outcomes and set up data capture for the next steps.

As the key outcome of the feedback river are the structured insights you’ll receive from the process, it helps to start with defining the insights you are looking to capture and work backwards from this point. First, structuring the high-level persona you’d like to capture in the feedback river is helpful in managing the feedback categorization. We keep our feedback rivers split between the high-level personas of our marketplace: talent and client.

We capture/categorize our insights using two main criteria:

User Journey Step - The main step in the user journey related to the feedback

Related Feature Area - The main feature/feature area that the feedback is related to.

We found this is a great way to refer to opportunities in the product, “We see lots of feedback in the onboarding user journey step” or “We see an interesting trend in feedback around our billing feature set”

As with any sort of categorization, finding the proper middle ground is challenging. We recommend less than 10 user journey steps and less than 15 feature areas. 

For our feedback river, we have defined the following:

Onboarding - New client in the onboarding process

Interviewing - Client actively interview talent to hire

Hiring - Client that has completed the interview process and is finalizing hiring details

Active engagement - Client that has approved the trial and actively working with the talent

Job creation - Features related to the job creation and posting process

Interview scheduling - Features related to client/talent interview scheduling

Contracts - Features related to contract management

User management - Feature related to user administration and management

Billing - Feature related to billing, invoices, and spend tracking

Document your user journey steps, related feature areas, and associated definitions for each in your feedback river spreadsheet.

Each user insight source is called a stream and it takes several streams to build a great river. Leveraging your insights template will help you uncover great stream sources. Plan a brainstorming session with your product teams and teams key to the journey and feature areas to identify stream sources. Tools like Miro and FigJam are your friend here. Add journey steps and related feature areas as cards on the board and have team members brainstorm on places where they’ve heard (or believe they would hear) user feedback related to the journey step/related feature area.


Customer Service Tickets (ZenDesk, Jira, etc.)

User Research Records (Confluence, Airtable, etc.)

Document each of these in your stream backlog. They will be useful to delegate stream building work in the future.

Now that we’ve identified the insights we want to capture and where we are going to get them, we are ready to start building streams. To start, we’ll need to create the Slack river channel where we intend to send all the streams. We recommend creating the “production” channel as well as a “testing” channel so you can work out the kinks of new streams without 

polluting the river or causing an erroneous stream to flood it with too much feedback

We developed 3 main methods to get streams into the river:

BigQuery → Google Sheets → Slack (via Workflow)

There are plenty of third party apps that would also integrate with Slack, but most of our streams are served by these methods. We’ll expand upon a few:

For data that’s readily available in BigQuery, we use this method to get data into the stream. Here are steps:

Write the query - Create a query in BigQuery that generates a single record the most recent time period (we usually use yesterday) that produces output of:

Use Google Connected Sheets and add the query and set to refresh at hourly

Create an Slack import spreadsheet that prepares your data to import to Slack

Create a Slack workflow that runs daily and pulls in the Slack import spreadsheet if there’s valid data


See the “Sample - Connected Sheet”, “Sample - Extract Sheet”, and “Sample - Slack Ready” sheets in the sample Google Sheet to set this up.

We wanted a way for our client services team to be able to quickly share relevant feedback in one of the key tools they use, Gong.io

Create an email alias that forwards to your email address

Create a new user in Gong.io with that email address that can be tagged in Gong.io comments

Create an email address for the Slack channel (Learn more 

When a tagged email is received from Gong.io for that alias email address, create a forwarding rule to forward to the email address for the Slack channel

If you use Segment.io and there’s a Segment event that would make sense to bring into your feedback river, the out-of-the-box integration between Segment.io and Slack can help get feedback flowing immediately. See the 

Insight capture is the one of the final pieces that makes the whole feedback river come together. We use a simple workflow here that allows anyone in the feedback river channel to capture an insight. Steps to set this up:

Feedback River Process at Toptal

Feedback River Process at Toptal

Purpose

How It Works

Sign up to see more