---
title: Gumroad Discover PM Challenge - Amos Goh
---

# Overview :world_map:
Hi, Amos here and this is my attempt at the Discover PM challenge for Gumroad
* You can find :
* the **Product challenge** below
* the **Editorial challenge** here: https://hackmd.io/@arentweall/rJ3_t3o5u
* the **Combination Figma** here: https://www.figma.com/file/IvACdDdqvOEmsHxmMdHBis/Gumroad-Discover-Combination-Figma?node-id=1%3A86
* **Source link** from Gumroad notion: https://www.notion.so/Discover-PM-challenge-f5d2da1d53ed4e9d96b222fbd41ad1cb

Thank you for your time, and I'd love to know your thoughts and feedback on my work :pray:

<br>
---
<br>

# 0. Objectives :round_pushpin:
The objective of this research is to analyse the current Gumroad Discover product and provide suggestions to:

| | Objective | Current | Eventual |
| --- | -------------------------------------------------- | ----------- | ----------- |
| 1 | Increase human touch | Qualitative | Qualitative |
| 2 | More accurately showcase what Gumroad is used for | Qualitative | Qualitative |
| 3 | Increase Gumroad Discover-powered GMV for creators | 5% of GMV | 20% of GMV |

Objective 3 is a lagging indicator for the first 2 qualitative objectives. As a discovery tool, if it is more humanely relatable and it accurately showcases what Gumroad creators are doing / selling, more users will be discovering products from Discover and buying there

<br>
---
<br>

# 1. Who are the users? :boy::girl:
Without data on the subdomain discover.gumroad.com, I will be using gumroad.com's [Similarweb](https://www.similarweb.com/website/gumroad.com/#overview) data to better understand who are Gumroad users.

| | Data Type | Findings | Assumption / Insights |
| --- | ---------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------- | ------- |
| 1 | User geography | 40~% of users come from USA, UK, Canada, France and India. 25% of users come from USA (top country) alone | Gumroad has a global user base with **majority in English speaking countries**. In the future, **localisation might be an important area** |
| 2 | Referral traffic sources | 42.17% of traffic comes from direct traffic, followed by 30.9% from socials, and 15.66% from referrals | Most Gumroad users either come from **an email (i.e. "direct traffic") or social link from their creators** |
| 3 | Referral & Destination sites | Top referring sites are Patreon, Product Hunt, Behance, Github whereas top destination sites are Paypal, Youtube, Google Drive, Twitter & Facebook | Verifying the point above, most users come from creators who are **product builders, designers and programmers**. Users either proceed to payment or onto the creators' social profile for further exploration |
| 4 | Social traffic sources | 56.7% came from Youtube, 25.13% from Twitter, 4.67% from Facebook, 4.05% from Deviantart, 3.51% from Reddit. | A majority of users interact with their audience via Youtube videos |

With an understanding of Gumroad's users, let's focus on the subset of users who will actually end up using Discover. Hence, from here, when I use the term "users", I am referring to Discover users only.

I have realised that it is not necessary to split users into existing and new users, because Discover is a relatively new product that Gumroad is not known for. Hence, it is safer to assume that all users who interact with Gumroad should be treated and onboarded as new users

Generally, after clicking through social or email links to the creator's Gumroad page, the user must have completed these steps to view Discover:
1. Click on the "Powered by Gumroad" button
2. Land on Gumroad's main page (gumroad.com)
3. Click on the Discover button or use the search bar
4. Land on Gumroad Discover (discover.gumroad.com).

<br>
---
<br>

# 2. What should Discover aspire to be? :star2:
When users do click on "Discover" or are prompted by "Search for products" in the search bar on gumroad.com, they are expecting Discover to help them do 2 things:

1. Discovery: For those looking to discover (still unsure of what they're looking for), it should **help them discover, browse and buy products from creators they are interested in**.
2. Search: For those who know what they are looking for, it should help them **narrow down their search and find what they want.**

Given that, here are the quality criteria I will use to assess the current Discover product

| | Purpose | Quality Criteria | What a perfect experience looks like |
| --- | --------- | ---------------- | ------------------------------------ |
| 1 | Discovery | **Relevance of results**: Can I find what interests me? | Discover products that are **relevant & interests me** |
| 2 | Discovery | **Quality of results**: Can I trust the results to be good? | Feels like I can **trust the recommendations** |
| 3 | Discovery | **Adherence to brand voice**: Does Discover portray who Gumroad is or wants to be? | Feels like I am **using Gumroad** (i.e. "approachable, clever, trustworthy, knowledgeable, optimistic, supportive, genuine") |
| 4 | Discovery | **Serendipity**: Can I discover something that I was not expecting, but will delight me? | Able to **find specifically what I want to find** |
| 5 | Search | **Effectiveness of search tools**: Do the search tools help me define and narrow down my search effectively? |Able to find **something that delights me, even though I wasn't looking for it**

Another important point is to give the user a **feeling of community, which is the essence of the Creator Economy.** It should not be just a listing site that spits out products from a database based on certain filters.

It should feel alive. It should reflect the ebbs and flows of what's hot and what's not in the creator and audience communities. It should represent the latest (everchanging) state of the Creator economy. Hence, users can find relevant products that they can trust because it is their tribe that thinks it is good. They will also discover things serendipitously which they did not know their tribe was looking at.

With the above in mind, I next assess the user flows.

<br>
---
<br>

# 3. How does Discover work now? :computer:
I have collated the user flows for Discover in the Figma below.
<iframe style="border: 1px solid rgba(0, 0, 0, 0.1);" width="800" height="450" src="https://www.figma.com/embed?embed_host=share&url=https%3A%2F%2Fwww.figma.com%2Ffile%2FIvACdDdqvOEmsHxmMdHBis%2FUntitled%3Fnode-id%3D0%253A1" allowfullscreen></iframe>

For more detailed screen flows with comments, please refer to the Figma below.
<iframe style="border: 1px solid rgba(0, 0, 0, 0.1);" width="800" height="450" src="https://www.figma.com/embed?embed_host=share&url=https%3A%2F%2Fwww.figma.com%2Ffile%2FIvACdDdqvOEmsHxmMdHBis%2FUntitled%3Fnode-id%3D1%253A469" allowfullscreen></iframe>

---
<br>

# 4. How does Discover fare? A UX analysis :crystal_ball:

## **General User Experience** :globe_with_meridians:
### **4.1 Website loads too slowly (16 seconds)**

For most users, their first interactionwith Discover is via clicking the "Everything" category as Discover is the first button on the list. However, the website typically takes 16 seconds before the user sees anything meaningful.


I think that this will result in **high bounce rates at the top of funnel and needs to be solved urgently.**

### **4.2 The category listing page is hard to connect with and extract useful information**
As a user, when I land on a listing page, I can see and understand that these are product cards. However, beyond that, **I do not relate to or have any emotional connection with what I see**. This is not good for a discovery tool as the product results feel foreign to me, and I cannot trust the results as a user.

| | Reason | Suggestion | Sample |
| --- | ---------------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------ |
| 1 | **Visual hierarchy and asset spacin**g is not clear, so my eyes do not know what to focus on | **Redesigned product cards with clearer visual hierarchy, asset spacing, colours**, so users can focus & understand what they need (Title > Reviews > Price > Creator) |  |
| 2 | **Colours are rather flat and plain**, so there is no emotional connection | Redesigned card | Refer to above |
| 3 | Components such as star scores and reviews **do not convey quality effectively** (perhaps because there are just **a lot of stars on the pag**e) | Redesigned card | Refer to above |

### **4.3 Product results do not feel trustable or quality**
When browsing through the products (especially on the main listing), I feel like **I cannot trust that these products are good** (despite star ratings).

A large part of this could be due to the way images are shown. As a user looking to discover or buy digital purchases that are untangible, images are important to help me visualise what the product does and why it is good. This not only helps show its relevance to me, highlight its quality and help me find things serendipitiously.

| | Reason | Suggestion | Sample |
| --- | --------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------- | ------------------------------------ |
| 1 | Images are **off-focus or cut off**, so users cannot tell what it is or why it is good | **Have greater emphasis of images that look better** (e.g. aspect ratio, cutoff approach) |  |
| 2 | Images are **off-focus or cut off**, so users cannot tell what it is or why it is good | **Train creators on best practices** on images and copywriting | Refer to above |
| 3 | **Product card does not have enough emphasis on image** (currently only 50% of card) | **Re-design how the product card looks** (as mentioned above) | Refer to above |

### **4.4 It does not feel like there is a community around the platform**
Community is foundational for the creator economy, but as a discovery tool for the creator economy, the community aspect is lacking from Gumroad Discover. For most users who are used to user-generated content on their apps such as Airbnb, Yelp, Tripadvisor, Google Maps, etc., community-based curation, discovery and feedback about what's good (and what's not) should be naturally woven into the product.

Currently, while there are some elements of social proof (e.g. number of purchases and reviews), the listing page feels pretty static and does not feel alive.


**Suggestions**:
* Increase **visibility of social proof components** (e.g. use of "trending" products, clean-up of star ratings)
* Introduce **themed badges which can be curated by editors first, followed by community in the future**
* **Showcase not just products, but creators too**. This helps users connect with their stories and have a deeper level understanding of their products

With the above suggestions, the category listing page should look like this:


### **4.5 Potential for localisation and personalisation in the future**
As a user, when I clicked into several listings (see Cooking category below), I see some non-English results featured on top which I do not understand. Given that most of our users are from English-speaking countries, this makes the product listing feels less relevant and helpful for users.


In the long run, when Gumroad goes more global, there is potential for **localisation where more localised results (based on region, language or interests) are bumped up based on the user's locale**. This is especially relevant for domains such as cooking and sports, which should increase the relevance, quality, trust and sense of community for the users.

However, in the short run, since I assume most of our users are from English speaking countries, it should be sufficient for us to **ensure that the top 9 featured products are in English.**

Secondly, the **range of categories is very broad, both from a top-level taxonomy and even for sub-categories (tags) within a category**. This variety for users introduces diversity and range, but also irrelevant noise at the same time.

In the future, I would suggest **greater personalisation or suggestions for peripheral products or categories based on what the user has selected**. This increases the relevance and trust for the product, and also increases the serendipity factor for the Discover tool.

## **Specific Gaps** :pushpin:
### **4.6 Placement of search bar is great, but it might be potentially confusing**
I like the placement of the search bar because it is **prominent and natural** in the user's flow. Also, I like the dropdown for easy access to all categories which new users might not be familiar with.

There are however 2 points which I think can be improved.

Most of the copy on site are creator-facing (i.e. "business owner" persona), whereas the search bar is meant for their audience. **For a first-time user who is not a creator looking to monetise, I would have thought the site is not for me and the search bar was meant for creators who want to get paid**.


As such, I would suggest providing some clues to users so that they know what is available and searchable on Gumroad. This would come in the form of a **rotating list of examples (e.g. "Procreate brushes, e-book writing guides, cookbooks, etc.") in 2 areas**:
* **Search bar**: To replace "products" in "Search for products"
* **Main heade**r: Add in rotating examples after "Gumroad makes it easy"... "to sell {{examples}}


The dynamic cycling through of examples would catch the users' eyes and help them better understand what they could discover, hence kickstarting their usage.

### **4.7 Discover Everything page needs to be revamped for better use of space**
The above-the-fold real estate of the Discover Everything page is wasted. Even though the main star of the page should be products by categories or collections below, the categories section is taking up a big part of the page while giving users marginal benefit only.


I would suggest to **reduce the category selection section** and show the rest of the categories shortly after.

Currently, each set of listing belongs solely to "Best selling in {{Category}}" which is rather stale. Moving forward, on the editorial side, I will suggest having **thematic collections** from categories to make it more delightful for users, such as "Asian American artists during AAPI month", etc.

Together with the redesigned product cards and social proof features (as mentioned above), I believe that the main listing page would look more trustworthy, relevant and welcoming to new users.


### **4.8 Search tools are good but can be improved**
The search and filters are useful in helping users narrow down their search, such as the **tags system as well as digital products-focused filters (e.g. file formats)**

However, some improvements can be made to this.

Firstly, when tags are selected, there is no feedback to show the user that the tag is selected, which is an easy front-end fix.

Secondly, I have realised that since digital products are very broad, each category is vastly different from other categories. This creates a **very different decision and narrowing-down process for different types of users**. As such, I believe that it will be helpful to have **category-specific filters for different categories** (such as comedy vs cooking vs sports). This will help users better narrow down to their desired results and find what they want.

<br>
---
<br>

# 5. How should we get to what Discover should be? :airplane_departure:
In summary, revisitng the 5 factors mentioned above, here is how Gumroad Discover fared to me as a user

| Quality Criterya | Score (out of 5) |
| ----------------------------- | ---------------- |
| Relevance of results | 3 |
| Quality of results | 2 |
| Adherence to brand voice | 3 |
| Serendipity | 3 |
| Effectiveness of search tools | 4 |
| Sense of community | 2 |

The Discover tool would be rather **useful for someone knowing what they are looking for**, but **not so useful for someone looking to discover new products.**

Particularly for a community-based domain like the Creator Economy, the **lack of community and social elements in the product needs to be improved** in order to speak to our audience - the creators and their audience.

### **5.1 Short-term Plan (Weeks)** :rocket:

Collating all the suggestions above, here is how I will start making Discover the best tool for discovering digital products in the Creator Economy. I do not believe that it will be entirely on Product to make this a better tool. In fact, editorial and community plays a big part in curating products, training creators and featuring creators.

Over the coming weeks, the first priority is to **ensure that the base product works well before introducing optimisations**. This includes:
1. Reduce loading speed
2. Redesign visual hierarchy of cards
3. Redesign main listing page
4. Other minor frontend improvements (e.g. tags frontend feedback, category tags, home page edits)
5. Editorial: Curate top 9 features to be English
6. Editorial / Community: Establish and share best practices with creators

Next up, we should **make Discover truly a community-powered tool by introducing greater signs of social proof**
1. Add "Trending" indicators
2. Add curator badges
3. Add creator sections in category pages

By this stage, the Gumroad Discover product should look like this:

Coupled with the new [proposed taxonomy](https://hackmd.io/@arentweall/rJ3_t3o5u) , the Gumroad Discover Everything page should look like this:

As we improve the general relevance and trust to the site, this should improve the trust, relevance and serendipity of a majority of users. The next group of enhancements would be.
1. Tags improvements
2. Category-specific filters

The reason why I scheduled this to the last is because these are category-specific (i.e. affecting only small parts of users). Also, I foresee that the new taxonomies need to be sorted out first, before the tags and category-specific filters can be worked on.

### **5.2 Long-term Plan (Months & Years)** :ship:

From a longer run point of view, we can then start pursuing **personalisation** where users are recommended more relevant products based on their click profile. When we have expanded globally and achieved mass in a few regions, we can kickstart **localisation by introducing local categories or even local ranking**. However, since the above has business dependencies and are good-to-haves, they are prioritised towards the end.

If we can increase the quality, relevance of results, as well as a sense of community, more users will use Discover as their go-to place for discovering new creators and products. **The goal would be to become the App Store for digital products - somewhere where you can find relevant things that you trust and are happily surprised by. As more users return to discover new products, I am confident that the Discover-powered GMV will go up in the long run.**

<br>
---
<br>

# 6. Time log :timer_clock:

| | Step | Timelog | Total Time |
| --- | -------------------------------------------------------------------- | -------------------------- | ---------- |
| 1 | Kickoff & planning | 05/06/21 4.15PM-4.30PM | 0.25 |
| 2 | Refine approach | 05/06/21 4.30PM-6.00PM | 1.5 |
| 3 | Flows / material exploration | 05/06/21 6.30PM-7.45PM | 1.25 |
| 4 | Analysis & Writing | 06/06/21 1H, 07/06/21 4H | 5 |
| 6 | [Taxonomy + Editorial Task](https://hackmd.io/@arentweall/rJ3_t3o5u) | 07/06/21 10.00PM - 12.30AM | 2.5 |
| 7 | Editing | 08/06/21 6PM-7PM | 1.5 |
| 8 | **Total time** | - | **12** |