PhantomBuster

Martin Tapia
Christian Seidel
Damien Neuberg
Karim Riahi

2023
Senior Product Designer

PhantomBuster is a lead generation and outreach tool for sales and marketing professionals. The product is made up of a broad range of automations, primarily for social networks like LinkedIn, Sales Navigator, Google Maps and Instagram. These automations are called Phantoms. Each Phantom performs a set of actions on behalf of the user, saving them time on lead generation and outreach activities.

How do you explain session cookies & rate limits to non-technical customers?

I joined PhantomBuster to stand up a new squad, alongside a new PM. Our mission was to improve the experience of how users connect to their social network accounts via a session cookie.

PhantomBuster uses session cookies to connect to a user’s account, without the need to request their login credentials. A session cookie is a string of numbers and letters. While visiting a site, this piece of data is sent from a website to your web browser. Session cookies are designed to be temporary.

Every time you log into your social network account, a new cookie is created for that session. If you log out or are disconnected, the session cookie expires.

Automations are closely monitored by the social networks. These companies implement rate limits to encourage human interactions and limit bot activity on their platforms. Rate limits are a strategy for limiting network traffic. It reduces the strain on web servers and puts a cap on how often someone can repeat an action within a certain timeframe, for example, login attempts.

An SSO solution provides a clean user experience, however, in the context of automations, this type of login request would flag the user’s account for suspected bot activity and potentially lead to a shadow ban. The challenge for our squad would be to improve the user experience given the hard constraints of rate limits.

Conducting qualitative research to validate business assumptions

To understand the nuances of the challenge, I conducted some research. The Product team had an existing research repository which consisted of ±3 years worth of user interviews. Although these interviews were related to other parts of the product, I knew there were insights related to cookie management, hidden in the research repository. I also analysed feedback gathered by the Customer Success team from customer service tickets, which were related to cookie management.

To make the collation of insights easier, I migrated the research repository and Customer Success feedback from Notion to Dovetail so that I could tag and group findings related to the topic of identity management, cookies and rate limits. Three key themes emerged from the qualitative research:

Session cookies are a foreign concept for new users. As they set up their first Phantom, they are pushed towards lengthy documentation. There is a steep learning curve during onboarding and the cookie system feels clunky.
All users want PhantomBuster to provide better guardrails for safe automation on the social networks. They want to maximise the number of leads and shorten the time it takes to generate leads.
Some users avoid any kind of outreach or outbound automation tools because they classify it as spam and feel they could be toe-ing a grey when it comes to GDPR compliance.

Data analysis to understand the impact on acquisition, activation and churn.

I wanted to assess whether the complexity of using session cookies had an impact on acquisition, activation or churn rates, so I used Amplitude to generate charts.

Acquisition: Conversion rate from Free Trial to Paid Plan

Of the free trial users within our ICP who successfully launched a phantom, only 1.22% purchased a paid plan within 14-days. The top automations used by these users were extraction Phantoms – automations which scrape enormous amounts of data from publicly available sites. This suggested that users may be trialing the product to complete one-off extractions or aren’t experiencing the value of the product during the free trial period.

Activation: Completion rate of a user’s first Phantom

76.9% of free trial users successfully finished at least one phantom during the trial period. Almost a third of the users who create a phantom experience an error, but most are able to troubleshoot the error and finish the phantom. 59.5% of all errors were cookie-related errors.

Customer churn

The top reason for customer churn was due to insufficient need for the tool (30.9%). This was followed by pricing (20.5%). Only 4.6% of churn was attributed to cookie management and shadow banning.

Of the free trial users who successfully launched a phantom, only 1.22% purchased a paid plan.
The top reason for customer churn was due to insufficient need for the tool (30.9%), followed by the pricing the product (20.5%)
Almost a third of the users who create a phantom experience an error. 59.5% of all errors were related to session cookies.
20.4% of those who signed up and validated their email, successfully launch a phantom.

Shifting our squad’s focus towards conversion

I created an opportunity tree to categorise the problematic themes (and the proposed solutions) I had surmised from the research. The purpose of this map was to create clarity around our squad’s focus and form the basis of a roadmap. I discussed the themes presented in the opportunity map with the Head of Product and the rest of the Product team. I learned that the key metric our squad should move the needle is the acquisition and activation rate.

It became clear to me that while the experience of using cookies was not an ideal user experience; this problem was not the most pressing issue for the company. The significant drop in conversion from free trial to paid was not currently being tackled by any of the squads. I was not convinced that working on cookie management would significantly impact our activation or acquisition numbers. Our squad negotiated with the Head of Product to shift our focus towards the conversion problem. We agreed to move ahead with the core mission of our squad, whilst also implementing smaller experiments to boost conversion.

Designing an identity management to manage multiple session cookies

PhantomBuster doesn’t provide solution for users to store their session cookies and reuse them. I worked closely with the CTO, lead developer and PM to develop an identity management system, which would store cookies in a database at an organisational level, rather than at a per Phantom level. We would also be able to attach an identity to the provided session cookie, allowing users to identify which accounts they are using and the last time the account was used by a Phantom.

I designed several iterations of the account management dashboard and the "connect to" step within the Phantom setup. The lead developer documented various technical solutions for the identity management system and we discussed the range of scenarios that the solution needed to cater for. We decided to implement this solution for LinkedIn Phantoms only, with the view to expand the solution for other social media platforms in the future.

Users are prompted to install the PhantomBuster extension to retrieve their LinkedIn session cookie automatically.
The browser extension retrieves the LinkedIn session cookie and displays the name of the user.
An account is created and stored for repeat usage.
Manually adding a LinkedIn session cookie, when setting up automations on behalf of a client or colleague.
Managing various types of accounts that have been stored by PhantomBuster for repeat usage.

The value of digging into the data before launching into design and development

As the project moved forward, free trial to paid conversion came into focus company-wide. The work on the identity management system was put on hold, while our squad pivoted towards focussing on experiments that would shift the needle on conversion.

The identity management system was a complex project to work on due to the number of user scenarios that needed to be considered and the restrictiveness of rate limits and session cookies. Simplifying this complexity was a challenging UI/UX puzzle.