MD Portfolio

case study

Automating Profile Tag Selection

Our team improved the Expert Profile by implementing an optimized tagging system and revising the matching algorithm. This allowed Experts to have more control over excluding specific tags and gain access to better matches and, consequently, better earning opportunities. 

We suggested high-demand skills and industries to provide Experts with business intelligence on supply and demand, thereby improving overall matching opportunities for experts. 

Our aim was to reduce expert rejections and increase expert earning opportunities in their service lines.

What problem were we solving?

  • The success of the Matching Algo is only as good as the data it’s fed
  • Experts were not able to indicate their preferences when it came to the type of work they were willing to pitch towards. Not having this ability caused a major business problem when the algo, used to match expertise with customers, was unable to accurately predict the Expert’s favorable response.
  • The failure led to a large number of unresponsive experts
  • No one was available to take on new client work.

Figma prototype demo

Impact and ROI

Within the first 45 days, we successfully reduced matching rejections, leading to a significant increase in the pitching rate of matched experts. 

  • Attained 80% Product Adoption within the first 45 days
  • Rejection Reason 1 – The job is not a fit for my skill
  • We reduced usage from a monthly average of g mid-twenties to low teens

Our primary business objective was to effectively enhance the matching predictability for Expert interest,, maintain a high net revenue retention rate, and increase the median earnings of each expert.

Constraints & Trade-offs

Garbage-in, Garbage-out

The data that was being utilized for tagging was extremely unreliable due to the rocky nature of a startup. The company did not put parameters in place at the beginning ultimately allowing users to custom enter tags. During our audit, we discovered duplicates, typos, and internal tags. Cleanup of the data was going to delay the project by half a quarter.

 

Leadership Weigh-in

During an impromptu presentation to leadership, suggestions for alternate technology and user experience were recommended. The suggestion required a redesign of the feature along with consideration for further automation within the workflow.

 

Matching App Updates

Before our feature was able to realize its success we needed the matching app to update its source for Expert profile tags. Our new system clearly bucketed the two; likes vs dislikes. We needed the algo to scan the list prior to making a match. If the algo was not updated the Expert would potentially receive matches for tags they are not interested in and have indicated as such within the new UI.

UX Insights

  • Experts are clearly stating what they do not want to work on
  • Experts prefer having industry experience related to the proposal match
  • Experts are not interested in working on proposals that require specific software (i.e., Quickbooks Desktop)

Determining User Needs

We needed all of the Experts to have an accurate profile so that our algo can make positive predictions in regard to matching with new client work.

user goal – intuitively complete the form, thus creating a more robust profile
business goal – 100% feature adoption by the upcoming seasonal quarter peak

In order to optimize new opportunity matches – Expert profiles need to be accurate in regard to Expert preferences.

Experts need the following to happen for their preferences to reflect their interests –

  • The ability to indicate which tags they are not interested in pitching towards
  • The ability to easily modify tagging preferences for profile flexibility
  • The ability to understand which tags are being matched and which are not – have a single source of truth

 

Overview of tasks –

  • Deep dive into the data around rejection reasons – qualitative and quantitative
  • Validated Business problems with full-funnel data
  • Affinity mapped the written feedback to better understand the larger contributing factors
  • Determined insight, and documented learnings in Confluence; presented to product team trio and director
  • Sole designer for the Matching App team and Expert Experience team