Turned Insights Into Revenue
Turned Insights Into Revenue
2023 • THUMBTACK • 6 MONTHS
2023 • THUMBTACK • 6 MONTHS
Thumbtack connects homeowners to service pros and helps pros find customers in homeowners. In 2023, pros wanted more jobs, but didn’t know how to improve their targeting. I led a strategy to surface actionable insights across 3 product surfaces. I led end-to-end design strategy from problem framing to MVP, facilitated workshops and prioritization during a PM gap, shipped 2 experiments, and scaled our learnings across teams. The impact included: +4.3% leads per pro and +1.08% global revenue.
Thumbtack connects homeowners to service pros and helps pros find customers in homeowners. In 2023, pros wanted more jobs, but didn’t know how to improve their targeting. I led a strategy to surface actionable insights across 3 product surfaces. I led end-to-end design strategy from problem framing to MVP, facilitated workshops and prioritization during a PM gap, shipped 2 experiments, and scaled our learnings across teams. The impact included: +4.3% leads per pro and +1.08% global revenue.
Pros struggled to target leads effectively
Pros struggled to target leads effectively
Our executive guidance for 2023 was to unlock growth opportunities for pros. To start, I led an experience audit and a cross-functional and cross-team workshop to ideate opportunities, which led one key pain point: pro targeting. Just like online ads find you through audience targeting, Thumbtack pros can reach homeowners through targeting, which includes: job types (house cleaning> deep cleaning), travel areas (for areas they want to get leads from),and availability (for days and times to get lead for).
With Research and Product feedback on targeting, we learned pros:
Set-it-and-forget-it for targeting
Have low targeting comprehension
Expected support
Have uncertainty about demand
Our executive guidance for 2023 was to unlock growth opportunities for pros. To start, I led an experience audit and a cross-functional and cross-team workshop to ideate opportunities, which led one key pain point: pro targeting. Just like online ads find you through audience targeting, Thumbtack pros can reach homeowners through targeting, which includes: job types (house cleaning> deep cleaning), travel areas (for areas they want to get leads from),and availability (for days and times to get lead for).
With Research and Product feedback on targeting, we learned pros:
Set-it-and-forget-it for targeting
Have low targeting comprehension
Expected support
Have uncertainty about demand
I organized workshop opportunities in a RICE table. Without a PM, I led prioritization with team leads (Eng and Data Science, Research).
I organized workshop opportunities in a RICE table. Without a PM, I led prioritization with team leads (Eng and Data Science, Research).
We roadmapped three experiments to address the sub-problems that led to the umbrella problem and hypothesis statements.
We roadmapped three experiments to address the sub-problems that led to the umbrella problem and hypothesis statements.
To ensure user satisfaction, I created design principles to direct solutions. Be insightful: to inform decisions. Improve comprehension: to enhance pro success with intuitive tools. Foster growth: to offer long-term value. Build trust: to deliver clear insights for with reliability.
To ensure user satisfaction, I created design principles to direct solutions. Be insightful: to inform decisions. Improve comprehension: to enhance pro success with intuitive tools. Foster growth: to offer long-term value. Build trust: to deliver clear insights for with reliability.
Experiment 1: Service Insights
Experiment 1: Service Insights
The problem was that pros are unsure which services to add. If pros add the wrong services and homeowners reach out to them for those services, pros will pay for jobs they can’t do and homeowners will be dissatisfied. Our hypothesis was that, providing service insights will guide pros to better service targeting and drive more leads. The impact was +70% new leads from added services, +2% leads per pro, and +0.6% revenue. We also learned local demand insights can drive leads. With that, we shipped this UX.
The problem was that pros are unsure which services to add. If pros add the wrong services and homeowners reach out to them for those services, pros will pay for jobs they can’t do and homeowners will be dissatisfied. Our hypothesis was that, providing service insights will guide pros to better service targeting and drive more leads. The impact was +70% new leads from added services, +2% leads per pro, and +0.6% revenue. We also learned local demand insights can drive leads. With that, we shipped this UX.
In the baseline, pros had access to a services page, where they manage their services. 70% of services were added in onboarding. Only 30% of services were added from the services page.
In the baseline, pros had access to a services page, where they manage their services. 70% of services were added in onboarding. Only 30% of services were added from the services page.
My leading question was, what’s the level of lift required for pros to add new services?
My leading question was, what’s the level of lift required for pros to add new services?
Explorations within the services page (discoverable) or in the add-service flow (timely).
Explorations within the services page (discoverable) or in the add-service flow (timely).
The final solution worked because of relevant services (recommendations algorithm) and local demand insights, per feedback from concept testing.
The final solution worked because of relevant services (recommendations algorithm) and local demand insights, per feedback from concept testing.
Experiment 2: Mismatch Insights
Experiment 2: Mismatch Insights
The problem was that when homeowners attempt to reach out to a pro who doesn’t offer the right service, it creates a mismatch, and pros don’t know it happened. If homeowners are converting from viewing pros, pros may fill the product gap with the assumption that Thumbtack is not good enough to find customer leads. Our hypothesis was that, providing mismatch insights will inform pros of high demand targeting and drive more leads. The impact was +2.3% leads per pro, +25% email engagement, and +0.48% revenue.We also learned FOMO psychology works best with specific data versus directional guidance. With that, we shipped this UX.
The problem was that when homeowners attempt to reach out to a pro who doesn’t offer the right service, it creates a mismatch, and pros don’t know it happened. If homeowners are converting from viewing pros, pros may fill the product gap with the assumption that Thumbtack is not good enough to find customer leads. Our hypothesis was that, providing mismatch insights will inform pros of high demand targeting and drive more leads. The impact was +2.3% leads per pro, +25% email engagement, and +0.48% revenue.
The problem was that when homeowners attempt to reach out to a pro who doesn’t offer the right service, it creates a mismatch, and pros don’t know it happened. If homeowners are converting from viewing pros, pros may fill the product gap with the assumption that Thumbtack is not good enough to find customer leads. Our hypothesis was that, providing mismatch insights will inform pros of high demand targeting and drive more leads. The impact was +2.3% leads per pro, +25% email engagement, and +0.48% revenue.We also learned FOMO psychology works best with specific data versus directional guidance. With that, we shipped this UX.
In the baseline, homeowners attempted to contact a pro by entering request form details only to face a mismatch wall. From this moment on, homeowner sentiment begins to feel disappointed because their invested effort of looking for a pro just hit a blocker.
In the baseline, homeowners attempted to contact a pro by entering request form details only to face a mismatch wall. From this moment on, homeowner sentiment begins to feel disappointed because their invested effort of looking for a pro just hit a blocker.
My leading questions were around three layers: Attention, where does the pro already engage for services? Action, where in the app could the pro act on insights? Time, when could the insights be the most impactful?
My leading questions were around three layers: Attention, where does the pro already engage for services? Action, where in the app could the pro act on insights? Time, when could the insights be the most impactful?
Wide explorations were key to find the right solution to ensure insights felt empowering because mismatches would introduce a new mental modal over just coming for new leads.
Wide explorations were key to find the right solution to ensure insights felt empowering because mismatches would introduce a new mental modal over just coming for new leads.
This final solution worked because we believed surfacing recommendations after new leads could facilitate pros to focus on new opportunities.
This final solution worked because we believed surfacing recommendations after new leads could facilitate pros to focus on new opportunities.
Experiment 3: Geo Insights
Experiment 3: Geo Insights
The problem was that when pros set travel area targeting, they have to browse a long list of counties and popular cities. This mattered because if pros are not familiar with a long list of areas, they may not set strong targeting. Our hypothesis was that, improving the usability and providing insights will guide better travel area targeting and drive more leads. The impact was +70% travel page engagement. However, the experiment trended -5% revenue per pro (did not validate hypothesis), so we stopped this experiment early because we learned that low demand signals was bad for retention and expansion of new areas.
The problem was that when pros set travel area targeting, they have to browse a long list of counties and popular cities. This mattered because if pros are not familiar with a long list of areas, they may not set strong targeting. Our hypothesis was that, improving the usability and providing insights will guide better travel area targeting and drive more leads. The impact was +70% travel page engagement. However, the experiment trended -5% revenue per pro (did not validate hypothesis), so we stopped this experiment early because we learned that low demand signals was bad for retention and expansion of new areas.
In the baseline, pros had to browse this list of counties or popular cities to select where they wanted customer leads from. Not all pros had the time for browsing or familiarity of their geo.
In the baseline, pros had to browse this list of counties or popular cities to select where they wanted customer leads from. Not all pros had the time for browsing or familiarity of their geo.
I started with a usability evaluation and competitive audit. I grounded my audit with Jakob Nielsen’s 10 heuristics, but with a growth-focused lens for conversion.
I started with a usability evaluation and competitive audit. I grounded my audit with Jakob Nielsen’s 10 heuristics, but with a growth-focused lens for conversion.
I moved to concept testing to narrow down on a concept and finalize with usability testing.
I moved to concept testing to narrow down on a concept and finalize with usability testing.
This final solution worked because we believed travel area targeting could be powered by an interactive map for targeting precision. We also believed preserving the area list experience could soften the impact of the redesign while testing the new UI.
This final solution worked because we believed travel area targeting could be powered by an interactive map for targeting precision. We also believed preserving the area list experience could soften the impact of the redesign while testing the new UI.
The final solution for desktop.
The final solution for desktop.
Scaling beyond the pod
Scaling beyond the pod
While the travel area experience underperformed, 2 of 3 experiments outperformed. With more external testing, we might’ve caught the long-term risks of the map redesign early. But our scalable insights pattern and cross-team enablement made the strategy a long-term win.
What I did to scale the impact:
Enabled adjacent team to launch the new Pro Performance Page using experiment learnings and wireframed concepts.
Positioned recommendations as a growth system to be modular, scalable, and trusted. For this, I created Thumbtack’s first recommendation card guidelines, which were adopted across multiple teams and handed off to the design systems team.
While the travel area experience underperformed, 2 of 3 experiments outperformed. With more external testing, we might’ve caught the long-term risks of the map redesign early. But our scalable insights pattern and cross-team enablement made the strategy a long-term win.
What I did to scale the impact:
Enabled adjacent team to launch the new Pro Performance Page using experiment learnings and wireframed concepts.
Positioned recommendations as a growth system to be modular, scalable, and trusted. For this, I created Thumbtack’s first recommendation card guidelines, which were adopted across multiple teams and handed off to the design systems team.
The wireframes that led to the new performance page.
The wireframes that led to the new performance page.
Other recommendation cards UX and guidelines I handed off to design systems team.
Other recommendation cards UX and guidelines I handed off to design systems team.
You made it. What's next?
You made it. What's next?
© 2025 LUIS YUJA
© 2025 LUIS YUJA
© 2025 LUIS YUJA