Removing customer lookup from support calls on an e-commerce platform serving more than 20,000 stores.
The solution connected telephony, internal administration and support history around the agent's actual workflow.
20k+
stores on the platform at the time
~400 / day
support calls each day
01
Prepare the context before the conversation starts.
Callers often did not know their account name or registration details. Agents had to identify the customer, find the right store and reconstruct earlier correspondence before addressing the actual issue.
Identification began while the caller waited for an agent. When the call was answered, the relevant customer, stores and support history were already available.
02
A support process spending its time finding the problem owner.
Initial state
Manual caller identification across disconnected context
The phone call, customer administration and previous support interactions did not arrive as one working context. Searching could consume most of the handling time before diagnosis began.
Constraints
The integration had to work inside a mature e-commerce platform
The platform served more than 20,000 stores at the time.
The support operation handled about 400 calls each day.
Caller information could be incomplete or different from account data.
The workflow had to combine telephony with existing customer and support systems.
Product decision
Remove the lookup step instead of making search marginally faster
Using the waiting period to identify the caller changed the process boundary. The agent could begin with prepared context instead of operating another search interface during the call.
My contribution
A product workflow informed by direct support experience.
I combined earlier customer-service experience with hands-on development to solve the operational bottleneck at its source.
Support workflow design
I mapped what an agent needed at call pickup and moved caller identification into the waiting period.
Telephony and context integration
I delivered the connection between incoming calls, customer data, stores and earlier support history.
PrestaShop migration application
Separately, I built an application for migrating online stores from PrestaShop to Shoper.
The measured result concerns the customer-context lookup step. It is not presented as a reduction of total call duration.
Key decisions
Design around the moment the agent needs to act.
The solution treated customer context as part of call setup, not a separate administrative task.
01
Identify during the wait
Caller matching starts before an agent answers, using time that previously produced no support value.
02
Join the operational context
Customer data, stores and support history arrive together instead of as separate searches.
03
Optimize the first useful action
The agent can begin diagnosing the issue immediately after accepting the call.
Delivery model
A workflow rebuilt around the incoming call.
01
Map the support context
Identify the customer, store and history an agent needs before the first question.
02
Connect caller identity to internal data
Resolve likely accounts and gather the available context while the caller waits.
03
Present one working view
Make the prepared context available at pickup and verify it in daily support operation.
Verified production outcome
The conversation could begin with the customer already identified.
~400 / day
supported by the workflow each day
Up to 60% → practically zero
the lookup step was practically eliminated after implementation
The platform and daily-call figures describe the period of the project. No current Shoper scale or endorsement is implied.
What this proves
Product engineering that begins with operational reality.
The case connects user insight, integration design and hands-on delivery to a measurable improvement for an internal team operating at scale.
FairyDeck uses required first-party storage to remember this choice. With your permission, Google Analytics measures site visits, sessions and calendar CTA clicks. Separate consent lets Google Ads record only a successful form submission — never its contents.