Product design case study · giffgaff broadband
Helping broadband members fix weak WiFi
A network check that shows members what's wrong before recommending a fix.
Overview
Context
Every giffgaff broadband member gets an eero router. When WiFi doesn't reach every room, there was no easy way to fix it with us.
The outcome
Testing exposed a trust gap, which I closed in the next version. It launched to every broadband member, driving incremental eero sales, with member NPS rising over the same period.
01 · Discover
One in three members were already fixing it themselves
33%
had added their own extenders or boosters by day 90
14%
rated their WiFi signal "Very Strong" by day 90
68 → 55
NPS fell between install and day 90
I had to add 3 WiFi extensions to get WiFi in the whole house.
Broadband member, survey response
Source: member surveys at install, 30 and 90 days
Competitor scan
Across the market, extra WiFi kit sits behind a phone call
Agent or diagnostic first
Most UK providers ask members to call, or prove a slow speed test, before sending another device.
Sold as a subscription
Extra coverage is usually a monthly "whole home WiFi" add-on, not a one-off purchase.
The opportunity
Very few let members diagnose and fix coverage themselves, online, in a few taps.
Members told us they wanted us to recommend the right kit, rather than leave them to work it out alone.
Source: review of public help pages from seven UK broadband providers
Business context
The current fix was costing the business
Free devices
Members who contacted support about WiFi were sent an extra eero for free, which we bought and gave away.
Support time
Every fix went through an agent, in a manual process, so help was slow and inconsistent.
Missed revenue
Members who bought their own kit spent that money with retailers instead of with us.
The business goal
Turn a cost into a self-serve sale, while making sure members only buy kit that will actually help.
02 · Frame
From problem to a testable bet
Problem statement
Members with weak WiFi in parts of their home have no self-serve way to understand or fix it with us, so they contact support, buy kit elsewhere or put up with it.
Hypothesis
We predict that letting members check their own network, and showing them what we found, will help them fix weak coverage without contacting us.
We'll know this is true when members complete the check unaided, can explain the result, and trust the recommendation.
How might we recommend the right fix without it feeling like a sales pitch, or like we're watching their home network?
03 · Ideate
Two directions, reviewed with the team
Direction A: straight to the fix
We tell members their WiFi is struggling and offer an extra eero, through an email nudge, a dashboard offer and a buy page.
Team review
We weren't showing members how we came to our recommendation.
Direction B: members run the check themselves and see the evidence before any recommendation.
04 · Design
Direction B: check first, then recommend
Key decisions
- Members start the checkThe dashboard never shows device details unprompted, so it doesn't feel like we're watching.
- Show the evidenceSimple counts of eeros, devices and weak signals back up the result.
- Say what we look atA plain note: signal strength only, not what you do online.
05 · Test
Easy to use, but the fix wasn't fully trusted
10/10
completed the check and the order
6.6/7
ease of finding out what was wrong
5.5/7
useful next time WiFi plays up
4.4/7
confident an extra eero would fix it
Method: unmoderated, think-aloud test on mobile with 10 UK broadband users. Two tasks: find out what's wrong with your WiFi, then follow the recommendation.
What we heard
A paid-only answer felt salesy. The data didn't feel intrusive.
How it felt
7/10
picked "Salesy" to describe it
In their words
"I would like multiple options and seeing the only option being one which costs extra I don't think is a good look"
Our worry
0/10
found it intrusive, and nobody raised privacy unprompted
Iterate
A free first step, before the eero
What changed
- Quick things to tryThree short, free tips: move the eero off the floor, away from other electronics, and restart it.
- Balanced against the goalOne compact card, so the eero stays the main recommendation. I tried a longer version first and cut it back.
Test, round 2
The free first step closed the trust gap
Salesy
3/10
found it salesy, down from 7/10 in round 1
Confidence in the fix
5.8/7
confidence in the fix, up from 4.4 in round 1
Same tasks as round 1, on the iterated design. Ease and completion held at 6.5/7 and 10/10.
Go-live
Live for every broadband member
06 · Measure
A cost became a sale, and members were happier
In the first 90 days after go-live
Incremental sales
>400
extra eero sales through the account, replacing free devices sent by support
Member NPS
+3
points of member NPS over the same period. We can't tie it directly to the check, but it's a positive indication.
Checks run
4.4k
network checks run by members
The hypothesis held: members could see what was wrong, trusted the fix, and bought it without contacting us.
What's next
Where the check goes from here
Proactive alerts
Flag a weak device before the member notices, rather than waiting for them to come and check.
Smarter results
Use more of the signals we already collect, once we've analysed real-world data.
Beyond weak WiFi
Explore how the check handles networks that are down, or problems that aren't WiFi at all.
Reflections
What I learned
- 1
Show your working
Letting members see the evidence turned a cold sales offer into advice they could follow.
- 2
Usable isn't the same as trusted
Perfect task completion hid a confidence gap. Asking how it felt mattered as much as watching what people did.
- 3
Test fears early, and gently
We assumed using network data would feel intrusive. Asking indirectly showed it didn't, without planting the idea.