Telecom
Pakistan
Telenor and Jazz
2010 to 2020
Telecom at scale: loyalty, personalisation, offer design and segmentation across a national subscriber base
I worked at Telenor (2012 to 2020) and Jazz (2010 to 2012, Jazz and VEON Group). At Telenor I led a team of marketing specialists across customer base management, voice of the customer, revenue, channel planning, product design and pricing for the post-paid high-value segment. At Jazz I built micro-segmentation models and frameworks for prepaid below-the-line campaigns.
At a glance
+5%
ARPU over a control group, from one-to-one personalisation
8%
engagement for the launched loyalty programme, built with a 10-member cross-functional team
5.4% vs 1.7%
uptake, under-pitched vs over-pitched 1-day bundles
30M+
subscribers covered by micro-segmentation at Jazz
01 · Loyalty and post-paid growth · Telenor
Growing post-paid share in the high-value segment at minimal operating cost.
The problem was to grow post-paid share in the high-value segment without adding operating cost. Bank partnerships and a loyalty programme were the two levers.
- Launched the loyalty programme to 8% engagement, with a 10-member cross-functional team.
- Built nationwide affinity partnerships with banks around co-branded hybrid products to grow post-paid share at minimal operating cost.
- A national bank partner co-funded the co-branded proposition and the post-paid brand promotion.
Schematic · The partnership loop
A partner bank funds a co-branded product, and the product grows post-paid at low operating cost.
01
→
Partner bank
02
→
Co-branded hybrid product
03
Post-paid growth at low operating cost
Schematic, not to scale. The only figure shown on this page for the programme is 8% engagement.
02 · One-to-one personalisation and real-time offers · Telenor
Matching each offer to the customer, in real time.
Offers should follow each customer's product affinity, purchase propensity, channel and price. So I owned the personalisation roadmap and ran it in real time.
- Owned the personalisation roadmap on SAS Campaign Management and RTDM.
- Offers built from product affinity, purchase propensity, channel and price; triggered in real time.
- Delivered across digital, USSD, IVR and retailer channels. Led a squad of 3.
- Result: ARPU +5% over a control group.
Schematic · Real-time decision fan-out
4 customer signals feed one real-time decision, which fans out to 4 channels.
Schematic. No volumes or rates are implied by the line weights.
Chart · ARPU vs control group
Personalised customers earned 5% more ARPU than the control group.
Control group
100
Personalised
105
+5% ARPU vs control group
Source: Relative index, control group = 100. ARPU uplift is measured against a control group.
03 · Offer pitch analysis and experiments · Telenor
Which offers converted better: over-pitched or under-pitched.
Some offers were over-pitched and others under-pitched relative to what customers activated. The team needed to find the best-converting mix of validity, bundle and price.
The work done
Analysed offer presentations against activations by validity, bundle mix and price. Ran validity, price and sale-event experiments.
- Under-pitched offers converted better.
- 1-day bundles: 5.4% against 1.7% uptake, under-pitched against the other group.
- 7-day bundles: 10.6% uptake (under-pitched).
- Longer validity raised uptake by up to 1.3 points.
- An 11.11 sale lifted activations by about 30%.
Chart A · 1-day bundles, uptake
Under-pitched offers converted about 3× better in 1-day bundles.
Other group
1.7%
Under-pitched
5.4%
Source: Internal uptake figures, 1-day bundles. "About 3×" is derived (5.4 / 1.7 = 3.2). Bars scaled to the larger value.
11.11 sale
about +30%
activations
Stated next steps
Portfolio offers
A/B pitching
Conjoint analysis
04 · Urban and rural value study · Telenor
Where to spend network capital.
Network investment had to reach the highest-value subscribers. The study mapped value share level by level to find them.
- Profiled urban and rural data subscribers.
- Mapped value share at district and city-neighbourhood level.
- Prioritised urban areas for network rollout CAPEX in 2018 and 2019.
- Presented at Group Technology and executive forums in Europe.
Schematic · Value share by level
Value share was read at three levels, from country down to city neighbourhood.
Level 1
→
Country
Value share
Level 2
→
District
Value share
Level 3
City neighbourhood
Value share
Schematic. No values shown.
05 · Electricity failures and customer value, Karachi · Telenor
When outages hit, when are customers most reachable, and what do they use?
Outages hit sites unevenly. Marketing needed to know when customers were most reachable and what they used.
Method
Joined site-level outage logs to usage and ARPU. Clustered sites with matching outage timing within 5 km. Segmented every site into one of 26 towns.
- About 42% of affected sites had daily outages.
- Median outage days fell as ARPU rose.
- Data and voice usage rose in outage hours, most for customers unused to outages.
- A live World Cup super-over lifted outage-hour data by 20%.
- Output: an outage-aware view for timing contextual offers.
Schematic · Town count
Every site was segmented into one of 26 towns.
26 towns
Neutral count marker. Within each town, sites with matching outage timing within 5 km were clustered.
Chart · Outage-hour data usage
A live World Cup super-over lifted outage-hour data by 20%.
Outage hours
100
Super-over
120
+20% data during a live World Cup super-over
Source: Relative index, outage-hour data usage = 100. Site-level outage logs.
06 · EasyPaisa wallet portfolio and telco bundle penetration · Telenor
Telco bundle penetration on the wallet, from 17% to 25%.
- Managed a fintech wallet portfolio on the EasyPaisa wallet with app-exclusive and regional offers.
- Grew telco bundle penetration from 17% to 25%.
Chart · Bundle penetration
Bundle penetration grew from 17% to 25%.
Before
17%
After
25%
Source: Internal figures for telco bundle penetration on the wallet. Bars scaled to the larger value.
07 · Micro-segmentation and management analysis · Jazz
Segmenting prepaid customers by the offers they had taken before.
- Built micro-segmentation models and frameworks for prepaid below-the-line campaigns.
- Included an offer-lineage model that segmented customers by the offers they had taken before.
- Business analysis team: produced weekly and monthly management dashboards and business-review presentations.
- Covered 30M+ subscribers (micro-segmentation at Jazz, as stated on the CV).
Illustration · Segments by offer history
Customers with a similar offer history group together into segments.
Illustration only. Dot counts and positions carry no data.
What carried over
The same method runs through all of it.
Segment the base, match the offer to the customer, measure against a control, and put the insight where the decision is made. I use the same pattern in my personalisation and AI agent work today.
Method notes
ARPU uplift is measured against a control group.
Uptake and activation figures are as reported internally.
Outage analysis used site-level logs; clustering radius 5 km.