Premier League Summer Series

The Premier League’s intent is to reach and engage fans not currently viewing PL games via other means of media, specifically, young millennial and Gen Z audiences that don’t watch linear live television. In order to engage this target audience, the PL selected XALT to deliver a direct-to-consumer live streaming experience featuring an innovative format with interactive, in-match features. This is the first step in a process to inform the Premier League’s development of its future Direct-to-Consumer (DTC) offering.


  1. Test, Learn, Incubate in partnership with Starhub.
  2. Directly serve PL fans who were not being engaged.


The XALT integrated solution offered:

  • Geographically-relevant influencers alternative commentators.
  • Hosted from a Premier League web domain, via an XALT (Premier League-branded) player
  • With a Discord integration to offer group chat around the live game¹.

The Premier League identified Singapore as the designated test market and 4 pre-season friendlies being played in the US (The Premier League Summer Series) were streamed on a live basis at 7 AM (Matches 1-3) and 12 midnight (Match 4) Singapore time.

1. XALT’s community team developed and cultivated a new Discord community of over 1k Singapore Premier League fans in just 10 days.

Executive Summary

Technology scaled and met all quality, DRM, and brand safety requirements by the Premier League. The experience worked on both desktop + mobile (android).*

We learned how we can automate to scale and continue to incubate the Premier League’s DTC experience. We’ve developed a list of key insights to help focus the XALT product roadmap.

*note: iOS deployment was not yet available.

Net new fans were gained. 59% of fans were not subscribed to the local Premier League rightsholder (e.g. Starhub), but are watching EPL matches on a regular basis (e.g. assuming Pirated feeds). Overall, 40% of fans only occasionally (35%) watch or have never watched (6%) a match before of which nearly all said they would come back.

The audience was engaged with chat and the interactive features. 85% engaged in the chat, 85% took part in the survey and 50% clicked on screen graphic pop ups.

We attracted and engaged target audience. We built a community of 1.1K fans in 10 days at a 50% lower cost compared to the current in market paid media solutions.

We reached and engaged the desired Gen Z audience that the league was trying to reach but had been unable to via traditional cable and satellite. 100% of fans were under 40 years old and 93% were between 18-35 years of age.

Commentator model drove efficient localization. The commentator pairings, that ranged from a coach and social worker to a podcaster and pop culture influencer were all able to hold audiences for the whole match. Nearly 80% of fans watched the full match on average.

Low cost half time programming from the arena worked! In match 4, when we added the half time programming, we retained 100% of the audience across half time vs. losing 20%-30% during halftime in matches 1-3 (they ultimately came back but keeping them at half time was a goal of the pilot).

Fans showed a propensity to buy. 8% CTR for Starhub promotion (1.5 x higher than average) and 2.3% of fans actually purchased a McDonald’s after it was delivered to commentators.

We tested new ad formats in real time. Directional results only, but in our testing, QR codes did not work, the most effective format was pop up graphics linked to commentator prompts. Our interruptive full screen ads performed somewhere in the middle. This is just the beginning of testing innovative new ad formats for sponsors.

And they want to come back. Both Commentators and fans will come back for the live experience. They also want to keep the Discord community going between live matches, largely via fantasy, sidecasts, ticket giveaways, and transfer window news activity. Fans and Commentators want to help shape the future PL experience.

Performance Insights


Performance metrics for the test were designed to be traced back to business intent, which for the purpose of this document, we understood as follows:

“The business intent is to increase engagement in underserved geographies by bringing new, relevant products to tap and retain an audience not serviced by existing distribution methods.”

Business Questions

  • Can we build a community in a targeted geography and drive new audience acquisition?
  • Can we show we can develop an engaged audience using this new community and a community based live stream distribution product?
  • Are we able to understand more about our community based audience than through traditional over the air or basic streaming services?

Solution Questions

  • Did we bring net-new audiences to the live game that could or would not have watched otherwise?
  • Do we have evidence that the new audience is exhibiting:
    • Engagement in interactivities?
    • Propensity to purchase goods or services?
    • Sustained engagement for the duration of the game?
    • Feedback on the experience?
    • Desire to have the experience again?
  • Can we determine the value and currency of this new community driven audience development?
  • Who is watching with their community, why, and what deep insights are relevant about them?

Analysis and Results

Did we bring and retain net-new audiences?

Cultivated a new Premier League Singapore Discord Community of 1.1K members within 10 days (without Discord bot activation).

  • Attracted GenZ and Millennial audience.
  • Over 59% do not subscribe to StarHub.
  • Approximatly 40% either never watch or only occasionally watch.
% Not Subscribed
Match 1 68%
Match 2 59%
Match 3 60%
Match 4 53%
% Under 25 % Under 35 % Under 45
Match 1 4% 80% 100%
Match 2 13% 77% 96%
Match 3 1% 82% 93%
Match 4 18% 81% 91%

lightbulb icon Key Learning:
The audience that joined would likely not have watched elsewhere.

Did we gain evidence the new audience is engaging in interactivity?

  • Almost 85% actively conversed in the chat.
  • Almost 85% took part in the survey.
  • Almost 50% of the audience clicked on a digital pop up on screen.
lightbulb icon Key Learning: The new community based audience is comfortable fully participating. Therefore the XALT platform activates lean-forward audience.

Did we gain evidence the new audience has a propensity to buy

  • Match 1 & 2 tested QR codes which were not successful. Audience did not engage with QR codes.
  • Match 3 & 4 tested clickable images on screen. Graphic A generated 6x more clicks than Graphic.
  • Match 4 only used Graphic A which generated almost 8% CTR to StarHub.

lightbulb icon Key Learning:
At 8%, the new community based audience shows 1.5x conversion performance v. regular audience. QR codes have no value on small screen devices but should be tested on Smart TVs². Full screen interruptive promotions do not yield increased performance v. smaller integrated graphic display items.

2. Wordstream benchmarks 2022 state 5.3% conversion



Did we gain user feedback on the experience?

Did we gain user feedback on the experience?

  • Audience loved seeing match 1 commentator on site during match 4 for the pre-show, halftime, and after-show.
  • Audience enjoyed the ability to interact with the community via Discord chat and influencers as commentators.

“Hi there. I would like to share the experience I had. I am a PC user and gamer, having a Discord as platform for football is so convenient so I don’t need to shut down my PC and switch to TV. Hahaha Also the commentator was great, very entertaining and keeps the crowd going.”

Could we determine the value and currency of this new community
driven audience development?

  • Initial community seeding (Organic)
    • 8 influencers (2 per match), ~2 reels per influencer = 35 organic posts in IG (primarily), TikTok.
    • 3 awareness-based influencers, 1 announcement each = 3 organic TikTok posts
    • Overall Learning:
      • 20 accounts earned value thru cross-platform promotion.
      • ~652K impressions, 7.6% Engagement Rate, 52.8K Engagements (all organic).
      • Total 85 publications across TikTok, IG, YouTube, Twitter.
      • Effective CPM = $8 (compare to $50 to $100 for influencer campaign; $10 to $20 for regular campaign)³.
      • PL paid 50% less than the market Earned Media Value ($120,000)

lightbulb icon Key Learning:
Influencers loved the experience and were keen on engaging with fans in the Discord community. Giving influencers the official opportunity to commentate allowed us to get lower talent rates (50% lower than market value).

  • Derivative reach (Paid Network Effect)
    • 7 click-optimized Ads ran on Reddit, $873 total spent
    • Overall Learning:
      • 2.6K clicks (paid clicks did not convert to Discord members on days that talent was inactive)
      • Effective eCPC = $0.31 (compare to $1 to $2 (est.) based on US $3.50)

Did we determine who is watching with their community and gain
insights about them?

Over the course of 4 matches we had ~2,000 join requests and estimated between 200 and 250 unique viewers. Overall, the audience that participated in the chat the most were gamers and Web3 community members. Specific additional insights include:

Match 1:

  • Commentators: Jack & Josh 229k combined followers on Instagram
    • Jack: New York based Aston Villa fan with approximately 5% of his audience residing in Singapore (based on IG, TT, and YT poll). Jack drove the majority of the engagement in the Discord chat during the match).
Jack Grimse comments 3
Jack Grimse comments 2
Jack Grimse comments 1
    • Josh: Singapore local influencer and actor who was a semi-pro football player and is followed by a lot of local football athletes. The majority of the audience that joined for Match 1 came from Josh.

lightbulb icon Key Learning:
The balance between a team super-fan and a casual Premier League fan made the commentary feel more accessible to audiences.

  • Tested using commentator 1 (Jack Grimse) as the integrated producer. Our hypothesis was that using talent as the interactive producer would drive a more seamless interaction between the commentators and audience. This proved to be true. He was very good at identifying luls in the game action and used the graphics to drive conversation or find natural moments for sponsorship.
  • Tested using QR codes to promote StartHub. The audience was not receptive to QR code graphics.
  • Tested running a full screen takeover promoting StarHub pre kickoff and during halftime. The audience was not receptive to interruptive ads and did not interact with the link.

Match 2:

  • Commentators: Dew & Ebi 38k combined followers on Instagram
    • Ebi: Local radio show host and professional singer/actor. Huge fan of Premier League and current subscriber to StarHub but no technical football knowledge. Not an avid Discord user or gamer. We wanted to test local radio voices and more “produced” style social content.
    • Dew: Local clothing designer and host of local podcast. Versed in streaming and familiar with Discord due to his gaming community. Most of the audience watching Match 2 came from Dew’s community but really enjoyed the Dew + Ebi pairing.

lightbulb icon Key Learning:
Produced social content did not perform. While Ebi had a strong football fan following and local radio presence, his content was not efficient in converting audiences. Dew was native to Discord and a lot of his followers were already on Discord which helped increase engagement.

  • Resolved technical issues from match one and received a lot of positive feedback about the quality of the stream.
  • Activated the Discord chat bot activation which commentators really liked and said it made it easy to engage with the community and keep conversation going.
  • Tested doing clickable stickers for Starhub promotion which drove more engagement than the QR codes (4.08% CTR)
  • Tested 2nd full screen takeover with clickable sticker vs QR code which generated a 2% CTR.
  • The first giveaway was announced. The winner came from a large Singapore based Web3 community which drove a large jump in Discord members (over 300)

Match 3:

  • Commentators: Al-Ansari & Coach Habil 11.5K combined social following.
    • Al-Ansari: (1829 Followers) Local social workers from the Children’s society and active in the local Singapore football community. His involvement allowed us to get a social post on as added value which drove new audiences to Discord. He used his personal WhatsApp network to convert audiences (121 WhatsApp messages sent). We expected more audiences to convert from his community. The overall feedback we received from him was that the students said it was too early and a lot of them did not want to wake up at that time.
    • Coach Habil: (10k Followers) Local football coach and the Director of the Albirex Singapore Football Academy. The audience immediately recognized him. There was a lot of conversation about the position he played and a lot of people were curious to hear his predictions about players and overall future fixture outcomes. He also used his WhatsApp network to announce commentating and received a lot of positive feedback about the event but a lot of his network said they struggled to get in because they could not figure out how to access Discord day of.

lightbulb icon Key Learning:
The audience on Discord was very engaged with the talent however those who were not native to Discord struggled with accessing the experience. A lot of the women from Coach Habil and Al-Ansari’s network said they were not able to access the stream because they could not set up their Discord account. In future activation, we would aim to activate micro communities like this at least one month prior to the event. 33% of the audience said that this pairing was their favorite because they loved the local feel. There was a lot of talk about organizing local football friendly matches so we would recommend using pairings like this in the future but giving non- Disord users the options to authenticate via traditional login.

  • Tested with talent that was located in Japan and Bali during the event. The tech worked, audience was able to see and hear both commentators very well.
  • Tested running the same Starhub graphics that we ran in Match 2 (same time slots) without having the talent announce it. The audience did not click on the graphics when the promo was not announced by talent.

Match 4:

  • Commentators: Chow & Dan Lim 57K combined social following. Chow: (55k Followers)
    • Chow was active in the chat during all four matches which helped get audiences familiar with her. The community was excited to see her commentate. The audience really enjoyed “teaching” her and those where we casual fans or had never watch a football match said they really liked her questions and Dan’s ability to break everything down for her and educate her.
    • Dan Lim: (2.5K Followers) Dan is one of 4 host from the Daily Ketchup and a huge football fan. He had great technical knowledge of the sport and drove a lot of conversation around fantasy football during the match. He was able to launch all graphics on his own and used the player stats graphics during luls.

lightbulb icon Key Learning:
33% voted for this pairing. The audience loves having one person who is an expert in the subject matter and has technical knowledge and someone who can add color and humor like Chow.

  • Tested having talent prompt conversation about food. We had a community member send McDonald’s to the commentators which prompted a lot of engagement and led to 2.3% audience members ordering McDonald’s.
  • Tested having player graphics with stats which the audience liked. Dan also liked using the stats graphics as content during lulls.

lightbulb icon Key Learning:
Tech provides a solid mechanism for alternative commentary, testing new advertising integration, and integration with Discord communities. The Discord chat was a great engagement tool and helped talent keep conversation going for the duration of each match.

Scroll to Top