UK Operator Sentiment Analysis
- Arnold Kuersteiner and Ky Pham Minh

- Sep 8
- 3 min read
Updated: Sep 9
UK ISP Sentiment Analysis
In this project, we performed a sentiment analysis of news articles related to broadband organizations. All articles are from UK broadband and telecoms news site ISPReview.
We looked at 7,728 articles spanning the last 13 years, covering a total of 200 organizations, which were categorized into themes as well as given a sentiment score ranging from 0 to 10. To do this we made a model combining data modelling and AI to analyse articles, perform the sentiment analysis and relate them to our business data. This was also one of the first projects using our broadband ontology with external data, something we will talk about more later.
The Data
Our source dataset consists of articles, with greater weight towards articles published in recent years. Starting from a modest base of 20-40 articles per month in the early 2010s, the volume grows consistently to over 100 articles per month in 2025.

As is to be expected with the continued investment and roll-out of new technologies in the UK broadband market, Expansion and Rollout was by far the most common theme of articles at 66.8%. "Regulation and Politics" and "Technology" are the next most significant themes, with 10.4% and 8.2% of the coverage, respectively.

BT is by far the most mentioned organization, with more than 2,200 articles, followed by Openreach, Vodafone, Virgin Media, EE, TalkTalk and CityFibre. These established players dominate the media space in terms of sheer volume.

Our model has settled in at an average sentiment score of about 6 to 7, which is where the scores for BT, Openreach, Vodafone and Virgin Media have settled. For smaller operators, there is also more variability in sentiment, while larger operators have more similar scores. We also see that smaller operators generally have a higher ratio between article count and premises count (i.e. they get mentioned more often relative to their size).
The challenger network builder CityFibre leads the top 10 with an average sentiment score of approxiately 7.5. TalkTalk appears among the most-mentioned but has the lowest average sentiment score of the group.



Notable Organisations Sentiment Analysis
While sentiment scores for smaller organisations can vary widely, the score for BT settles in at around the same value each year. The most positive year for BT scores was 2020, though not by much, mainly driven by a large number of articles highlighting successful milestones in fibre broadband and 5G roll-out, for example in Cumbria, Northumberland, Plymouth and Devon. Covid-related coverage was perhaps more positive than one might expect, with mention of support schemes for private and business customers driving positive scores.

CityFibre, the highest-scoring among major broadband operators, benefits from having a disproportionately high amount of news coverage related to their network expansion across the country. Compared to an average of 66.8%, coverage on ‘Expansion and Rollout’ makes up 89.2% of CityFibre’s coverage by article count. Unsurprisingly, sentiment is generally positive for this category; average sentiment scores for ‘Expansion and Rollout’ are 7.3, compared to ‘Pricing and Tariffs’ at 6.5 and ‘Regulation and Politics’ at 6.0. This positive exposure gives CityFibre’s sentiment rating a (deserved) boost versus other operators.

OFNL has had significantly less media coverage than BT and CityFibre. Still, coverage has been noticeably more negative than other organisations in recent years. A big reason is that in both 2022 and 2024, there has been significant coverage of competitors rolling out in areas which are already covered by OFNL (overbuild), which is seen as bad news from the perspective of the wholesale operator. Some examples:
2022:
2024:

Point Topic Ontology
The Point Topic Ontology is a project we launched this year to consolidate our different data sources on broadband markets in the UK and globally. Our goal is to develop a generalised ontology for broadband markets, which can act as a universal data language for humans and LLMs to interact with concepts and data related to broadband, and has all of Point Topic’s business data loaded into it.

By using our ontology, we were able to relate news articles to broadband entities and other market data like footprint coverage and subscriber numbers, which allowed us to perform this analysis.
You can find the full results graphic for our sentiment analysis here.
You can find more about our ontology here.
Thank you to ISPReview for allowing us to use their articles for this project.



Comments