Rear view of player batting while playing cricket on field against clear sky

Improving Live Event Coverage Online By Analysing Minute-By-Minute Traffic

In this post I’ll look at how studying minute-by-minute website traffic can help shape live event coverage, using the example of Australia v England in the Ashes.

When I joined BBC Sport as an online production sub-editor, my parents, while excited, thought this wasn’t a job fitting for my level of magnetic personality and charm.

They were right, just not how they intended.

They looked up the charisma scale – online production desk – journalist – chief reporter – on air reporter – presenter – Gary Lineker.

I actually went online production desk – front page editor – senior development journalist – editorial analytics lead.

I then left, but I think hologram was next.

Still, this blog isn’t a deep dive into my personality traits (‘the least Italian Italian you’ll ever meet’ – courtesy of Al; ‘the most boring man on Twitter’ – courtesy of a former BBC Sport editor.)

No, instead this is a post looking at detailed editorial analytics and how they might not be great party chat,but they can lead to some pretty interesting – or useful at least – insights.

Deep Dive Into Data

If you have the liberty to look at data in depth you can find stuff that is really easy to implement, yet makes a profound difference.

So, to get to the point, why on earth would we want to look at minute by minute data and what can be learnt?

When England toured Australia for the Ashes in the winter of 2017/18 we knew at BBC Sport that the live pages would get great browser numbers.

To be honest, the big figures never really excited me – we were a huge sport site, we had the radio rights, we were the natural place people head to. Getting hundreds of thousands of people to the live page was to be expected (though that’s not to say taken for granted).

For the first Test, a review session was booked in at the end of day one to see what could be learnt to shape the rest of the series. After all, there’s no point solely analysing post event when it’s too late to enact any change.

If you looked at the hour-by-hour traffic, you could see that numbers increased – not surprising games started at god awful o’clock. By 9am, as the day’s play would be wrapping up for the day, there would be many more browsers live in the cricket live text than had been the case at 3am.

The hour-by-hour traffic showed a fairly dull, steady increase in traffic. The minute-by-minute showed something altogether more interesting.

At 7.57 there was quite a lot of traffic, at 7.58 there was quite a lot of traffic. Ditto 7.59. I tended to be slightly more precise with the numbers in PowerPoint presentations.

Then, at 8am there was a lot more traffic and by 8.01 a huge amount more – more than 300% more to be precise.

By 8.05 it had largely settled back down to the levels of 7.58 and 7.59.

We would go on to see this pattern repeated every day, but we didn’t know that yet of course. Nevertheless, it was a pattern interesting enough to demand more attention

This posed two questions – why were these surges occurring and then what did it mean and how might it shape editorial.

The first question could be one you pondered over for weeks, or at least demanded a high level of proof to answer with confidence.

Without that luxury, we just came to a sensible hypothesis. People were setting their alarms and then the very first thing they were doing was checking the cricket score – the live text was the most prominent thing on site and so this was where they headed.

Supplementary data backed this theory up -the visits were all on mobile (the phone acting as an alarm, by the bed, instantly picked up and swiped into action…).

The visits were almost entirely new visitors – new in the last six hours at least. They were from people who had a need to check the cricket score, yet hadn’t been seen on the site during the earlier hours of play. Assuming they were asleep seemed pretty fair.

However, how could this be more than just data for data’s sake. The BBC is full of lovely people, but even there people can get a bit salty at nine in the morning when they’ve been up all night and the analytics guy pipes up in a meeting.

A Tough Crowd

Telling them that site traffic was higher at 8.01 than 7.59 might not go down well, unless there was something practical to be learnt from it too.

For day two, we couldn’t suggest anything that would improve any metric, after all no testing had been done, We had only seen a pattern.

What we could do is make changes that improved the user experience, and that surely has to be the priority. Make a site that’s great for the users and the on-site metrics should move in the right direction.

We looked at the live text and noticed something.

By pure coincidence the changeover between commentator shifts happened at about 8am – it seemed unlikely this was the cause for the surge in traffic – punters only wanting to visit the site when text commentator #2 was on duty!

The posts around this time were, dare I say it, a tad self indulgent. You know the type.

7.58. Right guys, that’s me done. I’ll be handing over to Seb – he’ll be here in a sec, he’s just finishing off his breakfast.

8.02. Ha ha ha! Thanks Chris. yeah, lovely breakfast it was too.

8.05. right, how’s everyone doing? Couple of quiet minutes here, Cummins has just bowled a maiden.

Now, imagine you’re one of those many people who have set their alarm for 8am and then head straight to the live coverage. Are these the entries you want to read?

It’s an obvious answer – of course they aren’t. They tell you nothing about the game, the match situation, or any of the talking points. They are just banal banter.

This one insight led to a change that still persists for overnight events. The posts for these sorts of times act as round-ups of the action, with all the key information and talking points.

If highlights are available, these might be embedded too. The 8am post is written in a way that it works as a summary for everyone who is new to the page, this a fundamental change from a problem that is still all too common in live text coverage – the author assuming that readers are following their every word, and not just dipping in and out occasionally.

Putting The User First

The post also has to be a bit before 8am – there is no point publishing at 8am and then seeing a short publishing delay scupper plans. So, at around 7.57 a summary post would go live and this would then remain the top spot for a few minutes.

Follow up posts would also be written as if people had only just joined at 8am, not endlessly referring back to social media banter from the small hours.

This is the sort of analysis I love – it uses in-depth data to change things for the better for the user. It isn’t data analysis with a preset agenda, trying to justify some editorial change or other – it is an investigation and chance to see where ideas lead.

And, as with all data analysis, it comes back to making the site better for users. Did it lead to massively improved engaged time? Not really, there was an increase, but equally there’s only so much you can do. People are setting their alarm and then getting up and going to work, you’re not going to keep them in the live text for an hour.

It is what it is – a small win for the data department and the humble user!

I’m going to tweet about this now.

Most boring person on Twitter? I don’t think so 😉

(oh – if you’d like any of this fascinating insight, wit or level of chat to help with your analytics please do get in touch. Preferably by a text format – I’m crap on the phone!)

Share this post

Share on twitter
Share on whatsapp
Share on facebook
Share on linkedin
Share on print
Share on email

Leave a Reply

Your email address will not be published. Required fields are marked *