Operators to pay compensation when broadband is slow

According to the Daily Telegraph, operators will have to pay compensation when consumers suffer slow broadband.

I wonder how this will work. My Post Office broadband can become almost unusable in the evening most likely because the back haul from the exchange is under configured.

Can I look forward to some compensation, or hopefully better performance?

If it's based on a minimum 10 meg then it's going to be impossible to implement. I could barely get that over FTTC so what would happen then? Would ISP's be forced to provide FTTP?

My broadband would almost be free at £75 per fault (think I've had three or four CAB disconnections and one HR fault in four years).

It'll be interesting to see what the operators do if the fault is not their problem - presumably if speed drops due to interference from another property they won't want to pay out!

How this works and the levels of compensation will be out to consultation in 2016.

Aim is more to push Openreach to offer compensation automatically, but for it to work needs retailers to play ball too, and as highlighted already not all 'faults' are down to Openreach e.g. backhaul capacity is often down to purchasing decisions of the retailer or their associated provider.

Also a danger that while Openreach is heavily regulated, the retailers may simply add another 75p to line rental, so that the chunk they pay out from their slice is spread across all users i.e. consider it an insurance payment.

if the regulation is wholesale side only its pointless, if ofcom want to prove they learning their lesson this needs to be done on the retail side.

e.g. its no good openreach pay a CP for missed appointments when that doesnt get passed onto the end user.

My Post Office broadband can become almost unusable in the evening

I think the journalist has jumped to conclusion there I don't think "reduction of service" = "slow".

That may mean one of your services is not working ie either PSTN or xDSL.


I don't think anybody is going to get ££ when the broadband is "only" running slow.

The 10meg target seems to say otherwise. However, until this is all ratified, we can't assume anything.

In reply to a post by Deadbeat:

The 10meg target seems to say otherwise. However, until this is all ratified, we can't assume anything.

There are always a certain amount of get-out clauses that relate to congestion during peak hours.

For example, within BDUK:
Even though the normal superfast speed target is 25Mbps+ or 30Mbps+, that refers to the access capability during uncongested period. The specifications for behaviour in congested peak hour periods end up specifying something like 15Mbps for 90% of the peak period - which obviously allows it to fall below 15Mbps for 10% of the peak period.

You can bet that a USO of 10Mbps will come with similar clauses.

And you can bet that any SLA terms that trigger a "slow speed" payment will come with similar clauses ... but they might allow a period for the ISP to rectify the problem.

You can be absolutely sure that one speedtest result will not be enough to trigger a payment.

You can be absolutely sure that one speedtest result will not be enough to trigger a payment.

And rightly so.

So many times I have seen people moaning about speeds, but investigation finds it due to the punters equipment, duff wireless etc.

And think of it, any ISP who had adslmax as a punter would be skint in weeks !

If that is the case Ofcom needs to get a LOT bigger, since there is a very long tail of small providers.

And paying £4 a month for their broadband...

Given the horrible complexity of actually measuring real world (rather than speedtest site) throughput and difficult in diagnosis this could be massively complex. Just putting it into some contractually enforceable wording is a huge job and not many people would understand the technicalities.

Just off the top of my head, slowdowns can be down to local WiFi congestion, poor household wiring (on xDSL), powerline noise issues, local segment congestion (on cable), local equipment incompatibilities, problems on domestic computers, intermittent faults on copper loops, contention on backhaul from cabinets to PoPs, congestion from PoPs to peering connections, congestion within peering networks, poor performance from end servers, slow DNS servers and hundreds of other things. Working out who is responsible for what is a nightmare, and extracting the required diagnostics yet another. That's even before you start with finding the sort of skilled workforce to sort it out.

I have had a very long career working on very large computer systems, much of it round infrastructure design, capacity planning, performance and analysis. I spent many a long hour working on performance definitions for even apparently simple elements (like enterprise storage infrastructure) and even that was difficult and frustrating, and that's in a relatively controlled environment.

Without getting into the technicalities, there are roughly to approaches to performance. On is using secondary characteristics (basically capacity planning on the basis that if you can remain sufficiently ahead of the demand curve all will be OK) or there is the measurement of primary performance (things like actual latency measurements all the way down an application stack from transactions to disk I/O). The capacity planning process tends to be the approach taken as it's far easier to do in bulk (albeit highly indirect and very broad-brush) and then there's the primary characteristic approach which measures individual performance which is amazing resource-intensive and requires a lot of skilled resource. Trying to micro-manage individual customers with the latter approach is a huge logistical issue, even if backed up with measurement agents. I used to work with major IT systems that, across the enterprise, would quite literally generate 100s of GB of performance analysis data a day just about the performance of the systems. I shudder to think what would be required if anything like this was required for a major ISP.

In reply to a post by Zarjaz:

... And think of it, any ISP who had adslmax as a punter would be skint in weeks !

In reply to a post by WWWombat:

You can be absolutely sure that one speedtest result will not be enough to trigger a payment.

It would help if Ofcom were to define a procedure whereby a customer could establish whether or not an ISPs performance in terms of bandwidth is adequate or not.

Ofcom should also specify what a customers rights are where performance is inadequate.

Since its still a work in progress it sounds great, if it works and does benefit the end user then even better, just would not like to be a shareholder with Virgin Media

Edited by deleted (Sat 27-Feb-16 16:50:00)

Possible ideas I can think of.

1 - Make openreach have it in their t&c that providers have to pass on any compensation to their customers. Providers dont agree? Then they find someone else not openreach. Problem with this is because end users have no direct contact with openreach, it probably cannot be enforced.
2 - Ofcom start accepting complaints when the policy is breached and fine the CP 10x annual end user subscription as a deterrent for further breaches.

Ofcom dont have a clue alot of the time alot of it is down to that they dont like consumers talking to theml, they make it very hard to send them complaints, instead trying to tell people to contact their own CP or another regulating body.