New tariffs

OK, this is on my personal blog as you can post comments and ask questions. But basically AAISP are looking at new tariffs.

However, we have a slight difficulty because we have so many choices. I am not sure of the best way to offer simple clear and understandable tariffs which are fair.

Basically, we have three types of line:-
1. 20CN, and these are gradually being moved to 21CN over several years
2. 21CN
3. BE

The lines have different base costs, with 21CN being lowest, 20CN costing maybe 30% more, and BE costing a lot more (about 4 times a 21CN line).

The lines have different usage costs now, and this will be even more the case from January. BE lines are the lowest. From Jan 21CN lines will be around 6 times as much as BE line and 20Cn lines around 15 times as much as BE line.

So we need to dream up tariffs and options that fit and handle people with multiple different types of lines, but are understandable.

For a start, I am thinking we separate line cost and usage. This also helps with min terms on BE lines (6 months) as that could be line cost only with usage stopping when the line is stopped.

However, usage is complicated. Current ideas include :-

1. Charging as now, based on GB, based on time of day, and preset usage levels and excess charging, but different rates for different types of line.
2. Having usage levels, e.g. 1Mb/s committed with bursting but with automatic limiting after an hour or two to ensure usage never averages more than committed rate and so no risk of excess usage charged
3. 95th percentile billing with a base commit level and higher rate over charges.

All are possible, but we don't know what people want to do.

It is also not clear how best to handle bonded lines with say 21CN and BE where the usage is different but we in effect have to commit to 21CN side for the total to allow for BE breaking.

So, I am open to suggestions on this.

Timescales are that we would probably launch BE lines on some sort of new tariffing system in October, but the 20CN/21CN lines on a new tariffing January.



  1. Personally, I like the simplicity of separate line cost, plus usage charging as now. I can see the appeal for some users of option 2, but I'd prefer the flexibility of saturating my link if I need to for hours, or days, and just be careful for the rest of the month - I'd get jumpy with the thought of "automatic limiting" lurking around the corner...

    Have to confess I'm not sure I fully understand option 3!

    Out of interest, how does BT/Be charge for usage - is it by GB (and is that time-of-day based too), rate, or what?

  2. At various stages we have to pay for capacity which is basically the peak usage, so normally business day (although some evenings are high now). This means if we can encourage usage to be at off peak times then that is also useful to reduce overall costs. So tariffs are not just a matter of costs+profit but also encouraging sensible usage throughout the day.

  3. Given that unlimited is not going to happen and throttling is a b-a-d idea, we quite like the present system though I'm a little flumoxed how to read the usage stats now we have a Be line bonded with two BT lines. Maybe you'll need to offer usage in "AAbytes" which equate to 2.5GB on 21CN, 1GB on 20CN and 15GB on Be, or something along those lines (though is it fair to penalise the customer for being in an area not yet on 21CN - BT seem to think so).

    The time of day works for us too - we do housekeeping evenings and weekends which can use a fair bit in one go (14GB so far today for example) and the Windows Software Update Server is set to synchronise after 2am during the unmetered period as it can suck a couple of Gigs in a night.

    Like Pete we need to be able to max out the link for a few hours now and then and don't want to have to worry about complex cost implications.

  4. Yes, making the tariffs "units" not GB, and saying how many GB each type of line clocks up is a way to manage it, yes.

  5. I'm in favour of the current way of working... the lines run max speed but you pay for the amount transferred, with either a different price per GB per line type (20CN/21CN/BE), or an abstract unit that translates into a different amount depending on the carrier... problem with the later is being able to easily explain that to people!

  6. I am thinking some sort of AAISP "units" may be the simple way of doing this.

  7. For the BE+BT case I'd like to see something where the BT line is treated as an emergency backup. I'm not particularly interested in bonding (since BT are offering to half the speed of my line when they get round to 'upgrading' it to ADSL2+ bonding zero onto the BE line is pointless), but would like to be able to get to the status pages and make a VOIP call to support@ if the BE line has problems. As such I'm quite happy for the BT line to be limited to 150K or such like, therey meaning you don't have to commit to BT for the same bandwidth.

  8. I actually like the 95th percentile option, but I think if you did it you'd spend half your life explaining how it works to people... so from a commercial point of view might not be the best.

    I think you're being overly pessimistic matching the Be capacity on BT. Be is much more reliable anyway, and accepting BT as a second best failover means that even if the worst happens and every Be/BT user switches over to BT at once, a slower/more contended service is acceptable for the short period that this would be required.

    Option 2 sounds good to me. I really like the idea of a system where it's monitoring usage and throttling as needed to prevent overusage automatically.

    I believe it's been done in the past where it was a toggle, so you can switch into burst mode where it works somewhat like it does now.

    If you combine that with 2 committed rates - a peak and off peak one (it'd be very hard for throttling to have to kick in during off peak, unless you were really hammering it) I can see that working well.

  9. Both 95th% and some automated shaping suffer the same issue - they do not encourage usage to be off peak and balance usage throughout the day.

    I am thinking we change tariffs to be "units" based. So we start with 20CN lines 9am-6pm Mon-Dri being 1GB per unit. Then evening/weekend can be say 100GB per unit and 2am to 6am can be 1000GB per unit. That means trade off between evening and daytime which some people wanted.

    Then we can make BE 10GB per unit daytime, and 21CN say 2GB per unit daytime from Jan maybe...

    Maybe we can have some automated user controlled clamping 9am to 6pm for people too?

  10. I like the sound of the 'units' approach - it keeps it nice and simple, and makes it easy to see the price differences between line types.

  11. I too like the 'units' idea, it will also allow people to see where their money go's, as in, if BT pulled their finger out I'd get more Gb per Unit, more download per pound. Says he on 20CN :(

  12. Having said I wouldn't like throttling (in the context of business lines) Tony raises a good point:

    > I really like the idea of a system where it's monitoring usage and throttling as needed to prevent overusage automatically.

    If you're doing a major overhaul, an option to automatically throttle when some percentage of prepaid usage is reached would be nice for when leaving the home connection in the hands of teenagers (or for small businesses when the boss doesn't want any surprises).

  13. I'm a home user, the tariff I have now suited my perfectly up to recently.
    Limited use during the day because i was at work and essentially unlimited use off peak (Given that my total usage for a month never exceeded about 20GB). A large percentage of my daytime useage was on days off or when I got home early from work and downloaded something without thinking.

    But after being made redundant that tariff really doesn't suit me any more until I find a new job as both my total and daytime usage have gone up a lot. I would like the faster upload speeds that may be possible on BE subject to price of course.

    So option 1) May or may not suit me. Depends on the rates and time zones.

    I worry about option 2, my usage tends to be very bursty. A low level of usage most of the time but constantly. With the occasional large usage. I can see myself seeing a game I like the look of and downloading 3GB off peak and then the next morning finding myself rate limited while wanting to download something I actually *need*. Most of the time that won't happen but really I want a line where if I want to download something I know that what I did the night before isn't going to prevent me from doing so.

    Option 3) just doesn't work for me. Even if it was cheap and useful I need to be able to predict the costs and this would make it very hard to do so.

    I guess I'd go for option 2 but reserve the right to switch back to option 1 if it didn't work out for me.

    What I'd ideally like is this -

    To pay line rental and usage separately and to have two usage options -

    1) A line which has essentially unlimited usage but which is rate limited in "real time" to suit the ISPS needs. So when the ISP was busy they could slow my line down and when it wasn't busy I could get my full line speed. Having a line on this basis would help the ISP control their costs and so ought to be relatively cheap.

    2) The option to buy 1GB usage credits and to be able switch on clueless to using them. So when that was switched on I'd always get full speed on my line but usage would count against my credit that I'd bought in advance. I guess it could be billed at different rates at different times of day too.

    I feel this would suit me as 90% of the time I don't care too much about speed, I'd be happy to take whatever suited the ISP as long as I could ask (and pay for) full speed when I wanted it. I feel it would suit you because if you had a reasonable number of lines on this basis you could shape your usage to control your 95 percentile costs significantly.

    Obviously there would need to be certain guarentees on the first part such as a speed below which it wouldn't drop, a certain number of hours where you could get full speed, and a guarenteed average speed over a month to be certain you wouldn't just throttle the lines all the time :)

  14. I constantly hear people at work moaning that they are not getting 20Mb 24/7 that they're 'paying' for with Sky, AOL or whoever. So I've no hope that they could ever understand anything even slightly more complicated, be that 95%, units or whatever you decide on.

    I'm sure you have a lot of customers who'll understand 95% and are capable of making up their own minds, but I don't envy you the task of making it understandable for other sorts of customers :)

  15. Having several lines, I'd be happy for a 95%ile as long as it was aggregated over all my lines. Otherwise it would get very messy.

    Which leads onto the obvious point - if the base cost is coming out of one pot and usage is coming out of another, then allowing usage to be pooled may be a sensible option regardless of how it is being billed - it would certainly be beneficial to us.

    The downside to doing this is it makes it far more difficult to work out what costs are likely to be - the 'rate your ISP' sites are bad enough at the moment, but just imagine the table:

    ISP - Usage
    Tiscali - 10GB
    BT Openwound - 10GB
    Sky - 15GB
    AAISP - Base cost per line depending on whether it's 20CN, 21CN or BE and billed on the 95%ile with a commit rate.......

    I think you'll be scaring people away.

    However, I do like the idea of AAISP units.


  16. AABytes works for me, too - as others have said, 95% is just too hard to predict, and really hurts people's ability to compare you with other ISPs (which in itself makes them less likely to choose you), and I really like the ability to do a 10GB download once a month without having any implications, as long as I don't go over the total allowance for the month.

    I find the current system of allowing carry-forward and -back very fair - averaged over sufficiently many users, hopefully it works out well for you guys, and it's certainly useful for me.


  17. Thanks for all the comments. It looks like we're going with a new units based system to be announced shortly.

  18. How about changing it so you pay per Gb, but with different prices per Gb?
    Peak BT 20CN £10.17
    Peak BT 21CN £6.78
    Peak BE £3.41

    With minimums as appropriate e.g. 8Gb for Be
    (If I'm understanding this correctly)


  19. interesting info on BE, shows how BTw price to gain a low porice point to attract customers but been subsidised by excessive usage costs.
    But here are my thoughts on what you ask.
    95th percentile is good when usage is not spiky, perhaps someone doing things like listening to a radio stream and occasional burst for windows updates and the like would benefit from it.
    charging per gig is the most consumer friendly I would expect, having zilch penalty for bursting and is a simple charge for usage. If I were a customer this would defenitly be my favoured option unless the 95th had far superior pricing.
    the commit option which I guess caps cost, but also caps burst I can see would be attractive for some people also, essentially you could see a Xmbit service which allowed an hour or 2 burst above that speed.

    I think the per gigs and the commit are the better 2 options you could offer side by side, I think 95th would have a very low takeup.

  20. Given that some prospective customers look at the current pricing and say "a 1GB limit for the same price I can get Unlimited from XXXISP? - Forget it!" I think anything more complex than is currently in place would be a Bad Thing.

    I haven't got the first clue how 95th%ile works, and I suspect that more than 95% of potential customers won't either, so I think that's best left well alone.
    The bursty-limit thing has a nasty feel to it - generally my usage is not highly bursty, but if I want to download a CD image (which I may do once every couple of months) I don't want it to drag things down afterwards. Again, I think it will scare off prospective customers.
    I think the current system of a fixed cost with a (time-of-day related) limit works very well, and by looking at Clueless I know exactly where I stand. Obviously the limit has to be dealt with if it's blown, and the carry forward seems to be an excellent way, with payable topups if needed.
    Basically, I'm saying that if it stays as it is now, with adjustments to the individual numbers for 20CN/21CN/Be, I'll be quite happy.
    So all that needs to be sorted out is the different "fixed" and "topup" costs for the three types of lines - Simples! (Cluck! :-)

    - HDRW


Comments are moderated purely to filter out obvious spam, but it means they may not show immediately.

NOTSCO (Not TOTSCO) One Touch Switching test platform (now launched)

I posted about how inept TOTSCO seem to be, and the call today with them was no improvement. It seems they have test stages... A "simul...