Sean gave me this advice on how to interpret the results and what to do next.

Now, I should say, that it really shouldn’t take this long and I will admit to not giving it 100% attention as some day job issues are pretty pressing right now.  I guess I took on some responsibility when I launched the service and right now I don’t feel like I’m living up to that responsibility.  In part I am relying on the kindness of friends to keep agtweet running and I don’t want to abuse that by pestering them all the time for help.  I’ve been getting a lot of questions about when the 085 number will be back … to which I’ll say I’m not totally sure right now.

For the moment, accept my apologies and bear with me.

thanks

James

All of which results in posts like …

and …

and one or two more that weren’t as polite.  These guys are right of course.  Even though these guys are using the service for free, you can still expect a certain amount of flack if you don’t do what you said you were going to do.

What makes the problem more acute is the fact that @agtweet is “line of … “, well, I was going to say “line of business”.  In other words, if you service isn’t working for any amount of time you can expect to get some flack, un-subscriptions and bad karma.  This was also a problem for smartnote.ie when I offered inbound SMS services.  You see, for someone like me, who operates from a laptop, quiet possibly at the other end of the earth, I’m not always around to look after these things.  Nor, to be honest, do I really want to be, and this is in a way what makes ‘line of business’ type applications not very virtual.  To support them properly, you need a support contract with an IT supplier at the very least, or even full time support, whenever your customers can expect to be using the service.

Piehole the Irish voice over directory doesn’t quiet suffer from the same problem.  If some over-zealous code push does mean can’t update their photo for a couple of hours/days, it is not the end of the world.  The same cannot be said for many ‘webapp’ style systems.  If they go down, customers go mad.  This makes them not all that virtual and maybe not suitable unless you’ve got a plan that involves enough scale to warrant a support desk.

Now who do I know that does that?  Oh yeah – me.  So is this good news or bad news?   Well, the good news is that it means more people are using Twitter although I’m still detecting a slow down in the number of Irish Twitter sign ups.  

So lets compare and contrast.  I’ll be unbiased I swear.

Usability

o2:  Text START to 51210 and you’ll receive prompts to get going.  Pretty easy.  The trick is, you need to be an o2 customer.  I’ll knock off a star because it doesn’t help you if you are with one of the other operators.

★★★★

agtweet:

Go to agtweet.com and register your number.  Then start sending texts to one of two numbers  Not quiet as easy.  You also need to either pay online or text agtweet to 57030.

★★

Pricing:

o2:  Receiving replies is FREE (whoot) and updating your account is charged at the regular rate.  Pretty good.  You can’t really argue with that.

★★★★

Agtweet:  In some cases Agtweet it FREE to update your status.  For example, if you are a meteor customer on the right plan you can use our Meteor number.  Some 087 customers can do something similar but it all depends on your plan.  Other than that – regular text rates apply.  Replies are from 4.5c to 25c.  Hmm.  Well, I guess it can still be better value if you are willing and able to sign up with a credit card.

★★

Irishness:

Lets face it, o2 isn’t that Irish.  In fact, its Spanish the last time I checked.  Oh dear.  Doesn’t look like they’ll be scoring too well in this round.

★

Agtweet:  Holy god – Agtweet is Irish to beat the band.  Look at that green branding, there is even a little Gaelge on the front page.

★★★★★★★

Well who knew – its a draw.  I guess it is a question of time before all the operators have some kind of Twitter service.  There are still other niches to compete on of course.  You can help people by allowing multiple people to update the same account etc.

I am i n blood, steeped so far that should I wade no more, returning were as tedious as go o’er

It has been just over a month since I started offering a way for people to pay for @replies on agtweet.  I went for a 3 tiered pricing plan of €3, €9 and €16.  Within a couple of days I had 3 €3 subscription and a week later there was my first €16 subscription.  Things were looking reasonable.  Full sure of its passive rise to stardom I pretty much headed off into the wilds of Argentina for month.

When I came back, there was some good news and some bad news.

The good News

Everything was still working.  The inbound SMS modems were holding up and the replies were being delivered correctly.  This is actually better news that it may at first sound.  I know from my experience running SmartNote that SMS services don’t always run smoothly.  There were no customer complaints waiting for me.   This might be partly due to the fact that agtweet isn’t a line of business service.  If a tweet doesn’t get delivered, its not the end of the world.  Having a service which people are reliant on dramatically ups your support cost which pretty soon can mean having full time support which would kill this little project straight away.  I’m interested in an automated income, not one I have to nurse maid.

The number of tweets sent through agtweet also seemed to be going the right direction. It is not quiet a hockey stick but there is at least healthy growth in the number of tweets sent via agtweet.

The number of tweets sent in via agtweet

So – agtweet is cheap to run.

The Bad News

I had kinda hoped for the mythical viral marketing effect that would skyrocket the number of new users.  As it turns out, with my decreased tweeting (as I was away) there seems to be a consumat decrease in traffic.

New subscriptions declined in October.

In total there has been €43 in sales made up for 3, €3 a month subs and one €16.  Unfortunatly the €16 sub failed on the second month which is the problem with subscription services – you are forever trying to keep up with cards that fail or expire etc.  More worryingly, there weren’t really any new sales.

Much easier than experimenting is sending out a poll and asking people why they weren’t coughing up.  It turns out that at least part of the answer is that many don’t have a credit or laser card and wanted to pay by ‘credit’ or ‘premium rate text’ as it has been rebranded.

Previous surveys have told me that people do what they do rather than do what they say so I wanted a cheap way to test this a little more before taking it at face value.  Onebip.com integration took me about 3 hours and about 12 hours later I had racked up another 5 subscribers.  Not bad going.  I don’t really like the onebip.com solution however as while the tool itself is lovely, it requires a login an people can only pay from in front of their computers.  Much nicer would be able to activate your account when you were out in about.  I did some phoning around and after getting no small amount of condescending feedback from a few premium rate service providers, Zamano eventually came up trumps in helping me get off the ground.

Now the trick with premium rate services is that the operators and service providers take a fair old whack of the revenue.  Out of €2.50 premium rate text, I’d be left with between 95c and €1.41 depending on the network operator.  I’d still have to pay for actually sending out replies which are now costing me 2.5 cent per message.  The table below goes through the differences in billing 30c a message for each reply and charging €2.50 for 10 messages.

The only other approach would be to bill each @reply using the 30c tariff.  The problem here is that the split gets less favorable, the smaller the amount, and its actually cheaper to sell 10 replies for €2.50 (25cent per reply) for the consumer and marginally more profitable for me to bill €2.50 once and pay for 10 outgoing text messages.

So why bother with the premium rate texts at all if the revenue shares are so harsh.

The answer is reach. The basic theory goes that more people will be able to pay by mobile phone than will be arsed to take out a credit card and enter details online.  This is because paying by credit card is difficult but also, in  theory, users can sign up from anywhere.  More importantly, they are more likely to be out and about and want to start receiving replies rather than stuck in from of their computer.  Using a premium rate text they can do this.  So far, I have 4 subscribers paying by credit card.  On the meteor customers the margin on this is about 95% after credit card processing fees are taken into account.  This compares with a 60% processing fee for paying by mobile.  However, so long as twice as many people sign up by phone as by credit card, its still worth while.  The trick will be in not canabalising the credit card sales.   To try and achieve this the credit card deal is considerably more attractive (5c a message) compared to the mobile rate (25c a message).

To subscribe, or not to subscribe

Frankly, there has been a lot less resistance to subscriptions than I had expected.  Polls seem to indicate that not that many people are put off by it.  One nice feature of charging by premium rate text is that you rebill whenever you like.  This means you can have a semi-subscription.  This means you only rebill people when their existing credit runs out.  For the consumer that means that they don’t pay unless they use the service.  At the same time, they don’t have to make a whole new buying decision every time they run out of text.  The regulator requires that you notify people if they have spent more than €20 as a reminder but this is a lot less onerous then requiring them to re-sign up every 10 messages.

Ag Tweet did something last week that 99% of web applications fail to do.  It garnered some paying customers.  Four of them in fact, eaching shelling out €3 a month.  I’m writing up a brief explanation of how it was put together and why I made various choices including pricing models.

So what was the objective of this bit of development?  The real objective was to simply fuel the number of people who are using Ag Tweet.  @replies (text messages sent to phones when someone is mentioned on twitter) were the top request from users and I had a hunch that by providing the service it would improve the number of people that sign up.

Pricing

I am working on the theory that pricing for subscription services is not set by value but instead by what the customer has available in a budget to spend on that item.  This isn’t a theory that has any grounding other than my own whim.  A lot of people, and certainly the most vocal, using agtweet are between 14  – 20.  These guys are great – they shout about agtweet and badger their mates into signing up.  I also carried out a survey amongst the beta testers.  This is what they had to say about a price.

I was pretty happy with this response.  Of course I would have been happier if people had said they would pay 9 – 16 euro a month but I guess turkey’s don’t vote for Christmas.  So why pick €3 and not €5?  Three reasons.

1. Three euro really sounds like not a lot of money.  I’d expect a lot of people to give it a whirl and a lot of them won’t be bothered with unsubscribing for such a small amount.  They’d rather let their credit cards expire.
2. What I really wanted to do is test to see if anyone would bother taking out their credit card at all. I know a lot of people talk about pricing high and discounting – but really these €3 guys are just there to spread the word (by tweeting about the service) and and provide a funnel for the higher price plans.  I only plan on breaking even on them.
3. I knew from the test group that 67% of users would have enough replies in 50 / month while 30% would need 150 and the remainder would need 300 per month.  My long term goal would be for the subscriptions to break down in that fashion.

That would give me a subscription breakdown something like this.

So all this work for 80 lousy euros?  Well actually  – no.  It is worse than that.  50% of agtweet users are on meteor while the rest are on some othe network.  I can MMS the meteor guys for free.  This was the original brainchild of @jmccrohan. I’m selling the text messages to the other guys roughly at cost price which means I should clear a cool €40 a month if all goes according to plan.  I could cut out the SMS cost altogether by making it a Meteor only service but for now I’m effectively using them to subsidise the others and help build numbers.

So why bother?  Well that is a good question.  Its pretty clear that on these prices it is not going to make a lot of money any time soon.  Instead I’m just stoking the fire.  This way it won’t cost me anything to run and hopefully adding the functionality will help to build the subscriber base.  Eventually the subscribers will start upgrading  and at that stage we’ll see some more revenue.

Payment method

I originally intended to take payment by Paypal, mostly for ease of integration.  The problem with Paypal is that they don’t support Laser and I knew a lot of users didn’t have a credit card.  I decided to use my Worldpay account instead.  There was another reason to use worldpay.  I have found that PayPal subscriptions are fragile little things.  I’m not sure what they put in the subscription emails but they positively beg subscribers to unsubscribe.  I’ve found way lower unsubscription rates with Worldpay.  Worldpay does allow you unsubscibe but the agreements tend to live longer than Paypal for some reason.

So here we have it.  Three price plans.  The integration was fairly straightforward although I have been getting a lot of requests to take payment by debiting phone credit.  I’ve been looking into onebip.com which seem to have a pretty transparent and easy to understand intration.  The only challenge is that they take about 50% commission on all payments.  That means I’ll have to make it more expensive ot pay via this method – unless I made it a Meteor only option.  Now that I write that – it may not be a such a bad idea.  Maybe the Meteor customers can drive awareness with their cheap tweets.

You can check out the (mostly javascript) tool at http://www.jameskennedy.ie/plannr.

So what does it do?  It aims to help you discuss price points for your subscription service.  It assumes there are 3 ‘levels’ of subscription and you can play with the figures ( prices, % of people on each plan, total number of subscribers etc).

When you are ready to you can add a comment at the bottom and email it off.  None of the data is saved on my server and all the information is held in the URL.

I’ve pre-populated the table with figures which look kind of like what I’m thinking of for AgTweet.  If you have any ideas on what pricing structure I should put in place for it, by all means fill out the sheet and let send it on to me.

First up it might be worth mentioning that it appears as though twitter take up in Ireland is slowing.  For the last three months I’ve been tracking new accounts and there seems to be a downward trend although the first months data could be skewed because I was busy finding all the new people.

So what was the strategy?

Rather than try to build up individual customers, my strategy was to try and find a sponsor who could effectively pay for all the costs and a modest profit for myself.  This started off very well.  Through twitter someone got in touch who wanted to sponsor a number.  I quoted €300 per month in return for adding a o2 number to the meteor and vodafone numbers already on offer.

I was pretty chuffed with my initial success.  I made an error though. I didnt’ close the deal.  I had started to line up meetings with ad agencies around town in the hopes of getting a major brand to sponsor the whole kit and kaboodle.  As a result I delayed asking for payment details.

“If it was that easy to get sponsorship, I should hold out for something bigger”, I cleverly told myself.  The meeting with the agencies were going well with lots of interest and excitement.  I had 3 in total.  It was actually great to get to talk to these guys but although the service seemed like a nice idea, there was no real way to tie it into any of the brands that they represented.  Actually, one company did suggest that I invest a thousand euros in a affiliate marketing scheme to help sell product for a customer.  Cheeky – sure god loves a try’er.

By this stage I had decided to dream up ways I could offer a number of different price points to potential sponsors.  I offered plans including

• basic sponsorship (logos on the site etc)
• headline sponsorship (exclusive sponsorship of the whole thing)
• naming rights

The name rights I put in at about €50k while the basic sponsorship rant to about €5k.  Here is template for the sales presentation I made each time around.

Of course as soon as I started this type of talk, I remembered why I don’t like big ticket sales in the first place.  After a couple of weeks, calls started to go un-answered, emails were being ignored and finally after about 2 months I realised that this wasn’t going anywhere.  In hindsight I can see how I didn’t nearly push hard enough or present a good enought case.  Why would’t anyone build it themselves?

Of course by that time my €300 a month sponsorship had disappeared.  A case of a bird in the hand is better than stonk load in the bush I’m afraid.

So now what?

Now, there is a case to be made to trashing the whole thing.  So far there are 1,006 registered users.  There was an initial spike as the press release was picked up by Silicon Republic.

The graph to the right shows the total nubmer of tweets sent by agtweet while the graph to the left shows the number of new accounts created both by month.

So now what?  I have 1,000 odd people who signed up with about 300 of those having used the service more than 20 times.  Within that 77 have used the service more than a hundred times.

The really nice thing about the service is that (I think) its pretty easy to figure out what people want.  Top of the list is

1. Replies to their phone by SMS or email.
2. The ability update multiple accounts
3. A tie with with facebook so that both update at the same time.

The replies are easily the most requested feature.  I’ve implemented a SMS reply feature but I’m reluctant to roll it out as (a) the texts cost me money and (b) scanning all 1,000 accounts for updates will bog down my machine which really has a day job doing something else.  I’ve split out the code so this scanning can be done on separate slave machines but they’d still have to be paid for.

A couple of people have come up with ideas for delivering updates without incurring an SMS cost but I’m still stuck with the server load issue.

I guess that I might get 10 people paying for this service.  I’d have three plans and the margin on text messags would be somewhere around the 100% mark up but that would still only amount to 10 x €10 for 100 credits * 100% margin = €50 a month.  Plus, there would be a stonk load of customer support to deal with (there always is with SMS) so it just wouldn’t work.

I spent quiet some time selling SMS and I’ve no real desire to go back there any time soon.

So what next?  Well the obvious answer is “walk away and do something more productive with your time”.  I might as well learn as much as I can while I’m at this though.  I like the idea of offering an ‘limited invite’ system that would help promote the service.  Maybe I’ll leave the SMS reply in the hands of a few chosen ones for a while and try and stir up some buzz that way.