Cute analysis, but a little biassed towards HJ.
Read on for a little more realistic look.
Before we start however, I should say that my opinion of this is that the price of
jfw95 is fair, but the price of jfwnt is not.
Despite the information contained herein, the simple fact of the matter is,
When you buy a copy of jfw these days, you are paying for a program that will run
under 95 and nt equally well.
There is absolutely zero difference in the program you purchase from HJ when you buy
the NT copy of jfw as to when you buy the 95 copy, excepting the information stored on
the keydisk.
Not withstanding the cost analysis alone, this charging model is unfair.
Personally, I don't feel juped since I only use the 95 version of jaws and wouldn't
use nt if you paid me,
but I imagine the people who have to use NT for whatever reason (poor souls) get
pretty irate when they realize that they've paid almost double to have a keydisk
changed.
Anyway, a few corrections on the cost analysis.
On Sat, Jan 09, 1999 at 12:40:31PM -0500, dave wrote:
> With all due respect, I'm afraid you guys don't have any more
> idea of what it takes to develop, market, and support software
> than Mr. Spider does.
Not quite true, There are many of us on the list who do, and to take such a
condescending tone before you present anything doesn't put you in a very good light
with your readers, that not withstanding...
>
> The $700 price differential of JFW for NT over JFW for Win 95
> is due to the cost of developing, marketing, and supporting JFW for NT
> having to be spread over a smaller customer base. It's just that
> simple. If you want the NT version and the 95 version to be close
> to the same price, HJ would have to raise the price for the 95
> version and then people would complain that 95 users were paying
> the ride for NT users.
This is one side of the coin.
The other is that the NT market is diminished by the simple fact that the screen
readers are too expensive.
When a blind person is faced with the choice of NT or 95/98 at purchase time on a new
computer, they are invariably going to choose 95/98 since they would be aware of the
cost involved with purchasing the nt version of their screen reading system.
Drop the price of the nt version to something approaching that of the 95 version and
you increase your market since more people will be willing to try nt.
The price differential is worse in other countries, here in Australia the difference
is closer to $1400.
$1400 to have a keydisk changed? When the Australian marketting and support isn't even
done by HJ (in most cases) I don't think so!
> JFW for NT is not just a re-compile of JFW for 95. When a program
> ( JFW or any screen reader) is so closely "married" to the operating
> system, it takes a re-write of the code, not just a re-compile.
> Even when every line is not re-written, every function call
> needs to be looked at. That takes time. And for every line
> that is changed, it needs to be tested.
You're right, its not a recompile, its the exact same product.
When HJ sells a copy of jfw for 95, they are selling a copy of jfw for NT also.
Testing and support costs aside, if you want to charge more for what you claim to be a
different product, it has to actually be a different product.
It isn't. Pure and simple. Different box yes, different keydisk yes, but same product.
>
> Take a look at what happened to GW-Micro's Window Eyes for Win 95.
> It wouldn't even run under Windows 98! That's further evidence
> that you can't just take one program and move it to another,
> even if similar, operating system. (And just to speculate,
> maybe if GW-Micro charged more, like HJ does, they could have
> afforded more programmers to get the job done on time. But that's
> just guessing on my part.)
No its not, its bad programming in the first place.
There is no reason to expect that a well written 95 native program (32-bit) shouldn't
run under 98.
That was one of the design goals with 98, otherwise Microsoft would have pushed us all
to NT straight away.
Now, if the software in question (which I haven't used) was 16-bit, then things get a
little more interesting.
But if its 16-bit, then its not a 95 native program, its a windows 3.1 program that
happens to run under 95.
JFW 3.0 wasn't, it was 95 native, and hence, it made the transition just fine to
windows 98.
Sure, some reworking was needed to get it to work with ie4 and such other features,
but it did at least come up running.
Lets not kid ourselves here, 95 and 98 are almost the same product (with the exception
of the inclusion of the wdm) from a programming perspective.
>
> First, Ted Henter had to hire programmers that knew the INSIDES
> of NT, not just Win 95. Then he'd have to pay them for several
> months while they produced nothing while they learned the in's
> and out's of JFW. Those guys pull in big bucks, so even
> at 6 months times two people, he could have spent $60,000 in salaries,
> just getting two new programmers up to speed.
This is overly inflated.
I don't know what you guys pay your programmers there, but thats close to $120k in
Australian currency.
I'd sure like to be on that sort of money to program.
Secondly, the apis used to access information aren't that much different between
windows 95 and nt, that was the point of nt4. Admittedly, you are correct in your
description of the process above, just your figures and time delays are a bit off.
>
> Then don't forget the overhead of office space, computers, software,
> benefits, supervisors, management, etc, etc. So $60,000 in salary
> equates to about $90,000 spent by the employer.
That much is true, so lets say closer to $70,000.
>
> Suppose he couldn't hire any because there is a shortage of GOOD
> NT programmers, he would have had to train his current programmers
> in the inner workings of NT. Again, that takes time and money.
> So it's still 6 months to get programmers up to speed, and about
> $90,000 spent while nothing is produced yet.
Again, correct, but overinflated. $70,000
>
> But we haven't even gotten to the support department yet. Same thing
> there. Either hire people who know NT and train them in JFW. Or
> take the current employees and train them in NT. And we're not
> talking about CASUAL users of NT. We're talking about people who
> know it well enough to do phone support, to read people's minds over
> the telephone, to know all the little quirks, to know the glitches
> gotchas and bugs of NT and how JFW interracts with it. To the CASUAL
> user NT and 95 may look very much alike, but to someone who has
> to take the phone calls and fix problems, the similarities are only
> superficial. And by the way, since almost all NT machines are on
> a network, the people who support JFW for NT better know NT networking,
> too. And NT networking is different from Novell, which is different
> from 95's peer to peer, etc, etc.
This part is way off mark.
You're placing the burden of complete nt support onto HJ. This isn't the case.
HJ's job is to support their product in combination with NT.
To set up a network, and configure nt to run correctly is the job of the local system
administrator.
Admittedly, HJ may have to work with the system administrator in some cases, but this
involves working closely with someone who already knows the ins and outs of NT
themselves and hence is much easier than dealing with normal users.
Training the support staff in the internals of nt (with the possible exception of the
video subsystem) isn't necessary.
Training them in the user interface is similar if not the same as 95.
>
> HJ's support department and programming department need NT
> workstations in addition to the Win 95 workstations. And don't
> even suggest dual boot. Because of how JFW is "married" to the
> operating system, how it reacts with NT in a pure NT machine is
> going to be different than how it reacts with NT in a dual-boot
> machine. So even in addition to the solely NT machines, the
> support department better have at least one dual-boot
> (95 and NT) machine because you just know some customer is going
> to want to do it.
This indicates a complete misunderstanding of how dual booting works.
JFW is not a low level device driver, its an application.
How it reacts to pure nt is just the same as how it reacts to nt with any other
operating system on the machine. It runs with the operating system in memory at the
time.
Since we have already established that jfwnt and jfw95 are identical products, it is
even reasonable to expect that on a dual boot system, the one copy of jfw could be
used under each operating system.
The interfaces are similar enough, some tweaking of the script files might be needed
but its minor.
Make the existing support machines dual boot and you're fine.
Or, if you want, buy some extra machines, thats all part of development.
>
> So let's budget another $40,000 to get the support department up
> and ready for NT. We've now spent $130,000. But wait, that was just
> to get them up to speed, that wasn't to develop anything. Suppose
Ha. lets not. Lets look at more like $20,000, if that, and we come out at $90,000
> it takes a man-year of programmer time to develop, TEST and DEBUG
> the NT version of JFW. That's a LOW estimate! That's another
> $90,000 on the programmers.
No its not, we already have the support infrastructure in place so your above estimate
is inaccurate.
Taking a more accurate look at the programming wage, and including maybe $10,000 for
support costs, we have $50,000 extra.
>
> Just to get JFW NT out the door, BEFORE EVEN ONE SALE IS MADE,
> Ted Henter had to figure on spending AT LEAST $220,000.
Try $140,000
>
> But wait! Our lesson isn't finished yet. In addition to development
> costs there are marketing costs (how much you have to spend to make
> a sale, and deliver the product), and support costs (the customers
> are going to call you on the phone).
We already have a support and marketting department, do we not?
>
> It costs money to send out salesmen to get the word out. Even if
> they don't physically leave the office, someone has to contact
> prospective customers and say "Hey, we got JFW for NT for sale!".
> Someone has to network with NFB and ACB. They need to contact the
> personnel departments and I S departments of companies and government
> agencies who are known to hire the blind. It's really a stab in the
> dark to guess how much your marketing costs PER COPY will be before
> you sell any, but a budget still has to be developed.
>
> Then for each copy sold, you can count on an average of X number
> of hours of AFTER-SALE phone support. Even if the customer
> doesn't purchase the SMA, you still have to provide some
> phone support. Because if the user can't use it, they're going
> to demand their money back and bad-mouth you to everyone else.
> (Like I've done to certain other companies. [grin])
> So you have to budget some amount of after-sale support into
> the selling price.
True, but the sma has nothing to do with this.
Its a discounted pay for upgrade service, nothing more.
Other companies offer the same sort of thing, only they do it differently.
They sell off their old stock with discount coupons for upgrades to the latest version.
HJ just chooses to offer the discount upfront at time of purchase.
>
> You also have to figure in the cost of money. That $220,000
> costs a lot more than $220,000 because it has to be
> borrowed from investors or a bank, or even Ted Henter may have
> borrowed it from himself. Even if he used his own money, he still
> deserves a return on the investment, because he could have put
> that money in the stock market and had it working for him.
> Because of the "cost of money" let's say that the $220,000
> costs $300,000 over the borrowing period.
Sheesh, I'm glad you're not my bank manager.
To save you all the time of working it out, thats 40% interest, if we assume the
borrowing time is 1 year.
Lets be more realistic and say 2 years. Noone charges 20% interest on loans, not even
in Australia.
and thats not at all taking into account the fact that there are profits being made in
other areas that could be bought into play.
Assuming that there aren't, however, our $140000 at an interest rate of 10% p/a
which is more realistic, over 2 years, comes out at
approximately $170,000
>
> Because the market for JFW for NT is so much smaller than
> the market for JFW for 95, the marketing costs per copy
> are going to be much more. I'll take a stab and guess $400
> per copy for marketing costs, and $300 per copy for support
> costs. (Yes, that looks high, but if you've ever had to
> support NT products, you'd know why.) $400 marketing
> cost out of a list price of $1500 is 27 percent, which is not
> out of line, and, if anything, is on the low side.
We're forgetting two important things here.
HJ doesn't have to do marketting for every copy of jfw sold.
And, jfwnt is the same product as jfw95.
If we take your figure at face value, which I disagree with, but anyway, and assume
that HJ sells 1 in every 4 copies of jfw directly,
then we have a figure of $100 per copy, plus your $300 for support (which is
reasonable), or $400 out of $1500.
Starting to approach the windows95 model? Good, you're still with me then.
>
> Based on my experience, I'm very confident in assuming that
> the $100 per year or $200 per year SMA does not come anywhere
> near the true cost of ongoing support and updates. I'm supposing
> that HJ bundled a major portion of the support costs into the
> selling price due to the nature of how customers fund the purchase.
> It's much easier to get funding for "up front" costs in government
> sales (either direct sales to the government for their blind
> employees, or in cases where state-run vocational rehab pays for
> individuals).
Correct, and I would assume that this is in fact the way HJ sees it too.
Like I said before, the sma is to ensure that there is money available for upgrades,
the support costs are covered up front.
>
> So there is $700 per copy that comes out of the price before we
> even begin to pay for development costs.
$400 per copy.
>
> We're still not done yet. We need to take into account
> that we need to keep those programmers on the payroll to come
> out with new versions and bug fixes. If you don't constantly
> update your product you go out of the software business.
>
> (As an aside: The rule for hiring programmers is: "You can't hire
> just one." There is so much job turnover it's ridiculous. If you
> put all your eggs in one basket and he or she leaves, you're
> screwed, that's why you need AT LEAST 2.)
>
> Two programmers at $60,000 per year plus the usual 50% overhead
> (floorspace, computers, benefits, management, etc.) equates to
> another $180,000 per year. (This figure is a little high because
> the NT programmers are also available to work on the 95 version,
> so their cost can be spread PARTIALLY over Win 95 sales and not
> soley NT sales.) Let's deal with a TWO year period, so we add
> $360,000 to the $300,000 development cost for a total two-year
> figure of $660,000. That's $660,000 for development and 2 years
> of product maintenance. (Did you notice how ongoing costs
> quickly exceed the already high development costs? That's why you
> need to charge for an SMA, and why $100 or $200 per year is CHEAP!)
Lets adjust this down ust a little.
$40,000 per year per programmer, plus an ongoing upkeep of $10,000
(You either have a drasticly underrated idea of how much it costs to get someone
started or you're trying to artificially enflate your figures, because you're quoting
the same 50% cost across the board, from start to finish, and that simply isn't the
way it works).
So we have our two programmers at $40,000 per year, plus our $10,000 overhead (there
are only 2, and a 2 year cycle for hardware/software replacement is way off target, 3
at the minimum) giving us a total of $90,000 per year thereafter.
On top of our initial $170,000 we have a figure of $90,000 making a total of $280,000
for the two years upfront. This is more like it.
>
> Do you folks get it yet? Ted Henter had to plan on spending
> at least HALF A MILLION dollars to put and keep JFW for NT on
> the market for just a two year period.
Just over a quarter of a million dollars.
>
> I'm almost done with this lesson, so bear with me just a bit longer.
>
> Let's take the $400 marketing and $300 support costs out of the
> current $1,500 price for JFW for NT. (1500 minus 400 minus 300
> equals 800). That leaves $800 per copy to be applied to the
> $660,000 figure.
Lets take the $100 sales and $300 support costs out of a current price of $1500.
$1500-$100-$300=$1100
That leaves $1100 to be applied to the $280,000 figure.
>
> Dividing 800 into 660,000 gives 825. So if my estimates are
> anywhere realistic, it means that HJ has to sell 825 copies of JFW
> for NT within the first two years just to break even.
Dividing $1100 into $280,000 means that hj has to sell 254 copies of jfwnt to break
even in the first two years.
Thats accepting the difference between jfwnt and jfw95 which as I've stated above, I
don't.
>
> There's even more. We've talked about paying interest on borrowed money,
> but we haven't talked about profit. You have to plan for profit.
> If you plan to break even, you'll lose money, gauranteed. Call it
> a "fudge factor" or a "buffer", but you have to plan that "extra" because
> life doesn't always go as planned.
Standard accounting practices here, and correct.
Furthermore, it is the capitalist way of life to want to earn as much money as you can.
Without implying anything at all about the staff of HJ, it is reasonable to expect
that they would want to make a considerable profit off selling jaws, I know I would if
I'd developed something with such a potential market.
>
> So let's round it up. To set a goal that will allow at least some
> profit, (and how to realistically determine that is another treatise that
> I won't bore you with), let's say selling 1000 copies is the goal for
> the first two years.
Ok, I'll be fair, and round mine up to 400.
>
> Someone who recently purchased JFW for 95 look at their serial number.
> That gives you an idea of how many copies of JFW (all versions)
> are out there. Now consider how much smaller sales of JFW for NT
> are going to be compared to all the other versions.
I did. the last version I saw was in the 20,000s
Now my copyright message under about jfw shows 1993-1998 which means its been selling
for 5 years.
If we average that out, thats 4,000 copies of jfw a year.
If we assume that 1/4 of those were nt copies, then thats 1000 copies per year,
quite a bit above 400.
The above paragraph is wildly inaccurate for two reason,
firstly that jfwnt hasn't been on the market all that time, and unfortunately I don't
have a copy handy to find out how long it has been selling.
Secondly it is not reasonable to assume that the number of new copies sold will
average out like that, it should grow exponentially as new feature sets are
incorporated and more people learn of the product's existance, but that isn't
necessarily the case,
and with a market like that of hj, it will also be effected by the number of new blind
people
(children becoming old enough to make use of jfw, or other people losing their sight)
who come in contact with jfw each year.
>
> My guess is that it has been and will be a real challenge for HJ
> to sell 1000 copies of JFW for NT during the first two years of
> that product's life. Also keep in mind that any large purchasers,
> such as a government agency, are going to want a substantial discount.
> So taking discounted government sales into account, the goal needs
> to be at least 1200.
noone's denying its a challenge, but that doesn't necessarily mean that the current
marketting strategy is correct, or even is the one that will make the most money for
hj in the long run.
Your article completely fails to take into account how the high price of jfwnt affects
the potential for sales, if you rethink that, you'll come up with vastly different
figures.
>
> The real amazing thing is not that JFW for NT costs so much. The
> truely amazing thing, and big credit to HJ, is that JFW for 95 costs
> so LITTLE!
Nope, the big credit goes out to whoever's idea it was to sell the same product at two
different prices, and with a copy protection system that allowed an easy,
over-the-phone upgrade from one to another, with a huge profit being made in the
meantime on the upgrade price.
>
> End of lesson.
I'd like to point out n summary that I dn't claim my figures to be based on fact by
any stretch of the imagination,
I merely present an opposing view of the situation, which I personally believe to be
more accurate.
The real truth I suspect probably ies somewhere inbetween the two.
It is unlikely that any of us will ever know the truth of the situation, and perhaps
that is for the best.
It is important however, when looking at a problem to realize that there are always at
least two sides to it (generally more) and that yours, or my, or anyone else's views
for that matter aren't necessarily correct.
To condemn someone's knowledge based on the fact that your opinion of how things
should be done is different to theirs based on your personal experiences shows a lack
of understanding and tollerence.
There are always going to be people who believe that HJ charge too much for jaws, and
there will be an opposing faction who believe they don't charge enough.
I rather lie on the side of those who think that jaws is too expensive, but I also
believe that it is HJ's right to charge whatever they want, and that until something
better comes along that is cheaper, to complain will get us nowhere.
For those of you who have ideas about how jaws could be improved, and how hj's
marketting strategies could be improved, I'm sure [EMAIL PROTECTED] would be more than
happy to hear from you about them.
Regards
Aaron
--
Aaron Howell Kitten Internet.
[EMAIL PROTECTED] Unix/Networking support/system integration
irc: DaRkAnGeL +61-417-625550
I love you Nessa, now, forever, always!
-
Visit the jfw ml web page: http://yoyo.cc.monash.edu.au/~nallan/jfw
