New - February 2nd 2007: A cross platform vision for Delphi - a detailed analysis and proposal on how CodeGear could increase Delphi's market share again by integrating a cross-platform compiler into the Delphi/BDS IDE.


New - June 12th 2007: Why the CodeGear newsgroup censorship sucks - a short rant about the fact that CodeGear has reached a new all-time low by not allowing their customers to complain and discuss about Delphi product plans.

The alternative Delphi roadmap to success

An open letter to Borland/DevCo/DTG

Written by Simon Kissel, October 12th 2006

Update Oct 30th:

Borland has now posted a public survey about Borland Developer Studio/Delphi to find out what their customers really are after. You can find the survey here.

It's a pretty good survey IMHO. However I'm a bit depressed that it doesn't include any Kylix/Linux-specific questions (to find out if people would want to use BDS for developing server-side Linux applications or GUI stuff etc) - after getting so much feedback from people interested in this area (yeah, that's my main interest, I admit it ;) it's sad they didn't include questions about it (the only question is if you plan to target Linux now/in the future, which right now will give a pretty skewed result with Borland offering nothing in this area - after all, a "yes" on this question implies that you are going to be an ex-customer...). I guess this still is one of those "we don't want to hear what our customers want as it conflicts with our roadmap plans" areas. But well, at least Win64 and Unicode and 'Native code vs .NET' are in the survey, so go take it.

Update Nov 3th:

If you care about having Linux as a deployment platform from inside the Delphi IDE (like CrossKylix, but with an updated compiler and supported by Borland) or care about having Delphi running under Linux, there now also is an unofficial Kylix survey created by Haruyuki Fujimaki and SiegfriedN. Over 250 people have already participated, and so should you! Realtime results can be seen here. Let's hope Borland/DTG takes a close look at this, too.


Dear Borland/DevCo/DTG,
dear Delphi community,

recently I've been very concerned about the roads taken by Borland/DevCo/DTG and how it may affect the future of Delphi, a tool I love and based my whole business on. After recent discussions on that matter in the borland.public.delphi.non-technical groups, I got the feeling that I'm by far not alone. I've therefore decided to do some hours of research and thoughts, and write all of this down.

The text below has been sent to my contacts at Borland, and afterwards been posted to the borland.public.delphi.non-technical. I'm very aware that Borland would have preferred if I would not have made this public, but while I have the highest possible respect for those folks (and I'd even go as far and call some Borlanders friends), I think that they are wrong. And I'm convinced they don't realize it. And I believe that the only way to get them back on track now will be massive pressure from the community, their partners and their customers. So, please understand that my intention is not to harm Borland/DTG, my intention is to put pressure on them so they start investigating if possibly their current strategy really is flawed.

If you think that I've got a point, feel free to link to this article. You may also provide Feedback in the borland.public.delphi.non-technical group, reachable with a newsreader like Outlook Express here, or by browser here. If you wish to contact me, you may mail me at kissel@computerman.de. However, I recommend that you instead mail to Borland and post to the newsgroup listed above.

Abstract

In this document I will show you that today, 80-90% of all Delphi customers are using native code development. I'll also show you that while many of them indeed are happy of the options given to one day being able to move to .NET, for the far majority features in the native code area plus improvements in the IDE itself right now are much more important then getting the latest and greatest of new .NET features today.

This document will also show that Borland's current roadmap contains of milestones relevant to only 10-20% of todays users, while the majority of 80-90% are not given important enough reasons to upgrade to newer versions.

Further, my findings will show that Borland's current strategy and roadmap includes a too high risk of losing large chunks of their todays customer base, with only little possible gains even in case their strategy works out.

I'll prove these points with detailed background information and feedback gathered, and I will propose an alternative roadmap that in my opinion will generate far better revenue for Borland, strengthen their position in todays markets while opening up interesting new markets for the future. My alternative roadmap does not imply Borland should be dropping its .NET offerings. But it clearly says that after 4 years of investments into their .NET visions, they now simply will have to focus on actual needs and requirements of the majority of their customers for a round or two again.

1.) Market analysis

I'll focus on the existing customers first. What do they use, and what are they interested in?

Let's investigate a few sources to find out.

a) Turbo download counts

  • Delphi Win32 48%
  • C++ Win32 24%
  • Delphi .Net 14%
  • C# .Net 14%

(After verifying with other download mirrors, these numbers are believed to have a maximum deviation of 5%).

72% of people interested into Turbo Editions are interested in native development.

b) Component market

Delphi components that have seen updates/new releases by their authors/component vendors during the last 2 weeks - data taken from torry.net:

  • Delphi/C++ Win32: 19
  • Delphi .Net: 1
  • Kylix: 1

For comparsion, we are taking the Top 20 downloads from delphipages.com:

http://www.delphipages.com/result.cfm?TD=20

Lot's of interesting data here. We'll have a look at the Top 3 downloads:

AVLLock Gold Download counts:

  • Delphi Win32: 18016
  • Delphi .NET: 1285

Interesting: The highest single-version download count is for Delphi 7: 4688.

DISQLite3 Download counts:

  • Delphi Win32: 21089
  • Delphi .NET: 0 (n/a)

VCL Skin 4:

  • Delphi Win32: 52109
  • Delphi .NET: 0 (n/a)

c) Community feedback

On Wednesday, October 04, 2006, Nick Hodges asked Borland customers who decided not to upgrade to the current product offerings what the reason of this decision was. There was an overwhelming feedback, both in that blog's comments, as on the Borland public newsgroups.

The most frequent reasons for not upgrading given were (not ordered, just the most frequent answers given picked out):

  • Missing support for Win64
  • It does not contain support for Unicode
  • Missing upgrade path from previous version
  • Previous product versions (Delphi 6 and 7 mentioned most often) are equally good or better than current product offerings (speed, stability)
  • No interest in .NET, which recent product version are focussed on
  • Bad product quality of recent product versions
  • Bad documentation in recent product versions (Help)
  • Switched to Visual Studio
  • Missing Linux strategy
  • Unsure about the future of Borland/DevCo
  • Switched to FPC/Lazarus
  • No Internet sales
  • High pricing outside of the US

If we filter out all answers not related to the roadmap, IOW sales, "no longer a customer" etc, this leaves us with:

  • Missing support for Win64
  • It does not contain support for Unicode
  • Missing Linux strategy
  • Bad product quality of recent product versions
  • Bad documentation in recent product versions (Help)

Further, having a look outside this specific poll, there are indicators that most people that communicate on the Internet about Delphi are talking about Win32 - delphipraxis.net, Germany's biggest Delphi forum (548.737 postings) has even closed it's Delphi.NET subforum completely due to inactivity.

d) Borland quality central

For Delphi, the highest voted report with 738 votes is "Create native 64-bit compiler/IDE that targets AMD64 and Intel's 64-bit (IA-32e) extensions". Also, Unicode is on the list (79 votes). An additional request not mentioned yet is shipping dbexpress sources.

The other high-voted reports are about product defects.

For Delphi.NET, the highest voted entry has 220 votes, and is about allowing the usage of VS plugins inside BDS.

For the TOP10 reports of Delphi, a total of 1651 votes was given. For the TOP10 of Delphi.NET, a total of 316 votes was given.

84% of the votes given on QC are about native Delphi. Only 16% of the votes are given for Delphi.NET.

e) Versions used

While we don't have an exact source for this, the findings above indicate that there is a high number of Delphi users that haven't upgraded at all during the last years - it appears that right now there are much more users using Delphi 5,6,7 in production than customers using Delphi 2006. Indicators for this are:

  • Frequently, if people are asking on the newsgroups and on different web forums, they mention they are using Version 5,6 or 7.
  • Download counters at component websites like delphipages and torry indicate that the most often downloaded versions are for Delphi 6 and 7.

This is interesting to note.

Market analysis summary:

Borland has often stated that voices raised on the Borland newsgroups are non-representive to their customer base at whole.

However, by investigating all available information sources, it seems like Borland is wrong about this.

We've seen that 72% of all Turbo Explorer downloads are about native Delphi/C++ code for Win32. We've seen that more than 99% of the Delphi component market are about native Delphi. We have seen that 84% of the votes given in Borland's QC feedback system are about native Delphi code.

We may safely assume that 80-90% of Delphi customers are interested in native code.

2.) Groups of Delphi customers

I'll now divide the possible buyers for new Delphi releases into the following categories:

  • a) Existing Delphi users, only doing native code, with no plans to move to .NET in the foreseeable future. Many of those are still using Delphi 5, 6 or 7. Some do this because they think these versions are good enough or even better than recent Delphi offerings for what they do.
  • b) Existing Delphi users, only doing native code with Delphi. For new .NET products, they have switched to Visual Studio. The most often reason given for this is that Borland's .NET strategy is hopelessly lagging behind Microsoft. Many of those probably are lost customers that will stick with their current Delphi version to maintain their Win32 code.
  • c) Existing Delphi users, that are in the progress or interested in moving their Win32 codebase to .NET now or in the near future, or who use Delphi Win32 for one part of their work, and Delphi.NET for others (for example ASP.NET)
  • d) Existing Delphi users, that now only do .NET with Delphi
  • e) New customers interested in doing .NET development
  • f) New customers interested in doing native code development

While we don't have numbers about sizes of the groups here (this would be an interesting poll topic, though), I'm assuming the following (yes, this part is speculative):

  • From the data gathered in 1.), we can be pretty sure that a) currently is the biggest group of people that could be interested in buying new Borland products.
  • For b), it might be possible to sell them new versions of a Win32-focussed for a limited time, if gives them a huge benefit for their remaining, non-ported native code. It might also be possible, but hard, to make them switch back from Visual Studio to BDS, if Borland manages to beat Visual Studio feature-wise. My impression is, that this group of people also is quite big - maybe of the once well-known community "VIP"s, including Book authors, TeamB members etc are in this group.
  • Compared to a) and b), c) seems to be a small group. There are a signs of life from them in the community.
  • There is not much evidence available that group d) even exists
  • For e) we have the data from the Turbo Explorer downloads. We may therefore assume that there might be interest for this area. We don't have any data on what potentially could be attractive about Delphi's .NET offerings for new customers, compared to the competitor Visual Studio.
  • Group f) seems to be small. However, this group has the potential to grow rapidly. As Microsoft is leaving the native RAD market behind (dropping native Visual Basic), Delphi now again is the best solution for Win32 development. However, this group probably right now is scared off because Borland is not offering native target support for other platforms becoming important to non-.NET-users, namely Win64 and Linux.

3.) The current Delphi roadmap

Highlander

Since the year 2002, Borland is focussing on .NET - for the last 4 years all product releases had a clear focus on .NET, with release cycles driven by .NET market needs.

The current Delphi roadmap lists Highlander as the next release, with focus on .NET 2.0 exlusively. The roadmap says: "There will be ongoing work for Win32 Delphi and Win32 C++ (such as unit testing and additional refactorings) but the most of the new work in Highlander will be focused on .NET 2.0.".

Comparing this with the results from our market analyis above, we come to the following conclusions:

  • Highlander does not include a single feature of the most requested features in QC for Delphi/Win32
  • The only highly-voted .NET feature voted for in QC, the integration of VS plugins into BDS will not be in Highlander
  • Native Win64, the No1 request from the customers, won't be in Highlander
  • Unicode, also one of the most often requested features, won't be in highlander
  • A Linux strategy won't be in highlander
  • Product quality isn't mentioned at all on the Highlander roadmap, neither is documentation quality.

Summing this up:

  • Highlander does not contain even a single of the most requested features of the native code customers
  • Highlander does not contain the single most requested .NET feature on QC.
  • In the .NET area, Highlander will not be competitive to Microsoft Visual Studio offers. By the time Highlander will hit the market, Microsoft will have released .NET 3.0.

Let's have a look at the groups of potential buyers for a new Delphi version again:

  • Highlander will offer next to nothing to group a). Especially all customers still sticking with Delphi 5,6,7 won't upgrade.
  • Group b) might be upgrading if they are currently using Delphi 2005 or 2006, in the hope of getting more stable IDE features. Many might instead stick to their current version, as most new work is done in Visual Studio anyway
  • Group c) might be upgrading
  • Group d) will upgrade
  • Group e) might buy the product
  • Group f) might buy the product if they haven't bought a previous releases (Turbo Delphi 2006) in the meantime, simply because Highlander then will be the current product offering. Without Highlander, those sales would instead go to the 2006 version of Delphi. This makes this group irrelavant for Highlander.

Summary:

Highlander will only sell to groups we have identified to be either small or nonexistent.

As we know that 80-90% of the customers who bought Delphi 2006 did so because of native development and are not represented on the Borland roadmap, many of them might be leaving Borland, looking for alternative solutions. In fact, from the feedback gathered many only still are with Borland because right now because they don't have a migration path away from Borland yet.

All in all, we can expect Highlander sales to be far worse compared to the ones from Delphi 2006.

The conclusion is that Highlander will be a failure.

"Delphi for Vista"

"Delphi for Vista" is planned to target the following things:

  • Windows Presentation Foundation (WPF)
  • Windows Communication Framework (WCF)
  • Managed C++ support

This is interesting, as it sounds like a gigantic effort. It basically means replicating everything Microsoft has done during the last years, trying to catch up. The fun part is that the features planned do no apply to any single group of current customers listed above. None of the features mentioned has been requested by anyone. The market for managed c++ is known to be a very small one, currently 100% taken by Microsoft.

Looking at it, this seems be be an insane plan - attack Microsoft in a small market they are years ahead with a 100% market share, while leaving behind your complete own customer base and all markets you are in yourself.

The motivation to put "Delphi for Vista" onto the roadmap appears to be a huge miracle.

"No code name yet" Delphi/C++ for Win64

It doesn't have a code name yet, it doesn't have a list of features planned yet, and the release date sounds highly unrealistic - just half a year after the biggest technological effort ever done at Borland for the the product with no market "Delphi for Vista", a Win64 release of Delphi is planned.

One has to think that there actually are no real plans for this yet, and that this was planted onto the roadmap simply to reduce the number of customers leaving Borland in frustration.

7, yes SEVEN years after the last native code innovation (Delphi 7, Kylix) from Borland, only one single feature of those required by the far majority of todays customers is going to implemented.

4.) Outlook

From the information gathered, the conclusions taken and from the impression the current roadmap gives, the conclusion is:

During the next few years, Borland/DevCo/BTP is going to lose about 80-90% of todays customers. This will be accelerated by the fact that one of the most important factors to Delphi's success is the component market. Due to the loss of Delphi customers, the component market already is very fragile as it is today, with already more than 50% of all commercial Delphi component vendors either going out of business or moving to other markets. As we have seen the component market also is highly dominated by native code components. Borland losing just another 20% or so their customers will mean that the component market will finally break down completely.

Borland's roadmap implies a high risk that the whole Delphi market and community will finally cease to exist. The roadmap appears to be a complete harakiri.

Summing all this up we can say: If you are native Delphi code user, a third-party component vendor, or part of the Delphi community, you will have to act now. You'll have to urge Borland to completely scrap the roadmap and create a new one from scratch, based on actual customer needs.

If you fail to do this, you should start working on a migration plan away from Delphi today. "Maybe they'll finally wake up in 2 or 3 years" won't help anymore, as it most likely will be too late by then.

5.) The alternative Roadmap

So, what could be done to turn the tides?

It's quite simple, actually: Simply do exactly what your customers ask you for. It's OK to have visions (and .NET clearly is one), and it's also OK to implement them if you can afford it.

The point is:

Borland can no longer afford investing more and more time, energy and money into visions instead of actual customer requirements.

Borland needs to completely scrap the current roadmap, and they need to do it now. Then they should do a detailed market survey just as I did, but with all the heaps of information they have in their hands.

And then, a new roadmap should be created. It could look like this:

Delphi 2007

  • Preview of Win64 compiler (Commandline, Win32 IDE, no Win64 debugger yet).
  • RTL and non-visual VCL classes support for Win64.
  • .NET 2.0 features that already are done and don't need much more energy to finish
  • Completely reviewed documentation. Working help system, possibly merging in all Help info that was lost after Delphi 7.
  • Partial Unicode support to prove that they are working on it
  • Continue improving IDE stability. Make sure the IDE is snappy and fast in the default installation mode, without the user having to remove parts of the IDE
  • Improve IDE support for C++
  • Improve the C++ compiler

In the Delphi 2007 timeframe, also re-activate the community efforts around Linux, so that a Linux RTL/VCL version can be done by the community for Delphi 2008. Ship CrossKylix again on the partner-CD, so that Delphi 2007 already inofficially will allow you working on Linux ports of server/middleware/apache/etc applications.

Delphi 2008

  • Win64 compiler and Win64 debugger inside Win32 IDE
  • Full Win32/Win64/Unicode VCL
  • Linux compiler integrated into the IDE
  • Partial Linux VCL provided by community project
  • Linux remote debugger integrated into the Delphi IDE provided by community project
  • If there still is market demand for Delphi.NET, and you have generated enough revenue with Delphi 2007 to be able to afford it, start working on advanced .NET 3.0 features. If not, feature freeze for .NET.
  • Improve IDE support for C++
  • Improve the C++ compiler

6.) Comparing the current and the alternative Roadmap

As we know, Delphi has lost market share massively. Therefore, it's important to boost sales number quickly again and to generate revenue before again investing into visions.

Let's compare the roadmaps:

The current Highlander plan is only attractive to roughly 10-20% of Borlands current customers. My alternative plan will be attractive to 80-90%.

My alternative plan comes with a nice bonus, too: As we know there is a huge chunk of customers that haven't upgraded at all during the last 4 years. We know that 100% of those customers are doing 100% native code only, because Delphi5/6/7 don't support .NET. You'll get revenue from many of them.

But there's more: The other big market player in native RAD Land, Microsoft, has just left the fields. Native Win32 RAD will bring in new customers. You are now able to conquer this market, instead of competing with Microsoft on a market they define.

I'm not a C++ guy, but from 1a) we know that there still also is a high interest in native RAD C++ development - the download counts for C++ are twice as high as those for Delphi.Net! Therefore, stop treating your C++ customers the way you did during the last years, and go get their money, too.

Due to the roadmap, there also will be another group of new customers:

Those interested in producing fast, native applications for multiple platforms.

I'll just give you one example: There is a gigantic market of PHP users. There is a gigantic market for Linux-based web applications.

Both typically run under Linux.

So, reactivate and refresh the old code you still got from Kylix, and enable those markets to take full benefit of the RAD toolchain you got (RAD IDE, dbexpress etc). Market Delphi as the ideal tool to write Apache modules and PHP extensions (buy in http://members.chello.be/ws36637/php4delphi.html, the technology is already there). You'll become market leader on a gigantic big market which still is growing every day, with next to no cost.

Summing this up:

Dear Borland,

Your current roadmap is based on implementing your visions instead of customer requirements. Your current roadmap is based on fighting for market share against the #1 leader, Microsoft. Your current roadmap is full of expensive investments, consuming all your resources. Your current roadmap is full of high risk, with very little to gain.

My proposed roadmap however is based on your customer requirements, generating confidence and revenue among those quickly again. It's also based on completeley taking over a still really big market with no competitor who's much stronger than you in sight. It also includes implementing paths to the future your customers will need. And it includes conquering new attractive markets with very low risk, but high possible gain.

Go, switch to the roadmap of success!

Greetings,

Simon Kissel

Meta-Information about this article

Document revisions

  • Oct 13th, 12:00 CEST: I've added an "abstract" section. This hopefully will make sure that abstracts about this article posted on other sites get a little more accurate - I've seen abstracts about this article saying that I'm asking Borland to drop .NET, which is not what this proposal is about.
  • Oct 13th, 12:26 CEST: In the initial version, my article was saying that if Borland would follow my alternative roadmap, "all" customers still using Delphi 5/6/7 would finally upgrade, generating revenue. Obviously "all" is an exaggration. I've changed "all" to "many".
  • Oct 13th, 12:35 CEST: I'll now add a FAQ, responding to frequently given comments and questions about my document. Thanks for all your feedback, keep it coming :)
  • Oct 30th: Borland has now posted a public customer survey about BDS' future

Trackbacks

Links to blogs or sites on the web who link to this article.