Communicating your project is an essential step for getting users. Here I summarize my experience from working on several different projects including KDE (where I learned the basics of PR - yay, sebas!), the Hurd (where I could really make a difference by improving the frontpage and writing the Month of the Hurd), Mercurial (where I practiced the minimally invasive PR) and 1d6 (my own free RPG where I see how much harder it is to do PR, if the project to communicate is your own).
Since voicing the claim that marketing is important often leads to discussions with people who hate marketing of any kind, I added an appendix with an example which illustrates nicely what happens when you don’t do any PR - and what happens if you do PR of the wrong kind.
If you’re pressed for time and want the really short form, just jump to the questionnaire.
Before we jump directly to the guide, there is an important term to define: Good marketing. That is the kind of marketing, we want to do.
The definition I use here is this:
Good marketing ensures that the people to whom a project would be useful learn about the project.
Good marketing starts with the existing strengths of a project and finds people to whom these strengths are useful.
Thus good marketing does not try to interfere with the greater plan of the project, though it might identify some points where a little effort can make the project much more interesting to users. Instead it finds users to whom the project as it is can be useful - and ensures that these know about the project.
Be fair to competitors, be honest to users, put the project goals before generic marketing considerations.
As such, good marketing is an interface between the project and its (potential) users.
This guide depends on one condition: Your project already has at least one area in which it excels over other projects. If that isn’t the case, please start by making your project useful to at least some people.
The basic way for communicating your project to its potential users always follows the same steps.
To make this text easier to follow, I’ll intersperse it with examples from the latest project where I did this analysis: GNU Guile: The GNU Ubiquitous Intelligent Language for Extensions. Guile provides a nice example, because its mission is clearly established in its name and it has lots of backing, but up until our discussion actually had a wikipedia-page which was unappealing to the point of being hostile against Guile itself.
To improve the communication of our project, we first identify our target groups.
To do so, we begin by asking ourselves, who would profit from our project:
Try to find about 3 groups of people and give them names which identify them. Those are the people we must reach to grow on the short term.
In the next step, we ask ourselves, whom we want or need as users to fullfill our mission (our long-term goal):
Again try to find about 3 groups of people and give them names which identify them. Those are the people we must reach to achieve our longterm goal. If while writing this down you find that one of the already identified groups which we could reach would actually detract us from our goal, mark them. If they aren’t direly needed, we would do best to avoid targeting them in our communication, because they will hinder us in our longterm progress: They could become a liability which we cannot get rid of again.
Now we have about 6 target groups: Those are the people who should know about our project, either because they would benefit from it for pursuing their goals, or because we need to reach them to achieve our own goals. We now need to find out, which kind of information they actually need or search.
GNU Guile is called The GNU Ubiquitous Intelligent Language for Extensions. So its mission is clear: Guile wants to become the de-facto standard language for extending programs - at least within the GNU project.
This part just requires thinking ourselves into the role of each of the target groups. For each of the target groups, ask yourself:
What would you want to know, if you were to read about our project?
As result of this step, we have a set of answers. Judge them on their strengths: Would these answers make you want to invest time to test our project? If not, can we find a better answer?
If our answers for a given group are not yet strong enough, we cannot yet communicate our project convincingly to them. In that case it is best to postpone reaching out to that group, otherwise they could get a lasting weak image of our project which would make it harder to reach them when we have stronger answers at some point in the future.
Remove all groups whose wishes we cannot yet fullfill, or for whom we do not see ourselves as the best choice.
Now we have answers for the target groups. When we now talk or write about our project, we should keep those target groups in mind.
You can make that arbitrarily complex, for example by trying to find out which of our target groups use which medium. But lets keep it simple:
Ensure that our website (and potentially existing wikipedia page) includes the information which matters to our target groups. Just take all the answers for all the target groups we can already reach and check whether the basic information contained in them is given on the front page of our website.
And if not, find ways to add it.
As next steps, we can make sure that the questions we found for the target groups not only get answered, but directly lead the target groups to actions: For example to start using our project.
For Guile, we used this analysis to fix the Wikipedia-Page. The old-version mainly talked about history and weaknesses (to the point of sounding hostile towards Guile), and aside from the latest release number, it was horribly outdated. And it did not provide the information our target groups required.
The current Wikipedia-Page of GNU Guile works much better - for the project as well as for the readers of the page. Just comare them directly and you’ll see quite a difference. But aside from sounding nicer, the new site also addresses the questions of our target groups. To check that, we now ask: Did we include information for all the potential user-groups?
So there you go: Not perfect, but most of the groups are covered. And this also ensures that the Wikipedia-page is more interesting to its readers: A clear win-win.
Additional points which we should keep in mind:
For whom are we already useful or interesting? Name them as Target-Groups.
Whom do we want as users on the long run? Name them as Target-Groups.
Use bab-com to avoid bad-com ☺
The mission statement is a short paragraph in which a project defines its goal.
A good example is:
Our mission is to create a general-purpose kernel suitable for the GNU operating system, which is viable for everyday use, and gives users and programs as much control over their computing environment as possible. → GNU Hurd mission explained
Another example again comes from Guile:
Guile was conceived by the GNU Project following the fantastic success of Emacs Lisp as an extension language within Emacs. Just as Emacs Lisp allowed complete and unanticipated applications to be written within the Emacs environment, the idea was that Guile should do the same for other GNU Project applications. This remains true today. → Guile and the GNU project
Closely tied to the mission statement is the slogan: A catch-phrase which helps anchoring the gist of your project in your readers mind. Guile does not have that, yet, but judging from its strengths, the following could work quite well for Guile 2.0 - though it falls short of Guile in general:
GNU Guile scripting: Use Guile Scheme, reuse anything.
We saw why it is essential to communicate the project to the outside, and we discussed a simple structure to check whether our way of communication actually fits our projects strengths and goals.
Finding the communication strategy actually boils down to 3 steps:
Also a clear mission statement, slogan and project description help to make the project more tangible for readers. In this context, good marketing means to ensure that people learn about the real strengths of the project.
With that I’ll conclude this guide. Have fun and happy hacking!
— Arne Babenhauserheide
In free software we often think that quality is a guarantee for success. But in just the 10 years I’ve been using free software nowadays, I saw my share of technically great projects succumbing to inferior projects which simply reached more people and used that to build a dynamic which greatly outpaced the technically better product.
One example for that are pkgcore and paludis. When portage, the package manager of Gentoo, grew too slow because it did ever more extensive tests, two teams set out to build a replacement.
One of the teams decided that the fault of the low performance lay in Python, the language used by portage. That team built a package manager in C++ and had --wonderfully-long-command-options without shortcuts (have fun typing), and you actually had to run it twice: Once to see what would get installed and then again to actually install it (while portage had had an --ask option for ages, with -a as shortcut). And it forgot all the work it had done in the previous run, so you could wait twice as long for the result. They also had wonderful latin names, and they managed the feat of being even slower than portage, despite being written in C++. So their claim that C++ would be magically faster than python was simply wrong (because they skipped analyzing the real performance bottlenecks). They called their program paludis.
Note: Nowadays paludis has a completely new commandline interface which actually supports short command options. That interface is called
cave and looks sane.
The other team did a performance analysis and realized that the low performance actually lay with the filesystem: The portage tree, which holds the required information, contains about 30,000 ebuilds and almost 200,000 files in total, and portage accessed far more of those files than actually needed for resolving the dependencies needed to install the package. They picked python as their language - just like portage. They used almost the same commandline options as portage, except for the places where functionality differed. And they actually got orders of magnitude faster than portage - so fast that their search command often finished after less than a second while. portage took over 10 seconds. They called their program pkgcore.
Both had more exact resolution of packages and could break cyclic dependencies and so on.
So, judging from my account of the quality, which project would you expect to succeed?
I sure expected pkgcore to replace portage within a few months. But this is not what happened. And as I see it in hindsight, the difference lay purely in PR.
The paludis team with their slow and hard-to-use program went all over the Gentoo forums claiming that Python is a horrible language and that a C program will kick portage any time. On their website they repeated their attacks against python and claimed superiority at every step. And they gathered quite a few zealots. While actually being slower than portage. Eventually they rebranded paludis as just better and more correct, not faster. And they created their own distribution (exherbo) as direct rival of Gentoo. With a new, portage-incompatible package format. As if they had read the book, how not to be a friendly competitor.
The pkgcore team on the other hand focussed on good technology. They
created the snakeoil library for high-performance python code, but
they were friendly about it and actually contributed back to portage
where code could be shared. But their website was out of date, often
not noting the newest release and you actually had to run
--help to see the most current commandline options (though you could
simply guess them if you knew portage). And they got attacked by
paludis zealots so much, that this year the main developer finally
sacked the project: He told me on IRC that he had taken so much
vitriol over the years that it simply wasn’t worth the cost anymore.
Update: About a year later someone else took over. Good code often survives the loss of its creator.
So, what can we learn from this? Technical superiority does not gain you anything, if you fail to convince people to actually use your project.
If you don't communicate your project, you don't get users. If you don’t get users, your chances of losing motivation are orders of magnitude higher than if you get users who support you.
And aggressive marketing works, even if you cannot actually deliver on your promises. Today they have a better user-interface and even short option-names. But even to date, exherbo has much fewer packages in its repositories than Gentoo. If the number of files is any measure, the 10,000 files in their special repositories are just about 5% of the almost 200,000 files portage holds. But they managed quite well to fraction the Gentoo users - at least for some time. And their repeated pushes for new standards in the portage tree (EAPIs) created a constant pressure on pkgcore to adapt, which had the effect that nowadays pkgcore cannot install from the portage tree anymore (the search still works, though, and I still use it - I will curse mightily on the day they manage to also break that).
Update: Someone else took over and now pkgcore can install again.
So aggressive marketing and doing everything in the book of unfriendly competition might have allowed the paludis devs to gather some users and destroy the momentum of pkgcore, but it did not allow them to actually become a replacement of portage within Gentoo. Their behaviour alienated far too many people for that. So aggressive and unfriendly marketing is better than no marketing, but it has severe drawbacks which you will likely want to avoid.
If you use overly aggressive, unfriendly or dishonest communication tactics, you get some users, but if your users know their stuff, you won’t win the mindshare you need to actually make a difference.
If on the other hand you want to see communication done right, just take a look at KDE and Gnome nowadays. They cooperate quite well, and they compete on features and by improving their project so users can take an informed choice about the project they choose.
And their number of contributors steadily keeps growing.
So what do they do? Besides being technically great, it boils down to good marketing.
⚙ Babcom is trying to load the comments ⚙
This textbox will disappear when the comments have been loaded.
Note: To make a comment which isn’t a reply visible to others here, include a link to this site somewhere in the text of your comment. It will then show up here. To ensure that I get notified of your comment, also include my Sone-ID.
Link to this site and my Sone ID:
This spam-resistant comment-field is made with babcom.