Open Source Startup Marketing: Initial Steps

As innovative open source startups gradually grow ambitious enough to start thinking about an enterprise-class client portfolio, marketing becomes more of a pressing priority. My observation is that marketing is overwhelmingly regarded by open source geeks (an accolade I aspire to myself despite my sub-standard code-fu) as either superfluous nonsense-ridden hand waving, as arcane dark arts taught only in the bowels of the netherworld, or somewhere in between.

For a smaller open source startup, it doesn't have to fall into either category. It can be simple and straightforward, and that is what I have written this article for: to reduce the dimensions of kickstarting the marketing function for an open source startup by emphasizing on key areas with disproportionately emphatic results. More bang for your unit of effort. Less of a marketing department, and more of an image and a strategy for how to access the enterprise client market.

Think of this as an "We're a small open source startup and we're all arsekicking engineers, but it looks like we need to market if we're going to attract lucrative clients" HOWTO.

This document contains a brief description of what is reasonable as initial steps in kick-starting the task of proper marketing for an open source startup. Where slightly more specific examples are useful, we'll refer to a fictitious wiki company.

Caveat emptor:

• The steps described herein are not canon; this is not a stone tablet, and I am not Moses. These are guidelines to be placed into context of the specific startup.
• The work items listed in this document are initial steps. This is the good reason why this document does not contain action items such as "Participating in $TRADEFAIR" or "Getting $SUPERDEVELOPER interviewed on itmj.com".
• We are assuming that the purpose of the marketing is to get the attention of enterprise clients. If the open source startup is targeting compatriot geek cognoscenti, then I have two conclusions:
  first that the marketing function isn't required since the engineers will do just find, and second that (in all likelihood, with few notable exceptions) that solid startup growth isn't on the list of objectives.

• The recommendations in this article are conducive to laying the foundations for a marketing function which is capable of accessing the enterprise market. Conducive to. You are creating the conditions which will achieve that, not running off and speaking fluently and capably to enterprise clients right off the bat. I'm not selling silver bullets.

General Direction

The first consideration is to provide coherent strategic context to actions. Understanding and managing data influx and corresponding outbound communication is in the first two steps. Building on the notion that marketing is about information, a small set of specific tools is described in step 3. Most importantly, a clear strategy is called for in the 4th step.

Steps

Control the information

By information, I mean that which is conducive to better marketing intelligence. This can include (and is certainly not limited to):

• Web buzz: Are we on del.icio.us? Digg? reddit? How many bookmarks, to which pages? Are people blogging about us? At what frequency? Who has tested or evaluated our system? How does all of this compare to the competition?
• Market feedback: Are we soliciting/safe-keeping/analysing feedback from the current client base? Are we acquiring feedback from leads irrespective of eventual lead results? Who uses the systems identified as competing for our niche, and why? What proportion of the installed base do not avail themselves of our products, and why?
• Access logs (and other manifestations of market interaction with our online presence): securing them, making sure they present an optimal array of information (e.g. useful information is not thrown away), etc.

The objective of information control as a first step is to create circumstances as early as possible where marketing strategy is based on intelligence rather than on instinct and educated best-guess common sense.

Observe the critical fact that we say “control the information” which here is semantically intended to be a subset of “acquire the information”. Controlling it means not only acquiring it but setting up a systematic and periodical information re-acquisition effort to enable trend-spotting and performance monitoring. This needn't be as complex as it sounds; it can be as simple as just checking the relevant websites or running trivial scripts against logs. We aren't setting up an IBM marketing department from scratch, we are just trying to maximize information control considering our scale.

Preliminary message control

As a first step in message control, we pin down what the message has been so far, and what needs to be changed. Usually, the first and largest change is that a message needs to be defined to begin with, or that he concept of marketing message should be changed from "what we communicate our product(s) to be" to "what impression of our products we leave in the market's mind".

A simple way of doing this is enumerating keywords central to the current message, evaluating them, and testing alternatives. Current keywords might be: "wiki", "platform", "open source", etc. To exemplify this exercise, all three can be considered sub-optimal:

• “wiki” might smack of wikipedia which to some minds has connotations with words such as “amateur”, “arachic”, “chaos”, etc. (We're thinking in terms of how others see the concepts here, not how we see them. Remember, marketing is about them not us.)
• “platform” also is problematic: almost any other enterprise information system will bill itself as a platform, which dilutes the market positioning. It is understood that the product is a fundamentally transformative platform, but the word “platform” has been diluted by industry overuse to the point where it doesn't effectively communicate this.
• “open source” can be an effective message constituent when the decision maker at a potential client is technical, or if they are non-technical but relatively enlightened. This is not always a given, and often the contact person at a lead will need to sell the solution internally. For this purpose, the message keyword “open source” is not guaranteed to be effective. It is more effective to proclaim the open source advantage after the sale has established a value proposition advantage (features/capabilities to outlays). “Look, it can do this (ooh!) and that (aah!), your organization will benefit like this (nifty!). Oh, and it's open source too (here, take my checkbook!)” rather than “This is an open source system (meh...) which implements an enterprise (argh!) wiki (oh, like wikipedia) platform (oh, like Oracle)”.

For the purposes of preliminary message control, it can be a simple way of making the change by identifying:

• Five things we should stop saying (e.g. “we make a very cool wiki”)
• Five things we should start saying (e.g. “people shepherd their information with our systems”)

It would also be correct to develop an “elevator pitch” (a 30 second spoken blurb) for the startup employees. Rather than hammer it into the startup employees' heads, it can be trained in in an hour-long marketing message session (which is not difficult to facilitate).

Establish marketing tools

Segment analysis

A set of basic marketing tools are required. The first is a segment analysis by SKU, which prerequires clear SKUs. The identification and definition of the product SKUs is usually already relatively clear. An initial segment analysis is required and a schedule for updated segment analyses likewise. The objective of this analysis is to get a granular understanding of the startup's product offerings and the relative success of each in relation to the whole.

Segment analysis examples

• Revenue/profit per SKU, or revenue/profit per SKU family
• SKU contribution to aggregate revenue/profit growth as a trend

Segment analysis sample use cases

• Triaging of product offerings (usually optimally conducted through simple Pareto analysis)
• Identification of sustained product under-performance in individual SKUs or SKU families to help define marketing focus on these
• Pricing decision support, as in the theoretical scenario where training demand is growing faster than the average of the other products, justifying price increases or staffing expansion

Visibility metrics and performance stats

At the risk of over-simplification, generating the type of visibility conducive to effective sales generation is probably the primary marketing output expected. As such, the product(s)/brand(s) visibility needs to be tracked and measured. Again, we aren't setting up an IBM division for visibility tracking, so the objectives here are:

• Sufficiently indicative without sacrificing innocent mammals at the altar of accuracy
• Low-overhead: minimal cost, minimal time invested in setting up and maintaining
• Correlatable where possible to sales generation (e.g. tracking proportion of contracts resulting from what form of visibility: who found out about our product(s) from freshmeat, ended up using it, and ended up on a support contract, who found it through a wiki comparison matrix somewhere, etc.)

Sample visibility metrics may be:

• Distance from top $SEARCHENGINE rank on key search terms (e.g. enterprise wiki); to make something like this a meaningful stat, it is best assessed as a bundle of relevant key search terms; see appendix A.
• Community activity (possibly benchmarked against competing systems; some simpler metrics might be mailing list activity or average IRC channel population)
• Other visibility metrics derived from the web buzz and market feedback generated from point one above, controlling the information.

Competitive analysis

A traditional SWOT analysis serves just fine here, and it doesn't need to be an exercise for which a platoon of crack MBAs is hired. Acknowledging that a SWOT matrix development exercise carries some moderate overlap with other initial steps and in one case would ideally be a prerequisite (as with the message development), a non-intensive first version should still be developed and then updated at regular periods (likely annually).

The SWOT analysis should not be the exclusive effort of one man who is charged with the marketing, but it is the function of this person to steward the development of a relevant, complete, and reasonably accurate SWOT analysis in collaboration with other senior management to satisfy the following conditions:

• The responsibility of the SWOT analysis remains squarely with the person charged with marketing
• Senior management input is solicited in a time-effective manner (i.e. building straw man or 0.1 analysis to facilitate input acquisition)

The critical marketing decision of “do we overwhelmingly do our thing” or “do we carefully consider the competition wherever possible” and the balance between the two attitudes should become clearer with a clean SWOT analysis. Though, it should be mentioned, in an industry with strongly differentiated products and services such as the enterprise platform/wiki marketplace, there should be a very strong reason if the startup considers the competition overwhelmingly in defining a marketing position, message, and strategy.

Formal marketing strategy development

The value of having a formal marketing strategy is this:

• Specific actions and initiatives have clearer expected outcomes since they can be traced more easily up the strategy hierarchy to objectives
• At the lower levels of the strategy will be actions. When actions are framed in a hierarchy it is easier to get more out of efforts (pushing aside the fancy talk with an over-simplified example, if one goal calls for better client feedback and another calls for more sustained evangelizing, then it is easier to recognize that an automated e-mail update sent out to the community to inform of package updates can be used to also solicit feedback).

The OGSM strategy format is simple enough for our purposes. The strategy itself should be updated on a rolling horizon as follows:

• Vision, objectives, and goals: every 3 years
• Strategies and measures: annually

Some sample elements could look as follows (notice how each subsequent level refers to the parent node through the assigned number):

Objectives:

• 1. Become known as the prime enterprise wiki player
• 2. Establish lead generation resulting directly from marketing initiative as the primary sales driver
• 3. Generate sustainable organic community growth

Goals:

• 1.1. Reach a visibility rank < 50% over baseline
• 1.2. Account for X% of new wiki deployments in market segment Y
• 2.1. Surveying new clients, initial contact resulting from marketing activity > 50%
• 3.1. > X% of community activity from actors outside the startup's employ
• 3.2. > X% of code contribution from actors outside the startup's employ

Strategies

• 1.1.1. Focus on improving visibility through independent community/user interaction (more sustainable and scalable than generating better ranking through own efforts – rather than working to create better ranking, working to create circumstances conducive to better ranking)
• 1.1.2. Better mention in publications/press (interviews, reviews, etc.)
• ...
• 3.1.1./3.2.1. Increase the install base of the product irrespective of sales generation (more installs gives more itches to scratch which gives more developers)4
• ...

Measures

• 1.1.1.1. X% of new search engine results originating from the community/userbase.
• 1.1.1.2. X% of new blog posts concerning enterprise wikis link to the startup's product, in this example a wiki (this is an example only; a matric like this might not be too fun to track)
• ...
• 3.1.1.1. X new independent deploys of the startup's product per $TIMEPERIOD
• ...

Appendix A

Example of bundled key search term stat (more accurate than maintaining simple single search stats):

Data:

Chart: