When I was selling open core ETL software @Talend a decade ago, we would be selling against hand-coding (customer doing nothing) vs. Informatica ($$$$ closed core software). That's it!

What has changed over the last decade? ...

For one, Informatica's growth has stalled around $1B vs. $600M~ a decade ago. Clearly a dying now legacy laggard in a market they largely created.

Meanwhile, Talend grew from $20M to $250M+ revenue with cloud growing 100% YoY today! Open core at the infra layer won.

Second, the ETL market / data integration market for commercial product has even FURTHER disaggregated from the fairly noisy env. even in 2010... Today, people use one or a combo of the following tools to do ETL: Spark, Flink, dbt, Fivetran, Beam, Flume, Hadoop (yep) + more.

What I learned in 2010 that is probably still true today:

Most ETL work is not done by ANY of these tools, but instead by hand-coding programs that manipulate and surgically transform complex datasets.

Talend also limited their market coverage due to their concentration on Java as the core runtime code generator... a great deal of ETL work is done off-JVM.

Another interesting thing I noticed:

Talend was very skilled at maintaining a carefully governed open core but only for connector/plugin contributions, not core runtime/studio (largely only Windows Desktop) product R&D...

This resulted in Talend being able to earn the trust of FOSS community stakeholders at the level of instrumenting and collecting leads for EVERY user of the FOSS core... which resulted in the sales organization seeing something on the order of 10-15K+ inbound leads per MONTH.

How was Talend able to convert against those leads in a marketing funnel powered by what I think was Eloqua at the time... they didn't! Conversion rates were HORRIBLE (< 1%).

However, what I learned by studying COSS companies over the last nearly entire decade is the following:

While COSS product revenue often results from horrible conversion rates, what is actually happening is a lot more nuanced...

Highly engaged FOSS end users are chatty. But they are not the buyer, they just want a support question answered (Talend didn't use GitHub issues and the FOSS projects were hosted on Talend's landing sites and forum).

The buyer is indirectly discoverable THROUGH the FOSS user.

Over time, this became referred to as "bottoms up" adoption.

A decade ago, a lot of people had to push against the inertia of legacy vendors selling top-down as THE ONLY motion for decades prior.

In reflecting on who the best reps were, it was not technical chops, qualification skills, strategic alignment, getting a good patch.... you just needed to focus on the 2-3 deals per QUARTER that would be in the top 20% globally and you would immediately be in the top 1%.

What most COSS companies that look/feel similar to this story need is exactly what Talend needed a decade ago.. and it STILL does not exist:

When you are generating thousands of "potentially qualified" leads per month, reps should be equipped with up-to-the-minute fresh data...

On the following: Identity discovery (instant name to employer "table joins" for lossy/incomplete leads with nothing but name/email)

Community engagement tooling (a kind of CRM purpose built to analyze open source engagement)

Will blog more on this. That's all for now! So many more thoughts here.