As I write these lines, I still don’t know who dunnit in “Only Murders in the Building,” and this is my late-summer source of anxiety (in addition to the war in Ukraine, the climate crisis, and the state of U.S. politics). So I have to say, writing this newsletter is rather refreshing.
Sure, there is no conclusive outcome to this week’s Part II into the “Zero to One of the Data Team,” but unlike the four above-mentioned topics, nobody dies. That’s the bar for me at the moment.
I hope your end of summer is restful and generally worry free.
All my best, Ariane
The org chart and the data team
In my last newsletter, we looked at the evolution of the data team across three stages — from having support from data in the company (though necessarily, in fact, a data team in name) through the third stage where the first data science employee joins.
This week, we’re looking outside of the data team itself to focus on how the data team lives in the media organisation.
When Sanda Loncar, the head of the product and data team at Kleine Zeitung in Austria, dropped by my inbox a few weeks ago, she was interested in discussing whether her data team (five people) was positioned in an optimal way in the organisation. Is there an org chart that suits data teams best? I’d say this is somewhat above my pay grade, but there definitely are certain trends and they speak both to the company’s past, but also its culture.
Without a doubt, data teams live in various places at media organisations. The below analysis was done without a proper benchmark but is observational based on companies I have interacted with in the past year, mostly in Europe and the U.S.
There are two broad cases:
Data is part of the product organisation. In my observation, this means that engineering is also under product. This is the product-led organisation, and it’s getting more prevalent (in purple in diagram above).
Data is part of the engineering organisation, and this seems to align with product also being under engineering. This is the engineering-led organisation and it’s prevalent in tech but not in publishing (in yellow in diagram above).
In both cases, there is a variant where the data team is split between its functional responsibilities (insights) and its enablement (engineering). You’ll find this variant in both product-led and engineering-led organisations. We will talk about this in a little bit.
Now, back in the day, marketing-led publishing organisations were actually a dominant model. Marketing owned product (basically, all your big media companies 10 years ago were in that model), and insights/survey, etc., were very much tied to marketing. You still do find data under marketing, but I would say this is on its way out. For example, The Economist used to be organised like this several years ago but then elevated both a chief product officer and a chief data officer. Marketing stayed in that direct link to the CEO, but the data and product functions moved from under it.
Why is there not a more dominant model?
I think the biggest reason is that data teams at most publishers are still on the smaller side and still fairly young (most data teams were not around 10 years ago). Therefore, the way they emerged — and what part of the company drove their emergence — is the proverbial tree under which the apple falls.
Unlike other parts of our organisations that have been the subject of deep reorganisations in the past decade — certainly hastened by the revenue changes we’re seeing around advertising and print — data was built mainly in support of the digital product and, from birth, oriented at digital revenue.
But two things seem to hold up:
Data, by and large, is where it first started and grew. This is why you’ll see data under product or engineering at media companies that have equally well-demonstrated data chops. There isn’t a “when you are real about data you do X, and when you’re still junior you do Y.” (We’ll look into examples of this in just a moment). My analysis is that this is because data is still young at most companies and, more and more, data will move to join up with the group it most supports.
For orgs where data grew from analytics and business intelligence, they have gone closer to product. For teams where data came from engineering and where analytics and BI haven’t been merged and stayed with marketing, data is still a bit split with insights sticking with marketing and data with engineering. My analysis is that this model is on the way out and that data-in-marketing will get more rare.
But there is a later stage model where data runs under its own steam, on the same level as product and engineering. It may look like this (above) and this speaks (counterintuitively) less about the data team itself than it does about the way the company uses and considers data.
Where data fits in the org chart speaks to what mission data primarily contributes
Let’s look at our org charts a bit more closely — the product-led organisation versus the engineering-led organisation:
➡️ Data falls under the business-side engine of the company … whether product or engineering
Data can fall under either. The question is: Is there a reason why an organisation is going to put data under product or under engineering? Let’s look at two cases here: Financial Times and Axios.
I think we can all agree that the FT is a long-time leader when it comes to having extremely advanced research and development for its data programme, usage, and engineering. I mean, from the outside, data seems to rule everything at the FT (back in the early 2010s, it already had a data team of 30+). At the FT, data falls under product. The chief product officer runs product, engineering, and data.
Meanwhile, Axios’ story couldn’t be more different than the FT — a young company, born digital, no legacy revenue. Axios runs data (and product) under its engineering organisation. The CTO drives product, engineering, and data.
But Axios isn’t product-led … it is journalism-led.
Its three founders come from a deep editorial background and (famously) built Axios to meet a need that expressed itself in very journalistic terms: “Why it matters.” So its choice of an engineering-led organisation on the business side is very logical: When journalism is a dominant current of the product’s development, it is the product. The New York Times was slower to build the world-class product team it now has because the newsroom for years, essentially, was the product team (I speak from insider knowledge here).
Looking at the enablement of its business, Axios wrapped all under tech because product was already heavily “represented” in the other force of the company, which was the journalism itself.
➡️ The same org chart can be used by a small or large organisation
Something that’s interesting here is that in both cases, these are organisations with both very solid and advanced foundations in data — both in terms of the relative size of the team and the ways in which data plays a part in how the product is built, how the company is run and markets its products, and how it uses data in its products.
So it’s not like one type of org chart speaks to “big companies” vs “small companies” or “companies that have large data teams” vs “companies that have smaller data teams.”
I mentioned Sonda Loncar, the head of the product and data team at Kleine Zeitung in Austria. Her data team runs under product and it works for them. That’s the model of the FT. It works for them, too. There’s no doubt these are pretty different organisations, but in this case, their org charts look similar.
➡️ Data under its own steam, reporting to the CEO
But there is a model of an org chart we haven’t really discussed yet and which does seem to reliably track with a later stage: The org chart model where data is under neither the head of product nor the head of engineering (whether these are C-level titles or VPs). This is the model at The New York Times — data reports directly to the CEO, as does product and as does tech.
So on the one hand, you could certainly point out that it can just be for practical reasons: the data team of the NYT is now about 100 people. But there is another reason (or incentive) for this one, and it’s not just headcount.
The New York Times has publicly stated (on several occasions) its belief that the future of its overall revenue was direct consumer revenue. But at the moment, it still has a significant share of anonymous digital traffic, supported by advertising (and of course, there is subscriber-generated advertising revenue as well).
For the NYT, the evolution of its revenue between digital consumer revenue and digital advertising is still hanging closer to a middle split. Digital consumer revenue is higher than digital advertising, and the Times has gone full speed ahead acquiring more digital properties like The Atlantic to further complement the value of its subscription bundles.
On the other hand, the NYT also did acquire Wirecutter, whose revenue model is advertising-driven for the most part. Even if the strategic goal of The New York Times is to continue to build its direct consumer revenue, it certainly isn’t discounting the opportunity of advertising revenue.
But consider then the case of the FT — whose foray into digital consumer revenue is one of the earliest successful demonstrations that it could be done. At the FT, there is a clear path forward that consumer revenue can be the lion’s share of the company’s revenue. As of the end of 2021, the FT’s digital content revenue is three times bigger than its next biggest component — print.
This doesn’t mean advertising revenue is discounted. Of course, the FT may have a lower volume of pageviews than, say, CNN, but its revenue per page for advertising is very high because most of its traffic is logged in and seen by valuable subscribers, which fit the juiciest advertising demographic, to boot.
In other words, even where advertising revenue is concerned (and rich), it is so because product data is rich, too: “It’s the result of deeper audience data and, as a result, an increasingly effective marketing proposition. So it isn’t ads versus subs — there is strong growth to be had in both,” said FT’s CEO John Ridding at a conference last year.
A word about these somewhat imprecise statements: Nikkei Inc., the corporate owner of the FT, is employee-owned so doesn’t publish financial reports. There isn’t therefore a perfect way to reconstitutes this data, but I can confidently say that at the percentage of paid traffic the FT generates, there isn’t a realistic way to generate advertising revenue that would compare.
So we have two companies with very mature direct consumer revenue strategies on digital whose advertising revenue is comparatively becoming less important. But for the NYT, this is a cake still split toward the middle and with continued diversification pursuits that lean on advertising.
And to be sure, these are more than two-sided strategies: Events, print, and other ancillaries are a significant play at both the FT and the NYT — and Axios, too (for the events business).
But for our purposes, let’s consider advertising and product as the two main consumers and drivers of data for these companies.
- At The Times, there is a more even split between the various goals of the companies, and data would be in a tricky spot if it was run under product.
- At Axios (well, speaking about its activities prior to its recent acquisition by Cox), there is (or was) no ambivalence that the revenue play was direct consumer revenue, too. Axios’ traffic, while impressive for a young company, was never going to sustain its large newsroom via advertising.
And so the NYT put data under general leadership. But for the FT and Axios, where data has a clear “dominant” partner or customer, data falls under these umbrellas.
The case of the chief data officer
More publishers have been elevating data to the C Suite, but in terms of sheer numbers, that’s still uncommon. And it’s definitely something that really maps to larger organisations. The New York Times named its first chief data officer less than a year ago, for example — and its business-side head of data is a SVP who serves under its CEO.
If anything, the title seems to be more about parallelizing titles at the C-Suite when data has a direct seat under the CEO. Again, contrasting a NYT with a FT, the difference is likely not how “important” data is to the organisation. I think we can safely say data is important in both these organisations.
But where the NYT has brought out the data function to occupy its own independent seat, it also makes sense to reflect the parallelism of this decision in the titles given to functional leaders.
Further afield on the wide, wide Web
It’s been a hot minute since I linked to something from Cassie Kozyrkov, the chief decision scientist at Google.
This week, she explains, in very accessible terms, some of the risks and downsides of AI. It’s one of these articles that explains something readily transparent, except for the part where most people who have tried to explain this to you before were never as clear.
In related considerations, I’d love to have her for an INMA conference, but I know it’s going to take some doing.
Signed: A fan girl.
About this newsletter
Today’s newsletter is written by Ariane Bernard, a Paris- and New York-based consultant who focuses on publishing utilities and data products, and is the CEO of a young incubated company, Helio.cloud.