Against the Grain Vol. 33#4
If a machine is running well, it is said to be “humming” — meaning all things are running in unison and at optimal performance. Basically, this is the idea behind a new company started by Silverchair called Hum. It offers a new data analytics platform (also called Hum) that analyzes interaction across an entire organization, currently restricted to societies and associations, and helps engage and enrich the user journey. No data scientist required.
This is a fairly new category in publishing technology called a CDP or Customer Data Platform (yay, another new acronym!) — and we will likely start seeing things like this with more frequency if recent polls from the annual SSP meeting are any indication. Hum is a bit more than a standard CDP, though, which seems mainly a place to gather data; they also have some cool tools that help with the analysis, like machine learning and the potential for smart discovery. There are several examples of CDPs outside of publishing that can be quite costly and come from familiar names like Adobe, Microsoft, and Salesforce. Some come with enterprise-sized, one-market-fits-all approaches, others are more in tune with the retail markets they serve.
Hum seeks to be a better-fitting alternative to the publishing market, starting with societies and associations and a model that helps them scale and grow as they realize benefits from the product. They are also more aware of things that publishers have been paying attention to recently like GDPR and CCPA, having grown out of their sister company — and publishing industry stalwart — Silverchair.
I recently caught up with Dustin Smith the President and Co-Founder of Hum, who spoke to me from his home office in the Charlottesville area, near Hum’s corporate headquarters.
John Corkery: Could you tell us a little about your background in technology?
Dustin Smith: Prior to joining Hum, I was Director of Innovation Programs for the Inovo Group, a boutique innovation consultancy in Ann Arbor doing super interesting work with Fortune 500 companies and startups, helping them succeed at strategic innovation. I really enjoy opportunity discovery, bringing strategy to life, and market development.
JC: How did you come to join Silverchair?
DS: In my previous position we would innovate, then hand the results off to others to develop. The opportunity to take the most promising thing out of the Silverchair opportunity discovery work and then build a business — to take it forward and make something happen with it — was of interest to me.
With Silverchair, the journals and books hosting market is more mature and that’s why they brought me on initially, to find something complementary and adjacent, so I took them through the same sort of process I took the Fortune 500 folks. We conducted lots of primary and secondary research to uncover a whole lot of opportunities and invest time into the most promising ones. We also spoke with some of our clients in healthcare and the interesting things they were doing between the nexus of content and data. This seemed where the biggest opportunity might be: societies and associations have separate digital platforms, without much usable data, so the opportunity to unify their data into something useful was how the idea of Hum came into being.
JC: Why societies and associations?
DS: A very good question. The core of Hum is a Customer Data Platform (CDP). It’s not exclusively a CDP as people might define it, but at its core it is a CDP. Historically, CDPs come out of the consumer packaged goods and e-commerce industries where what you’re really trying to do are very crude sort of transactional sales — like trying to sell you Pepsi Cola, trying to sell you shoes — there’s not that much to talk about, you’re basically just a hawker standing on a box. Content marketing is a kind of cover for that sort of very commercial motive. They just want to talk to you so they can mention Pepsi or shoes. A CDP helps them target the right people for that message so they’re not just yelling at everyone.
The thing about societies and associations is that content is the way they deliver value, particularly in digital forms. It is really high-quality content that is pure in terms of its motive. So they are already engaging with their audiences offering really great content. Discoverability is definitely an issue and I think you suggested [some of the tech is] long in the tooth, and some of the formats are maybe a little bit futsy and not so engaging. As you’re able to get away from the myopic focus on, say, the scholarly or the very long form, you get to a place where you’re developing more interesting, engaging, and general purpose content which then starts to open up a broader set of audiences, so it’s not just the sort of core membership you are able to engage. And that gives you a lot more options.
As you’re engaging with more people, as you’re following those interactions, you are cultivating “first party” data [data around observations about your audience] and you’re more attuned to what people are interested in. You can develop smaller product offerings, particularly on the professional association side, but also on the society side. You’re able to engage with sponsors better. Instead of saying, “We send a newsletter to 70,000 and 5% of those people open it,” it’s “We know there are 20,000 people who are extremely interested in thoracic surgery.” And we’re ultimately going to work with the people who are at the vanguard of medical instruments or in pharmaceuticals to reveal some of the interesting work that they’re doing, and to potentially connect some of those partners with the content. Ultimately, Hum allows you to serve your existing membership and constituents better as it also opens up new business model options for them.
JC: From a technology standpoint, how was Hum developed? Does it sit on the Silverchair stack?
DS: Hum was developed from scratch. We’re a sister company of Silverchair, so we integrate with the Silverchair platform but are not strictly part of it. Hum is a different organization, with a different technology stack altogether.
We’ve developed Hum with a lot of knowledge from our CTO, Niall Little, who was Director of Architecture for Silverchair. He brought a lot of deep knowledge about how to handle content, how to create search and personalization recommendations, and discovery. We were able to bring that learning to a clean-sheet design, which has been very helpful.
We looked at the landscape in terms of pulling an open-source CDP off the shelf. Unomi by Apache ended up being really the only option, but they are much more heavily focused on the sort of CPG [Consumer Packaged Goods] side of a classic CDP use (which is a retail and e-commerce technical methodology). We treat audience members (users) and content as first-class objects in our CDP, so ultimately we felt that we had to build something from scratch. It provides a lot of interesting context where you’re able to look at what sort of individuals are engaging with various pieces of content, and also look at segments or groups of individuals and what content items are of most interest to them or what topics are trending.
JC: So Hum is a data repository and analytics platform that generally resources a broad spectrum of customer data, is that fair to say?
DS: Yes, but there’s one point that’s missing there: Typically, a classic CDP has elements of data unification, analytics, engagement, and activation. Hum brings all of these together. For us, it’s how Hum brings all those together and gives you the ability to personalize both emails and onsite content, and so we have widgets that pull from that data, index content, and deliver sort of personalized content in a more automatic fashion.
JC: So does Hum utilize artificial intelligence, machine learning, those types of things?
DS: Increasingly so, and as we get more data we will be continually building our model library. We have a few dimensions of growth and improvement there. One is that a lot of the ‘grey literature’ (like blog posts, white papers, and some marketing literature) doesn’t necessarily get the taxonomy treatment that the scholarly content does and is somewhat invisible to machines. Hum enriches the content. This is one place where artificial intelligence exists and we’re using transformer models to make that more discoverable.
JC: Does Hum work just with “grey literature”? What about the hard scholarly publishing content like journals, DOIs, etc. which many societies and associations publish? Is Hum more focused on the marketing and grey literature side of the house or do you have something for the journal publishing side of the house as well?
DS: We focus on all content. “Grey literature” and journals content really play different roles, so we really see a continuum of depth of content. At the shallowest end, you have a tweet or perhaps a notification email, and then you get deeper as you get through blogs, and then even deeper content pieces which may be a journal article or a white paper. You’re able to have a virtual event/webinar sort of thing through physical events and conferences that have deep signals about what people are interested in. From the Hum perspective, people interacting with content teaches you about their potential interests.
We know that in some societies 90% of their data is generated by the journals division, so ignoring that would be very foolish. You’d be getting shallower, incomplete data.
JC: Are you also pulling usage data from COUNTER?
DS: We actually go for raw data and track at the individual level and then work to the sort of statistical aggregate level. We want to be able to know as somebody moves around the entire society or association, as they are engaging with various things, what that truly teaches us about that individual or groups of individuals.
JC: Societies and associations have members, so they are knowns, I suppose, as opposed to unknowns.
DS: We have knowns and unknowns, though we call them identified and anonymous. The members are identified, but you have profiles on anonymous users as well. These anonymous users have more shallow interactions over time. We can personalize for anonymous users, but ultimately, we want to nurture relationships and hopefully over time they become known users and people you’re able to have contact with, even by email. That’s a very rich sort of interaction even if you’re not monetizing them in any way — the ability to contact and engage regularly and deliver value on your own schedule — not just push something at social media — as well as the ability to understand membership and follow what they are doing and engage with them, especially if you care about the amount of money they’re spending or what that might indicate.
It’s not just population and aggregate statistics, it’s all the way down to the user. The individual user is not the most useful way to engage with that data, it’s really the segment as the smallest atom or component that you want to engage with, providing the ability to create segments of different facets of data. Taking what people are topically interested in, their demographic data, other behavioral data, and having that reflected in real time or very near real time (basically within three minutes) ultimately means you can get direct insights as well as take another look at your membership structure. You might be able to look at the data and see what young people are engaging with and how we might reform a program or see what the “identified” non-members are engaging with and how we might be able to bring them into the membership fold. In terms of planning for events, knowing the most trending or searched-for topics provides the opportunity to ultimately create deeper and richer sort of event content. That real-time behavioral data of individuals, and not just what Google Analytics tells you, is quite rich.
JC: So Hum is aimed very much at strategy, marketing — in that top layer of things like product development. It is not in editorial, publishing, submissions and subscriptions — although it may obtain data from those systems or data from journal usage. Is this right? So who uses Hum? Who is the use case for?
DS: Yeah, I mean you’re right in terms of the adoption scenario. We’re not at the level of maturity where we’re going to have a marketing coordinator say, “We’re going to adopt Hum.” It’s just not the model at the moment, but the use cases are quite wide.
The leading use case is probably on the marketing side of things. They’ll likely have the broader mindset of looking at more data sources, thinking beyond the membership or experimenting with business models, but also (and it sounds kind of trivial) the ability to create segments to power some of their email marketing as opposed to blanket campaigns. You know, the “spray and pray” approach. We’ve seen some people go from basically 5-20% open rates to 80%+ because they’re targeting the right people. That’s something that everybody can use because basically everybody’s sending blanket emails and it becomes so much noise. It’s only gotten worse during the pandemic. So that’s a clear place we can add value.
We also unify things coming from the AMS (Association Management System) or CRM with financial information to look at some segments and match some of that financial information, not just how many members do we have or what their growth looks like, but the behavioral characteristics of that membership.
We certainly see events as an interesting feature within Hum that allow the user to take the segment, (say, people who attended an event) and create a content fingerprint on those people and then match for people who are reading the exact same thing, and generate phenomenal open rates. Engagement and yields improve for events because you’re targeting people who are similar in character, but not just based on demographic characteristics. You’re not just targeting academics or doctors of a certain type.
JC: Is Hum a service? How is the BI delivered? (e.g., dashboard, application, API, etc.?)
DS: Hum is a SaaS platform. There are others that are more of a service or a BI dashboard, but that’s not us. If you’re a Hum customer, you share the same core SaaS application, but you get your own instance. Your data is sequestered and protected. Internal users are provisioned on the Hum dashboard, which gives them the ability to get a level of insight to monitor the integration and health of the widgets as well as do things like create segments and campaigns.
Hum has an admin/superuser level type of role and a frontline person’s role, but you’re able to delegate access. There’s a lowest level where you’re able to see some things but not the deepest level personal user data. It’s more at the segment level or population level. Then there are widgets set up through initial onboarding and configuration. We’ve seen a real desire not to do a lot of the technical stuff so we’re bringing a simplified user experience to the table; we basically make it so that a relative technology novice marketing person can use Hum with very minimal training. Hum has kept the complexity on our side; we put a lot of time into that.
JC: What is the Hum sales approach?
DS: We’re finding that this is really a CMO, CTO, COO and CEO level conversation. If you want just slightly better email marketing orchestration, there are other solutions. If you want to double down on your membership, there are other solutions for that. But if you want to bring everything together in a unified approach — how you look at expanding your reach, building your audience, and improving the way you configure your digital systems so you create the most value — you really need a stakeholder who’s able to take that broader look and not be confined by the whole membership and physical event legacy mindset.
JC: Is Hum an all-or-nothing service or is it multi-layered?
DS: Our core, what we call a standard integration or quick start, consists of connecting with CRM/AMS, marketing/email system, and whatever the dot-org CMS uses. The dot-org website does a lot of aggregation and has a lot of content already. That gets a client started with data flowing into the platform. They will have profiles created, data enriched, and the ability to reach out to people. This is the basic Hum core implementation. At the same time, we work to find use cases where the client realizes value in the near term. For instance, some organizations will say we’ve had so much interest in these courses we’ve developed that are pandemic-related, and we want to help people find that content easier, or we are hosting a virtual event and want to create targeted activities to promote that. We build a plan and show results. So, there’s natural urgency internally to continue to grow beyond the core implementation.
JC: How does Hum charge for their service? Is the relationship based on charges by event? How does that all work?
DS: It is driven by the number of active audience profiles in Hum. We tried to get as close as we can to matching value. We want you to be able to build your audience, but we want you to build an engaged audience. Companies like HubSpot or Marketo charge if you have a record in the system, whether they’re active or not. For Hum, we charge on a monthly basis if you have an active individual who has interacted with your digital properties. It incentivizes us to help the society or association realize value. If they want to start small and have a small audience, that means it’s cheaper to get started. We grow with them, as opposed to saying this is a big enterprise deal. It makes it a bit easier for them to get started.
JC: Who are your competitors? You mentioned Hubspot and Marketo, are these your competition, or is there somebody in the STM world?
DS: Frankly, the biggest competitor is inertia. This is a newer category — there’s not another CDP in the association space.
When addressing the ‘jobs to be done’ level you can have a Salesforce or Adobe monolith, but these are incredibly expensive. You can do a similar sort of thing with HubSpot if you go all in on the HubSpot platform and try to put as much as you can into their ecosystem. That keeps the data with them and allows you to draw on it. It also means you have to have connectors with society-specific systems, which you would have to build yourself. It’s a challenge with bilateral data and is very difficult to connect with things like journal platforms. This is not their sweet spot.
JC: How does Hum deal with things like GDPR, CCPA and data privacy?
DS: We have a composable contracts architecture and are in the process of getting our SOC 2 certification, which allows us to continually stay abreast of privacy concerns. But we’re finding we don’t really have to adjust agreements too much because individual applications are already collecting that sort of data and are compliant with cookie notices and the ability to opt in or out. We validate that everything is in place, but the ability to opt out is managed more through the society or association, and we have the compliance process and machinery in place to maintain this.
JC: When did Hum launch?
DS: Quiet launch was in December of 2019, with the formal launch in early February 2020. We have been building out the early pilot customers and have now moved to our charter cohort. We’re being patient because we’re still figuring out the exact business model and go-to-market strategy. At the same time, we’re refining the machinery to make the solution for multiple stakeholders, but you know if you try to boil the ocean you’re going to fail. So we are being patient.
We’re finding the entry point and a tractable deal structure and deal flow, then we’ll fine-tune and optimize the delivery model. Our target for this year is 10 to 12 customers. We’re trying to stay modest; as we figure things out we’ll start to accelerate. There’s a ton of interest but we’re being selective about our charter cohort because some potential clients may not fit the ideal profile. We’re looking for customers who have budgets, use cases, and internal capacity.
We believe we can be viable and focused but also extensible and able to be integrated. We’re going to be powerful and valuable to the cohort for a CDP-plus orchestration, and the inside solution that we’re offering these cutomers will inevitably ask more of us in development, but it will benefit everyone.
JC: Does Hum see other markets outside the society and association market space?
DS: There are already people coming knocking from outside, but we’re committed to staying focused. Many of the CDP competitors are completely generic and say, “if you have data, we’re happy to handle it.” But they lack subject matter expertise.
We know there are close adjacencies, for instance with alumni associations, foundations, and other not-for-profits. People with a lot of content or audiences they want to engage have use cases that are similar: like an alumni association that wants to keep engaged with their alumni and keep them updated, or a foundation that wants to see how they’re influencing audiences on various topics. Certainly publishers, and of course we’ve had some conversations with publishers who are interested, but it’s the same sort of audience-building notion for customers with an institutional business model, so this can be a potential for society publishers selling into libraries, for instance.
I think there’s a broader opportunity for these high-quality organizations, institutions, and publishers to build out their audiences and you really can’t do that effectively if you don’t have the data side of things handled.
Hum is going to focus on the thousands of organizations we can serve within the association and society market, as well as the adjacent markets where there are opportunities and folks with similar problems to solve.
It seems Hum may have found an opportunity to help societies and associations hum a new tune — with the many disparate platforms they have been using for years, it may well be time. We will be watching as they grow. — JC