Tech stacks are large. The empirical stack data we recently shared from Zylo, a leading SaaS management platform, showed that even after a year of belt-tightening, the average SMB (500 employees or less) still has 162 SaaS apps. Mid-market companies (500 to 2,500 employees) have 245. And large enterprises have 650.

This isn’t particularly surprising any more, is it?

Oh, and by the way, those numbers don’t include:

(1) Any custom apps the company has built, including with low-code or no-code platforms.

(2) Any apps that are personally used by employees without being expensed. Mobile apps are the most common examples here: social media, learning, personal productivity, creative tools, etc.

(3) Any apps that freelancers or hired services firms — agencies, consultancies, or other outsourced providers — are using. You can say that’s not part of your tech stack, but in a lot of cases, inputs and outputs flow between their stack and yours, even if it’s through manual processes.

(4) A tremendous number of free or freemium websites that employees use that nobody thinks of as “apps”, even though they’re delivering data or functionality that help run your business. Do you consider Google search an app? Probably not. But it’s one of the largerst and most sophisticated pieces of software on the planet, and no doubt your employees rely on it every day.

All this is to say: software permeates everything. It’s hard to get a true count of all the apps in play for a company, because the further away an app is from central IT’s “managed” part of the tech stack, the less visibility we have.

This foggy frontier is where shadow IT lives. But the border of visible IT has been steadily shifting outward. It used to be that any app not directly managed by IT was considered shadow IT. Now, department-owned apps have moved from the shadows into the daylight and make up the largest percentage (48%) of officially managed apps in tech stacks. And they’re the majority (69%) of the spend.

In contrast, IT-owned apps account for just 17% of apps in stacks and 28% of the spend.

Fascinating, isn’t it? Department-level apps — formerly known as shadow IT — have now overtaken IT in total count and spend. More than a decade ago, a pioneering analyst at Gartner named Laura McLellan predicted that CMOs would spend more than CIOs on technology. A lot of people thought that prediction was nuts. Not me. She and I wrote a joint article for Harvard Business Review in 2014 explaining the dynamics driving that shift. I think we can safely say her predictive insight has been thoroughly validated.

Who is… The Shadow?

So what is shadow IT today? Zylo, whose empirical stack data I’m citing here, defines it as apps that are expensed by individual employees — perhaps for themselves, perhaps for their teams — that fall outside the official procurement and governance process.

It’s super interesting that such (redefined) shadow IT accounts for 35% of the number of apps in tech stacks — yet only 3% of the spend. It’s a lot of small apps.

The assumption is that such shadow IT is bad, like trans fats. The three main reasons:

1. It may be wasted spend, duplicative of existing IT-approved licenses.
2. It may be ungoverned by IT, presenting security and compliance risks.
3. It may be disconnected from the stack, creating data and process silos.

These are all legitimate concerns. However, the first one seems less egregious when we recognize that it’s only 3% of the spend. The second and third are harder to quantify, but that cuts both ways: the expected costs of those issues may be small or large, and may only be revealed over time or from a probabilistic “Black swan” event.

But we really should consider the other side of the equation too. Why do people buy such shadow IT? Is it just to rebel against the Empire? With a SaaS subscription? Not exactly the stuff of Jedi legend.

Weighing the upsides of Shadow IT

Individuals and teams adopt SaaS products outside of their company’s official tech stacks for one primary reason: to enable them to better perform in their job.

It may be that there isn’t an app in the official tech stack that does what they need it to do. Or perhaps there is, but the way that product works is undesirable on some dimension: it’s too hard to use, it doesn’t have the right features, the outputs it delivers are subpar, it takes too long, it costs too much, they haven’t been sufficiently trained or enabled, etc.

I don’t have quantitative data to prove it, but everything in my experience and everything I’ve ever heard from other people who go outside their official stack to use other apps is that the benefits in creativity, innovation, and productivity are meaningful to them. It helps them Get “Stuff” Done. It pushs the frontier of the firm’s processes and capabilities. It helps prevent stagnation in talent and technology.

Now, that doesn’t eliminate the downsides. But it does present a non-trivial trade-off. There’s reward as well as risk — for individuals, but also for the company, which is ultimately the sum of its individuals and teams and their impact — balancing on the Scales of Shadow.

In fact, one of the reasons that such Shadow IT is so popular is because tons of SaaS companies have now built their products and go-to-market engines around the proposition of giving free, freemium, or low-cost/high-return value to individuals and teams. They prove their worth in the trenches, and then scale up to become officially adopted across the enterprise. Such “bottoms up” product-led growth (PLG) strategies have proven highly effective.

Yes, it’s a strategy that benefits those PLG apps. But they only achieve that benefit by delivering value. Consider the top factors that PLG companies focus on, for both classic seat-based licensing but also with usage-based pricing:

Build for openness and build to meet users where they work: they need to easily plug into existing ecosystems and workflows. Build for the end user: make users happy and successful. Deliver instant product value. Monetize after you deliver value.

You can see the appeal. Particularly because, in the eyes of many users, big legacy-ish enterprise-wide platforms haven’t expressed as much concern for their happiness and personal success. Now, that’s changing. But frankly, it’s changing because these PLG apps have created competitive pressure in the market, raising the bar for department-wide and enterprise-wide solutions.

One other major benefit that I believe comes from these bottoms-up PLG apps: better utilization. People use the apps they want to use. They resist using ones they don’t like. And the advantage of individual users and teams paying for their own licenses, inherently out of their own budgets, is that the buyers and the users are tightly coupled if not the exact same humans.

Those big, enterprise-wide deals for sweeping seat purchases? I suspect you’re far more likely to have unused seats lumped into that pile.

Taking this even further, PLG products that are leaning into usage-based pricing are driving the ultimate alignment between expense and utilization. You only pay for what you use, and you only use what gives you value.

Thank you, Chuck Norris Shadow IT apps, for pushing these usage-based models into the competitive dynamics of the market.

Eliminate Shadow IT by redefining it

Still, the downsides remain. And compliance, security, and siloization are heavy stones on the other side of the scales. But are there ways we can mitigate those downsides without losing the upsides?

I believe it’s possible.

One step is to de-couple technical approval and financial approval for apps used by individuals and teams. We’ve already done this at the departmental level. Marketing is responsible for covering the cost of the platforms they officially use, but those platforms increasingly go through an IT review for security and other compliance requirements.

Push that model further out to the edge of the org. Any app that an individual or team wants to use should undergo a security and compliance review. But the choice to pay for that app is up to the individual or team — and their ability to secure budget and justify its use. Don’t get me wrong, there should be pressure to justify the expense. But for small expenses, the pressure should be closer to individual and team, not in a distant department that likely has no direct stake in the use case.

But does that create more burden for reviewing a larger set of apps for security and compliance? Yes. But this doesn’t have to be one extreme or another. It can be a continuum, where there is a larger menu of apps that become approved. It’s not every app on the planet. But it’s not limited to just one in a category. And hey, maybe teams should “pay” to submit a new app to that review process.

I actually think this is a fantastic opportunity for SaaS management platforms, such as Zylo, to provide more vetting-as-a-service for popular apps. It could accelerate or optimize the review process for IT teams.

Other ideas might include a “sandbox” structure for new apps on the edge, that let users experiment with free or freemium apps in a limited fashion to determine if it’s even worth nominating them for review.

Users are experimenting with apps this way now. It’s just in the shadows because most companies haven’t created a good framework to let them do that experimentation in a way that’s visible to IT.

I’ll wrap this post up here, as a comprehensive write-up of all the possible ways to evolve the management of the apps-formerly-known-as-shadow-IT would like be a book. (Hmm.) But dismissing the upsides or ignoring the situation in the trenches is not, in my opinion, a sustainable strategy for companies competing in a rapidly evolving digital world.

We kill shadow IT for good by making all software visible.

And I didn’t even get to the invisible tech stack the lives beyond the boundaries of the firm, with all of one’s software-enabled services providers. A topic for another day.

The post Visible and invisible tech stacks, and the upsides and downsides of “shadow IT” in martech and beyond appeared first on Chief Marketing Technologist.

First, one more reminder: please take our Martech Composability Survey this week. When you see the questions, I think you’ll agree that having a statistically significant dataset for a “no BS” view of this topic would be super valuable for the whole martech community. We’ll share the full results publicly. But we need your participation. Please and thank you!

I’ve been adovcating the benefits of aggregation platforms in martech stacks for a couple of years now. These are platforms that provide cohesion to a diverse set of apps or data sources in your stack. You get both a unified, integrated system — or more accurately, an internal ecosystem — that easily adapts to apps or datasets being added or removed from your stack.

Such aggregation can happen at different layers: data, workflow, user experience, or governance. Some platforms aggregate at multiple layers. But the clearest — and arguably most flexible — examples are those that specialize in aggregating at a single layer.

The quintessential example is a cloud data warehouse, which aggregates at the data layer. You can have many different data sources contributing data to the warehouse. And you can have many different apps pulling data from the warehouse. No matter how many sources and apps you have, the warehouse serves as one common hub for all of them.

A key characteristic that distinguishes an aggregation platform from other kinds of software: the more things you connect to it, the more value you get from it. An aggregtion platform delivers “network effects” within your stack. (This is in contrast to point-to-point integrations in a stack that become more complicated and fragile as you add more things to your stack.)

Cloud data warehouses have this characteristic: the more things you have pushing and pulling data to and from them, the more value you can derive because they make such a wide span of data accessible to so many applications. (With the caveat that you need to actively manage this so it doesn’t turn into a data swamp.)

This isn’t hypothetical. Cloud data warehouse use is on the rise in marketing. Winterberry Group recently published an excellent research paper, Demystifying the Data Layer: The Transformation of Marketing Data Infrastructure, that sheds light on this shift.

Across the US, UK, and Europe, marketers are increasing — or significantly increasing — investment in their data infrastructure:

Hey, as the saying goes, “AI strategy is data strategy.” And everybody needs an AI strategy now. So having 93% of US companies, 95% of UK companies, and 99% (!!) of European companies beefing up their data infrastructure to enable AI-powered operations isn’t that surprising.

What may be surprising to some — at least to ardent advocates of austere consolidation — is the graph at the top of this article that shows most companies (79%) are not using a “singlular cloud” solution for managing and leveraging this data. Most are either using a best-in-class stack (35%) or a hybrid stack (44%).

Hybrid architectures “[blend] the strengths of centralized platforms with specialized best-in-class solutions.” This sounds to me like an aggregation platform: one central platform, connected to multiple specialized apps. That 44% of marketers are taking this route — more than doubling with those who are transitioning to this model — is a testament to its suitability to our currrent tech and data environment.

There was a related finding in the report that caught my eye:

Asking marketers which solutions they use to store, manage, and analyze data, you can see from the percentages that there are obviously multiple solutions working in parallel most people’s stacks.

Almost everyone (83%) uses a CRM. Another 59% use a customer data platform (CDP). And another 57% use a data warehouse or data lake. Even though a case could be made for each of these to be the “system of record” or “source of truth”, you’ve obviously got a lot of stacks out there with two or three of these in them.

While some architectural purists may object, I think this can be a pragmatic design pattern. If these platforms are sharing the same data with each other — which they should, at least to an extent — real-world operations can benefit from having that data locally available in the context of each particular application. It can speed up performance, and it can adapt the “shape” of the data to the functional model of that application.

A CRM empowering salespeople has different uses than a CDP orchestrating a marketing campaign or a data warehouse that’s being mined by an analyst. As long as the common data is synchronized across them, such a “federated” architecture can let each optimize interactions with that data for their users and use cases.

A cloud data warehouse as an aggregation platform can be particularly effective here, serving as the Grand Central station of data integration. Data that is relevant to multiple applications is unified in the warehouse. But the subset of that data used by individual apps may — emphasis on the word may — be federated into other apps when that delivers the greatest operational efficiency.

But what about silos? Surely they’re universally bad?

Well, don’t call me Shirley. Here’s the thing: data is expensive to move, expensive to store, expensive to compute on, and — in the cost of human time and talent — expensive to manage. Granted, with a data warehouse as a data aggregation platform, it is much less expensive than it was before. But the costs are not zero.

So you only want to send data to the warehouse if there are legitimate cases for how that data will be used. There’s a ton of operational data buried within the mechanics of how different apps work that doesn’t have relevance elsewhere in your stack.

What’s important is the option to share data. Any data that an app collects or generates should be able to be piped into a warehouse (or other kind of data aggregation platform), so that you can choose to leverage it elsewhere in your stack when a use case arises. It’s a silo with many doors.

Unified, federated, and even siloed data all have roles in martech stacks.

Last reminder: please take the Martech Composability Survey! Participation in the occasional study like this is my one ask of readers. You’ll be able to see the full results, which I think you and the rest of our community will find interesting and useful. Thank you!

The post Can martech data be unified, federated, and siloed all at the same time? Yes, and each serves a purpose appeared first on Chief Marketing Technologist.

First, a quick ask: please take this 5-minute survey on martech composability. We’ll share the full results back with everyone. I bet it will be very interesting. Thank you! Okay, back to today’s post…

I prefer empircal data over gut-feel prognostications. For 12 years, I’ve been told that SaaS in general — and martech in particular — is going to dramatically consolidate. People can be quite vehement in such predictions. Yet year over year over year, the data consistently fails to show that outcome.

According to the 2024 SaaS Management Index Report just released by Zylo, a leading SaaS management platform that manages over 30 million SaaS licenses for companies of all sizes, the average SaaS portfolio shrank from 291 apps to 269, just a hair under 8%.

For context, 2023 was a brutal year for SaaS companies, as customers sought as many opportunities as possible to reduce spend by cancelling unnecessary app subscriptions. And indeed, average SaaS spend shrank closer to 11% over this past year.

Yet the average stack size for SMBs with 500 or fewer employees is still 162 apps. For mid-market companies with 501-2,500 employees, it’s 245 apps. Yes, these are down by 6% and 4% respectively, year-over-year. But that’s not exactly the scorched earth scenario that the prophets of consolidation were promising we’d see at this point.

I’m sure they’ll say, “Just wait until the end of this year!” Maybe they’ll be right eventually.

Don’t get me wrong. I’m all for companies being cost-conscious and proactively managing their stacks to maximize ROI. I’ve also got no problem with market forces that tend to drive a small number of competitors in a category to pull away from the crowd.

But the reality is that we’re in a time of continuous technological innovation. If anything, the pace of change in tech has been accelerating — and is accelerating even faster now in the current wave of all things AI. In an environment like this, new opportunities for startups abound. And the pressure for companies to adopt new technologies to keep pace with their competition are real.

You can see evidence of this by the number of new SaaS apps that keep getting introduced into a company’s SaaS portfolio every month:

This continuous evolution is a significant counterweight to the forces of consolidation.

It’s worth noting that, while average stack sizes have been shrinking these past few years — albeit by relatively modest amounts — the number of distinct SaaS apps in the market has continued to grow. Of course, you’ve witnessed the upward trajectory of the martech landscape for the past decade. And a couple of years ago, the review site G2 shared they had over 100,000 SaaS products in their database.

But as another source to triangulate app diversity, the above graph from Zylo shows the number of different apps that they keep track of. It’s only been growing, now up to more than 20,000 apps. Yes, growth has slowed. But it’s still been growing, even as average stack sizes have started to shrink.

There may be (slightly) fewer apps in stacks these days. But if you do the math, it’s an increasingly diverse collection of apps across those stacks.

Shadow IT, a shadow of what it once was?

Once upon a time, “shadow IT” was the pejorative phrase given to any software app that wasn’t owned by the IT department.

We’ve come a long way. Now, the majority of software — both by total number (48%) and spend (69%) — is owned by business teams. Marketing is running most of its own martech apps (although the amount of IT-owned, company-wide data and aggregation infrastructure used by marketing is growing).

These business-owned SaaS apps are no longer considered shadow IT because, although business teams are footing the bill and running these apps, the IT department still has full visibility and can insist that they comply with their governance and security requirements. In many ways, this is the best-of-both-worlds scenario we long sought in martech. It balances central control and distributed empowerment.

But shadow IT still exists. Its definition has narrowed though, and it now mostly refers to SaaS apps that individual employees or teams expense on their credit cards, outside the watchful eye of IT governance.

As you might expect, most of these shadow IT apps are small enough — or at least inexpensive enough — to fly under the radar as a credit card expense. According to Zylo, only 3% of the total spend on SaaS falls into this category. What’s fascinating, however, is how many of these small apps there are. While they’re only 3% of the spend, they’re 35% of the number of apps in people’s stacks.

Yikes! More than 1/3 of a company’s SaaS apps are ungoverned shadow IT! This sounds like a four-alarm crisis.

Maybe it is. But when you look at the most popular shadow IT apps, most of them seem relatively harmless. Udemy, LinkedIn, CliftonStrengths — as an aside, I’m a huge fan of StrengthsFinder, and highly recommend it — Kudoboard, Coursera, MasterClass, etc.

Yes, there would be benefits to having all of these subscriptions be official “daylight IT” instead of shadow IT. Mostly to keep expenses down by getting rid of unused licenses or negotiating corporate-wide discounts. (Although I’ll note those two levers are often trade-offs of each other.)

But we’re only talking 3% of total SaaS spend. So even if you cut those costs in half, it would be a mere 1.5% savings. Hey, 1.5% here, 1.5% there, next thing you know, we’re talking real money. But it’s no white whale.

A couple of these are more problematic. Using SurveyMonkey to collect customer data? That should be governed. Using Grammarly with confidential documents and emails? That should be governed. Using ChatGPT and the OpenAI API? Hell, yeah, that should be governed!

But this seems like a gap that can be closed by putting in streamlined mechanisms to officially purchase approved “personal productivity and development” apps (which is how I’ll broadly generalize most of that list).

In fact, the gap is closing. In 2022, Zylo reported that 20% of employees were expensing SaaS in this shadowy way. By 2023 the number had dropped to 15%. And here in 2024 it’s down to 7%.

In summary:

1. Tech stacks are still large, although healthily trimmed.
2. The universe of diverse apps in tech stacks is still expanding.
3. Business teams definitively own most of their apps, albeit governed by IT.
4. Shadow IT is fading into the shadows, mostly because we redefined it.

Lots more great stats in the Zylo report. Highly recommend picking up a copy.

One more reminder: please, please take our martech composability survey. When you see the questions, I think you’ll be as curious as I am to see what the results will reveal. But we need you to participate in order to have statistical significance. It’s my humble ask in exchange for the free content I publish. Thank you!

The post Well, SaaS tech stacks shrank from 2023 to 2024… but only by 8%. You were expecting more? appeared first on Chief Marketing Technologist.

It’s that time of the year. The Northern hemisphere looks forward to spring. The Southern hemisphere looks forward to fall. And everybody in martech and marketing operations looks forward to the annual Stackie Awards.

This will be our 10th edition of The Stackies: Marketing Tech Stack Awards. Wow, a decade of hundreds of martech stack slides: 2015, 2016, 2017, 2018, 2019, 2020, 2021, 2022, and 2023. We’d love for you to join this martech community tradition with us this year.

To enter the 2024 Stackie Awards, all you need to do is create a single slide (16:9 aspect ratio) that illustrates your marketing stack — the collection of martech software that you use. Your slide can be as simple or as elaborate as you like. The best slides successfully communicate a clear concept about how your stack is organized and how it fits into your marketing operations.

Then just upload your slide by April 26, 2024. There is no fee to enter. On the contrary, we’ll make a $100 donation to UNICEF on behalf of every legitimate stack entered, up to a total of $10,000. We’ll the announce the winners at an online #MartechDay event on May 7, 2024, and release a deck of all the slides that were submitted.

“So wait, your headline says everybody wins?”

Well, only five entries will be selected as winning stacks: those that we subjectively believe contributed the most to the community. But everybody wins because we all get to learn from the complete collection of stack illustrations: what do the stacks look like at different companies and how do they think of them? Martech vendors win by getting recognized in the stack slides shared by their customers. And the donations made to UNICEF on behalf of these entries help in some small way with things more important than any stack.

Airstream, Autodesk, Cisco, IBM, Microsoft, Nationwide, Paychex, Philips, Sargento, SAS, United Healthcare, Verizon, Washington Post, and Whirlpool are some of the well-known brands that have shared their stacks in The Stackies. For example, here’s the entry from Juniper Networks in 2021:

Or, for a B2C example, this entry from Casey’s in 2023:

Or, one more example, this one from Rabobank in 2022:

You can have as much fun with your stack illustration as you like. But all three of these examples capture a sense of the journey — either for marketers or customers — in which these martech tools are used. I like Juniper’s and Casey’s, because they also break down the different categories of martech products they use and how they map to that customer journey or marketing production line.

But you don’t have to follow what they did. Any way you want to represent your stack, we’d be delighted to see it.

That said, every year I like to make a couple of suggestions for entries that I hope a few people will consider. This year, I’d love to see two things:

First, where is generative AI being used in your stack? For content creation? Data analysis? Customer service? Streamlined workflows? Idea generation? Are you using specific gen AI tools, or gen AI features built into larger martech platforms? If you can help us see the real-world use cases of AI, that would be a cool contribution.

Second, I’ve been writing about aggregation theory in martech stacks for the past year or so, how certain platforms unify data, workflows, user experiences, or governance across your company’s tech stack (i.e., not just limited to marketing). The quintessential example of this is a cloud data warehouse or data lake, such as BigQuery, Redshift, Snowflake, or Databricks, that aggregates data from many sources and makes it available to many apps to use. I’d love to see a few Stackies this year that illustrate real-world examples of this.

Are you using a cloud data warehouse as part of your martech stack. Please include it!

But you don’t have to constrain yourself with either of those ideas. More than anything, the Stackies are about educating each other about the different ways martech is implemented across industries and geographies. Whatever you want to teach with your stack slide, we’re ready and eager to learn. And we greatly appreciate the contribution.

Just be sure to send in your slide here by April 26, 2024. Time to get stackin’!

The post The 2024 Stackie Awards are now open for entries — share a slide of your martech stack, everybody wins appeared first on Chief Marketing Technologist.

Dear marketing readers: hang in with me here. I have a point. Promise.

I started programming as a kid, writing multiplayer games for dial-up bulletin board systems (BBSs) — a precursor to the web and social media as we know it today. It was the late 80’s, early 90’s, and I mostly wrote in a language called C, with some occasional high-performance components written in 8086 Assembly language.

For those of you who aren’t software developers — or for you younger developers who grew up thinking Java was a low-level language — this is what assembly code looks like:

You are literally spelling out individual instructions to the CPU, moving bytes from memory to registers, executing simple mathematical operations on them, and creating program flow with primitive logic gates of AND, OR, XOR.

While acknowledging the cosmic power of that low-level foundation upon which all software has been built, let me just say that writing Assembly sucked.

As a game designer, I had storylines in mind, new worlds I wanted to create, interactions between players I wanted to enable. Translating that down into code, especially the parts in cryptic Assembly instructions, was like diving from the heights of Mount Everest down to the depths of the Mariana Trench. One’s head could explode.

A few things I recall about that experience:

First: it was tremendously time-consuming. A set of game mechanics that I could explain to you in 30 seconds — one of my games let players mix different potions to create new, more complex ones, laddering up to ever more powerful magical elixirs — would take 30 days to develop. Yes, coding enabled these things to be created. But the high cost of coding was also a barrier to materializing many other ideas.

Second: it was very error-prone. Even great programmers end up with “bugs” in their software. I was, at best, an average one. So I had a lot bugs in my code. Discovering them was hard. In more complex programs, like a multiplayer game, even being able to test all the possible ways in which something might go wrong was an exponential task. Fixing them could be even harder.

Relatively few of my bugs were flaws in the game design. Almost all of them were accidental mistakes made in the coding process, translating a perfectly valid conceptual flow into thousands of lines of code, where a simple typo could crash the whole thing. Code is fragile, my friends.

By the way, there is no Lake Wobegon. Half of all programmers are below average. And the below average ones can write such bug-infested code that if the program were your house, your friendly Orkin professional would recommend gasoline and a match.

Third: it was easy to lose the plot. Head down in the code, trying to implement some crazy recursive function, juggling pointers to a mess of different data structures, I was no longer thinking about the game or its players. I was consumed by wild pointers, dangling pointers, memory leakage, the impact of modifying the base pointer, and address access out of scope errors.

However, despite all of that resulting in mediocre code, the games themselves became quite popular for their time, played by over a million users on BBSs around the world. Their success was in spite of my deficencies as a software developer.

The moral of this nostalgic walk down memory lane: code and coding have always been an imperfect means to an end.

Back to the Future to Really MOVE CX, DX

Fast-forward a few decades to today — and the proliferation of no-code tools.

There continues to be debate about the wisdom or folly of letting non-IT employees build workflows, apps, agents, or composed customer experiences with no-code platforms. Whenever I publish an article about growing no-code use in marketing and martech — such as last week’s Every marketer a data analyst and an engineer… delusion or destiny? — I get an earful from people who insist that non-developers don’t have the skills to build anything good.

With all due respect, they’re wrong.

First of all, code is not inherently better than no-code. The history of programming has been a steady march to higher levels of abstraction. We started with punch cards. I came along with Assembly language, which I’ll reiterate was as much fun to work in as a salt mine. Then higher-level languages like C, C++, and Java. And then higher (and safer) languages yet, such as Python, with over 350,000 packages you can use to “compose” amazing programs in a fraction of the time it used to take back in the salt mines.

No-code is the next step in that progression of abstraction, where more than ever, it’s about understanding what you want to build and your ability to describe it — visually or with a natural language interface (thank you, gen AI). Not having to concern yourself with things like uninitialized pointers and segment faults is a pretty big improvement to a builder’s quality of life. Most non-code tools have excellent guardrails that raw code doesn’t.

“But even in no-code, it’s about logic and programmatic thinking,” critics will respond. And they’re right — about that first part. But then they finish that sentence by adding, “And only software developers and IT professionals have those skills.“

Sorry, but that’s incorrect in both directions.

On one hand — reflecting back on my sloppy code slinging days — being a software developer doesn’t inherently make you great at logic or programmatic thinking. No offense, but there are a lot of below average developers out there whose grasp of logic and programmatic thinking is not great. Just because you bring in a “software developer” to build something, doesn’t mean it’s going to work as intended (or at all).

On the other, not being a software developer doesn’t mean logic and programmatic thinking inherently elude you. Finance. Operations. Legal. Many business professionals have great skills with logic and programmatic thinking, even if a screen full of Javascript would be as indecipherable as sanskrit to them. (Quid pro quo: have your average Python programmer try to grok the mechanics of an Excel budgeting spreadsheet from a senior financial analyst at a public company. Then have them explain it to me.)

Marketing ops, sales ops, rev ops — these disciplines thrive on logic and programmatic thinking. Customer journey maps, orchestrated by marketing automation tools, handed off to algorithmic sales sequences driven by conditional triggers and actions… these are savvy software “programs” that are crafted by experts in their domain.

“Experts in their domain” is key. A good marketing operations professional can build excellent programmatic workflows with a no-code automation tool first and foremost because they deeply understand the context of what they’re building. It’s not just that they are proficient at designing a logical program flow. They know what those triggers and actions actually mean. They know what outcome they’re trying achieve. They know what can go wrong with the business process.

In an age when no-code tools — and even gen AI “copilots” for building with code — are making it easier and easier to do the technical implementation work of creating an app or automation, the real value is knowing what app or automation to create. What does it need to do? Why? How (from a process perspective, not at a raw code technical level)?

Domain expertise is the linchpin here.

No-code tools empower people with deep domain expertise — and perfectly fine logic and programmatic thinking skills, whose only “flaw” is that they didn’t learn to code in pythonic Python — to quickly and efficiently turn their knowledge into better digital operations and experiences.

Can no-code tools also empower people who have no idea what they’re doing to make some really crappy workflows and automations? Sadly, yes. But it’s not that different from the chaos a software developer who doesn’t know what they are doing can sow.

Don’t conflate knowing how to code with knowing what you’re doing.

The greatest benefit of the no-code era is the disentanglement of those two things.

And now, a word helpful data from our sponsor…

As the costs rise for publishing this blog and newsletter, I’m experimenting with ways to accept sponsorships that are non-interruptive and hopefully useful to you, dear reader. One idea that I’m trying today is highlighting a report from a sponsor that I believe has data and insights that you might find valuable. I’ll only share it if I genuinely think it’s good, but I am getting paid for including it. You can help support my writing by checking it out.

Our inaugual sponsor is MoEngage, who just released their 2024 State of Cross-Channel Marketing Report. A study of 700+ B2C marketers, it includes useful data about marketing channels and priorities, challenges and solutions. For instance, this distribution of the channels B2C marketers are currently using:

What are the top challenges marketers face in adopting new technologies for cross-channel marketing? Well, the #2 answer is — surprise, surprise — “integration issues with existing technologies” (27.6%). The other Top 5 challenges? Well, you’ll have to get a copy of their report to find out.

The post Developer vs. non-developer is the wrong divide; what matters with no-code is knowing what you’re doing appeared first on Chief Marketing Technologist.

“Everyone within Publicis will become a data analyst, an engineer, an intelligence partner, with all the information they need at their fingertips to supercharge client growth.”

Publicis Groupe made that bold statement last week in a press release and presentation celebrating their performance from last year “after shifting from a holding company to a platform” and charting their course for their future in the age of AI.

Now, if you’ve raised a skeptical eyebrow to that claim, you’re in good company with most of the people I shared that with on LinkedIn and the artist-formerly-known-as-Twitter. But, hey, if the world’s third largest agency holding platform company can’t paint an audacious picture, you’d worry about their core competency.

But here’s the thing: directionally, I think they’re right.

Empirical evidence supporting that vision came with the release of Workato’s 2024 Work Automation & AI Index report earlier this week. Workato is a leading enterprise automation company that provides a low-code/no-code (LCNC) platform for automating processes and workflows across your tech stack. (Disclosure: I am an advisor to them and a co-author of their CEO’s new book, The New Automation Mindset.)

Using anonymized data sampled over 36 months from 1,055 customers on their platform, they captured the ground truth of what companies are automating — looking at over 82,000 automations — but arguably more remarkably, who is building those automations.

The big reveal in my opinion: 44% of all automated processes are built outside of IT.

Business ops teams — such as marketing ops, sales ops, revops — build 27% of all these automations. Non-IT project managers, product managers, app adminstrators (looking at you, CRM admins), build another 10%. And then another 7% of “other” builders, which I’ll guess as intrapreneurial power users. (I have a couple of them on my partnerships team at HubSpot, where we also use Workato, and it’s amazing what they can do.)

It’s also notable that these non-IT builders — who are experts in their domain, but at least organizationally are not experts in IT — aren’t just automating simple processes in their function. They’re tackling complex automations too.

“Scores of 1-3 are simple, point-to-point integrations with 1-4 steps. They contain no logic, and interconnect simple SaaS applications. Complex processes (4-6) involve conditional rules, logic, looping, data transformations, and cross-reference data. Sometimes they involve batch processing. Highly complex processes involve a combination of SaaS, on-premise, ERP, and enterprise applications. They often contain 30+ steps, with conditional rules, advanced transformation, humans in the loop, and more.”

Now, I know there are a few cynics skeptics out there who think non-IT people using LCNC tools is a recipe for disaster. We could have a vigorous debate over the relative value of domain expertise vs. IT expertise when it comes to implementing digital operations within a business function. But instead of rhetoric, let’s look at data:

If business teams implementing their own automations was going to create a train wreck, you’d expect to see that show up in the first year, when those business builders are least experienced. Or at least the first two or three years, right? Things would go pear-shaped, and the CIO would drop the hammer, “Look at the mess these amateurs created! We’re taking over.”

But that’s not what the data shows. Quite the opposite. On average, after having 31% of automations built by the business in Year 1, the organization leans into that model even further and has 41% built by the business in Year 2. By Year 3 that number is up to 48%.

Keep in mind, this is not an IT versus business situation, with the two different sides using competing tools. They’re all unified on one common platform — a quintessential example of a workflow-layer aggregation platform — that is almost always owned by IT. Empowering these business teams is part of IT’s strategy.

As Workato notes, “IT is evolving into a player-coach role: 56% of automations are still built by IT personas, but IT is also being tasked with governance and guidance for the 44% handled by business teams.”

Can we have both empowered domain expertise and good IT governance harmonized? The evidence here suggests yes.

The really exciting thing about this? This multi-year data from Workato only includes the effects of generative AI at the tail of the time period analyzed. Last year, they released a natural language copilot to empower builders even further — as well as a governance framework and academy course to teach it. A year from now, we’ll see the emprical impact, but I bet it will show an acceleration both in total automations as well as the percentage of business-led automations.

Speaking of generative AI, the Workato report also includes interesting data about where organizations are incorporating generative AI into their automations:

48% of the use cases are in revenue operations, with an additional 12% in customer support and operations. The most common of those use cases is with conversation intelligence — summaries, sentiments, next steps. These have been some of the most time-consuming and error-prone manual facets of managing the customer journey. Intelligent automation here is clearly a huge win for both companies and their customers.

So let’s circle back to where we began, with Publicis’ bold aspiration.

The facts are that businesses are rapidly increasing the scale and complexity of their digital operations. They’re empowering more and more people outside of IT to shape and adapt pieces of those digital operations that are closest to their work. The Workato report didn’t cover the democratization of analytics directly, but a significant number of their use cases involve intelligently distributing data to enable more context-specific analytics. That might not make every employee a data analyst, but it’s sure going to enable a lot more people to productively use self-service analytics in their jobs.

Admittedly, Martec’s Law still holds: there’s going to be a ton of difficult organizational change required to harness the innovation this technology now makes possible. But that’s why this is such an amazing time to be working in martech and marketing operations.

The post Every marketer a data analyst and an engineer… delusion or destiny? appeared first on Chief Marketing Technologist.

Welcome to 2024!

I expect this will be a transformative year in martech. The Martech for 2024 report we published last month (video presentation) covered several of the major trends underway, such as aggregation and composability in stacks — especially in the data layer with cloud data warehouses, lakehouses, and lakes.

The common theme of those trends is a breaking down of the boundaries between different martech apps and platforms. The silo’s days are numbered. (Can I get a hallelujiah?)

It’s a profound change. But there’s an even more profound change rolling in.

Growing complexity as a constant for two decades

For the 16 years that I’ve been covering martech on this blog, martech has been strictly increasing in complexity. And before you blame that on the explosion of the marketing technology landscape — now over 13,000 products! — recognize that all those martech tools were built in reaction to the growth in complexity of marketing itself.

When customers embraced search, marketers had to embrace SEO to persuade both people and search algorithms. When customers embraced mobile, marketers had to embrace apps, responsive web design, SMS, push notifications, WhatsApp, geofencing, etc. When customers embraced social media — Facebook, Twitter/X, LinkedIn, Instagram, TikTok — marketers had to embrace social content, social advertising, social customer service, social reputation management, influencer marketing, etc. When customers embraced online video, marketers had to embrace video content and video advertising, YouTube channels, Instagram reels, etc.

And that’s just the customer-facing side of martech. New marketing approaches such as account-based marketing (ABM) and product-led growth (PLG), intent-based and digital behaviorial marketing, customer journey mapping, unified RevOps at the intersection of sales, marketing, and customer success, ecosystem marketing, and so on.

All these inventions and innovations have made significant contributions to the discipline of marketing. But they also massively expanded the diversity and scope of marketing. And all these new and different moving parts, all interacting with each other, expoentially grew the complexity of marketing.

Martech systems — all the different software in your martech stack — grew in complexity as a function of this inherent complexity in modern marketing. And even if you chose a bundled suite of apps over a best-of-breed combination of tools, you couldn’t escape the constant flood of new features and functionality to serve the ever-widening span of your requirements and responsibilities.

It might seem like a given that the more complex martech systems became, the more complex the user interface to those systems became too. Whether you were switching among an infinite row of browser tabs for different martech apps or hunting for specific options within a mega marketing cloud, the cognitive load in just figuring out how to do any one particular thing — out of all the things that were now possible — became a significant mental tax.

This has been a major factor in plummeting martech utilization rates.

As marketing became more complex, martech systems became more complex, and the UX to those systems became more complex. It was effectively a tautology:

Marketing Complexity = System Complexity = UX Complexity

Granted, there were differences at the margin. Martech products increasingly invested in better UX as a competitive advantage. Yet even the best were swimming against the tide.

But could a new innovation reverse UX complexity?

Generative AI user interfaces turn the tide

Generative AI is the first major innovation in the history of marketing technology with the power to fundamentally reduce UX complexity.

One of the first examples of this early last year was the release of ChatSpot, a generative AI “chat UX” interface for HubSpot built by Dharmesh Shah. (Disclosure: I work at HubSpot as their VP of platform ecosystem.) Through a simple ChatGPT-like text prompt, you could ask ChatSpot to do something for you in HubSpot — create a contact, write a personalized email, run a report, etc. — and, voilà, it would just do it for you.

You didn’t need to navigate any menus or screenfuls of buttons, dropdowns, checkboxes. Dharmesh eloquently summed up this new mode of software interaction as a shift from point-and-click to describe-and-done.

Over the past nine months, ChatSpot rapidly expanded its capabilities to support sales prospecting, SEO consulting, rich content creation, business analysis, deal optimization, etc. Yet as all these powerful features were enabled, the user interface for ChatSpot remained a simple text prompt.

Hundreds of other martech companies followed suit, adding natural language “chat UX” interfaces to their products. Many of these new AI-powered UIs are still in their beginning stages, just scratching the surface of what will be possible in the year ahead. But as they make more existing product functionality more accessible through this UI — and then add even more new functionality — they actually reduce their UX complexity.

It’s almost counterintuitive. The more capabilities you add, the simpler the UX becomes?

The underlying complexity of what their products can do remains. In fact, as we’ll discuss in a minute, the complexity of these martech sytems is actually likely to increase further. But the martech UX on top of those systems has become much simpler. And it promises to get even simpler yet.

How do you get simpler than a natural language UI? By enabling that UI to serve outcomes more than tasks.

The first generation of chat UX interfaces required you to be fairly specific in each task you wanted to do. For instance, “give me a report of sales by country”. The newer generations – and they’re iterating rapidly — will be able to understand higher-level requests, such as, “which countries are lagging in forecasted performance and how can we boost close rates on their most important deals?”

This next generation of martech UX will be legit AI agents, capable of pursuing higher-level goals. They’ll be able to plan multiple steps, adjust their plans based on feedback from the actions they take, leverage any data or API functionality they have permission to access, and even know when to consult a “human in the loop” for clarifications or confirmations.

AI agents will serve as uber-aggregators in the martech stack, providing an aggregated user experience through their chat UX, while also aggregating process functionality and tapping into aggregated data warehouses behind the scenes to achieve their goals. Governance will be embedded into or layered on top of these agents.

For marketers, the distinction between process, control, and experience will be blended behind the interface to these agents.

Removing a big constraint on system complexity

Ironically, while AI agents reduce martech UX complexity, they will almost certainly increase underlying martech system complexity.

In many ways, the limits of how much martech UX complexity humans could deal with had become a significant constraint on how much additional martech system complexity could be implemented effectively. There was only so much data, so much functionality, and so many combinations of those elements chained together that even an expert marketing ops pro could keep track of in their head.

Humans have many wonderful qualities. Comprehending exponential complexity is not one of them. But AI agents can excel at processing hundreds, thousands, even millions of data and functionality combinations and sequences.

As we unshackle what martech systems can do from the martech UX we use to manage them, underlying system complexity will be free to grow at a much faster pace. It will look more like the chart above than the one at the top of this post.

Will such exponential growth in martech system complexity be a good thing?

Truthfully, it’s hard to predict. As noted, humans aren’t great at comprehending exponential complexity. But I think the answer will be yes. Why? Because customers and markets are complex. Enabling our martech systems to adapt to that complexity with far greater fluidity than we could have ever dreamed of in the past has the potential to dramatically improve the effectiveness and efficiency of marketing.

Customers are complex.
Therefore, markets are complex.
Therefore, marketing is complex.
Therefore, martech systems are complex.
But martech UX doesn’t need to be as complex anymore.

I’ll reaffirm: this will be a profound inflection point in martech.

The post The amazing turning point when martech systems complexity and martech UX complexity diverge appeared first on Chief Marketing Technologist.

Last week, Frans Riemersma and I published our Martech for 2024 report, an in-depth analysis of the evolving martech landscape in the gen AI era and the underlying forces of aggregation and composability that are shaping it. You can download a free copy here.

We also hosted an hour-long presentation, discussing many of the themes of the report, with the added color commentary of two giant martech nerds armchair industry analysts weighing in on some of these findings.

Missed it? Fear not. You can now catch the replay on YouTube — or watch it directly below if you’re reading this on my website. (Reading this in email? First, thanks for subscribing. Second, click here to jump to the playback on YouTube.)

Don’t forget to download a copy of the report to have all of our graphics and charts, as well as deeper explanations of the topics we cover in the video.

Once agian, we’d like to express our gratitude to GrowthLoop, mParticle, OfferFit, SAS, and Snowplow for sponsoring this report and webinar. Thank you!

The post Missed our Martech for 2024 jam session? Catch the replay and get the 89-page report for free appeared first on Chief Marketing Technologist.

There was a meme circulating around social media last week of an agency marketer rambling in a “thought leadership” interview on TikTok. I have no idea if it was parody or real. Or more accurately, if it was intentional or accidental parody. It was a jumbled jubilee of buzzword bingo – personalization, data, brand, experience, customer-centricity — that wound up saying absolutely nothing at all.

There but for the grace of God go I in any 60-second interview clip.

I feel for the speaker. It’s hard to explain why this is such an incredible time of technology-powered transformation in marketing in a few brief remarks. And if you try to tie it to the keywords that people are overindexing on at the moment, it’s all too easy to blend into the same backbround-radiation platitudes we’ve been observing for two decades of digital marketing. Now with “AI” inserted in every other sentence.

Which is my — ha — awkward segue to announce that Frans Riemersma and I have just published Martech for 2024, an 89-page report on the state of marketing technology as we head into 2024.

You’ll be the judge, but I hope you find it a different — and useful — read. Rather than short, listicle-like predictions of what we imagine might happen next year, we started with empirical data of what is already happening today. We try to explain those phenomenon in plain language and then extrapolate where those trends are logically taking us.

We discuss AI, of course. But we spend more time describing major archictectural shifts underway in martech — aggregation and composability — that will enable much more advanced AI use cases. These are the giant shoulders that gen AI will stand upon.

As a kind of “easter egg”, we also include a mid-year update on the marketing technology landscape. Thanks to a flurry of AI-powered startups, the landscape grew 18.5% in just the past six months. It took four years for the landscape to grow from ~150 to ~2,000 solutions, from 2011 to 2015. Now we added that many apps in a few short months.

Now, I know this will be immediately rebutted by people arguing that the half-life of many of these apps is likely to be quite short. They may be right. Indeed, the forces of consolidation are hard at work across the entire SaaS universe.

But there’s more to the story. We dive into a thorough examination of what’s happening in the long tail of martech apps. It is far from homogenous, with high variance in both their longevity and growth potential. It is an engine of continual renewal, driving the evolution of our industry.

That hot mess of startups and even pre-startup hobby horses, rapidly iterating new ideas on the frontier, is the star foundry from which red giants — or blue giants or orange giants, if you get my drift — will be born, fed, and inspired.

To appreciate this cycle of renewal, consider the distribution of martech apps in over 1,000 stacks that Frans and I collected over the past 7 years:

Despite the enormous forces of consolidation by major platforms in the head — Adobe, HubSpot, Microsoft, Oracle, Salesforce, etc. — we see a relatively constant balance with torso apps (large specialist apps, generally north of $100 million in revenue) and small apps in the tail (everything else).

This isn’t a cost distribution. Or even a scale of usage distribution. But it is a distribution of the count of apps in marketers’ tech stacks that empirically show the torso and the tail have remained vibrant parts of the equation.

It’s important to recognize that the composition of our martech stacks is never static. We constantly see apps in the tail — and to the lesser extent in the torso — shuffled out as either their relevance or uniqueness wanes. But then new apps in the tail, pushing some new boundary in martech innovation, emerge to take their place.

This dynamic is what brings us back to the themes of aggregation and composability.

But if you want the full explanation of that, you’ll need to download the Martech for 2024 report, where another 85 pages await you beyond the surface I’ve barely scratched here.

One more thing. We’re incredibly grateful to GrowthLoop, mParticle, OfferFit, SAS, and Snowplow, who sponsored this report that we’ve been researching for the past five months.

They had no say over the body of our report. However, we did include interviews with each of them to elicit their perspective on these trends. While they obviously have commercial interests in their points of view, they also have tremendous domain expertise in the areas in which they’re competing. We guided these interviews to tap their insights and experience in a non-promotional manner:

• Aggregated audience and customer journeys (GrowthLoop)
• Customer data infrastructure in an AI era (mParticle)
• Overcoming the three bottlenecks of personalization (OfferFit)
• An important time for responsible marketing (SAS)
• Operationalizing first-party data in the customer experience (Snowplow)

I think you’ll find these interviews fascinating in their own right. I certainly did.

Again, you can download the full report here.

The post Major trends in martech for 2024: the real changes underway in a 99% platitude-free report appeared first on Chief Marketing Technologist.

Want to know what 2024 has in store for martech?

I’m not talking about wild-eyed, hand-wavy, link-baity prognostications that proliferate in every year-end prediction season like that one Mariah Carey Christmas tune. I mean real-world, fact-based, happening-now trends that will inform your marketing operations and technology strategy for the next 12 months.

On December 5, Frans Riemersma and I will release a 50+ page report on Martech for 2024. Based on our research over the past 6 months, it will provide updated empirical data on the current state of the martech market and examine major trends underway in the evolution of companies’ martech stacks.

We’ll share:

• How much the martech landscape has grown — or shrunk? — since May 2023
• Which martech categories are seeing the most new AI-native ventures
• How the composition of martech stacks have evolved over the past 5 years
• Why use case analysis is key to utilization and smart consolidation
• Understanding aggregation at the data layer with cloud data warehouses
• How generative AI is defining the future of workflow and UI aggregation

You can be the first to get this report and hear our colorful color commentary about it by joining us for a free webinar on Tuesday, December 5, at 11am EST (5pm CET and 8am PST). Sign up here, and we’ll send you the all-access link.

We’re grateful to GrowthLoop, mParticle, OfferFit, SAS, and Snowplow for sponsoring this — and to Goldcast for powering our groovy virtual event. Their support has enabled us to invest in this research and make it freely available to you and the rest of the martech community. Thank you also to Clearbit and G2 for providing enrichment data on the martech landscape.

Don’t miss this informative and entertaining session. Sign up now to get that all-access link, which you can view live or on-demand for the rest of December. Make our wish come true. All we want for martech is you.

The post What will happen in martech in 2024? A no-BS report and webinar by two giant martech nerds appeared first on Chief Marketing Technologist.