The marketing and advertising technology industry has been growing following a familiar pattern for the last 15 years.
Each year, many new startups or mature companies identify a “new” problem or highlight an element of a long-standing problem — ad fraud, viewability, identity loss, privacy regulation, made-for-advertising sites, inventory transparency. Then, each year, many companies create solutions that solve the problems but in the process require more integration, more ID profile reconciliation, more overhead with more anomalies to reconcile and more platforms to manage. As a bonus, we often get more acronyms that no one remembers what they stand for.
In sympathy, at every digital marketing event, scores of digital marketing companies tout new features, new integrations, new ways to bag, tag and monetize every online interaction with every “profile” it can touch.
The stories and case studies get jazzier at these events but does all this jazz make marketing better?
Ask marketers and they will, publicly or privately, say no. The reason is clear. Marketers are on the front lines of managing all this complexity.
It is marketers who have to pay tech tax.
It is marketers who have to figure out how much (not if) budgets are getting siphoned off due to fraud.
It is marketers who struggle to show ROI because the share of media budgets that make it to active media is conservatively estimated to be only about 40% after tech tax and fraud have taken their big bite from the budget.
After a few decades of transformative innovation, is the industry actually moving forward?
History tells us what is really happening and it isn’t pretty. Plainly, the industry is stuck in a complexity quagmire. Transparency initiatives have stalled. Outcome resolution continues to be MIA. And no one can agree on what digital media actually delivers.
This rinse and repeat pattern of creating niche solutions to niche problems which are then layered on top of everything that already exists creates a landscape so complex it is incomprehensible.
This complexity accretion pattern is not unique to adtech. It is, in fact, a pattern with a long history. By learning the history of complexity and its systems, we can reveal a functional future for martech and adtech.
Tainter’s framework
The concept of the “complexity conundrum” is well understood in data and networking circles. “The ‘complexity conundrum’ refers to the paradoxical challenge where increasing system sophistication (in organizations, technology, or physics) aimed at enhancing performance actually makes systems fragile, unmanageable, or impossible to fully understand and control. This creates a situation where the effort to fix issues introduces more risk and inefficiency, demanding a shift toward simplicity, resilience, and optionality over pure efficiency.”
This concept was well documented decades earlier, in 1988, by anthropologist Joseph Tainter in his published work; The Collapse of Complex Societies. It examined roughly seventeen civilizations that rose, peaked, and fell. He argued that collapse is not really the work of barbarians, climate, or moral decay — those are proximate causes. The deeper mechanism is complexity itself.
Tainter’s thesis runs like this: societies solve problems by adding complexity. They build bureaucracies, hire specialists, raise armies, pass laws, layer on institutions. At first, each new layer yields enormous returns — the first irrigation canal, the first standing army, the first written record. But complexity is subject to diminishing marginal returns. The second canal yields less than the first. The hundredth bureaucratic department yields less than the tenth. Each new investment in complexity costs more and delivers less.
Eventually, the marginal return on complexity goes negative.
The system requires more energy, taxation, and coordination to maintain itself than it returns in benefit to its members. At that point, collapse becomes the inevitable outcome. People stop participating. Subsystems morph into bureaucracies that impede progress. The expensive scaffolding begins to creak and sway dangerously like some real-world version of Jenga.
Ultimately, the whole ecosystem collapses making an opening for a simpler arrangement that actually works.
That is the framework. It explains a lot of things — including, I think, where martech/ adtech is headed.
Historical precedents
History provides ample context to understand this dynamic that spans both major human events and business case studies. Each example will foreshadow digital technology’s future.
A. Civilizational Events:
Rome. The Western Roman Empire is Tainter’s central case. Faced with persistent threats — barbarians on the borders, civil wars, economic stagnation — Rome’s response was always more complexity: a larger army, larger bureaucracy, larger tax base, larger administrative apparatus. Each solution worked, briefly, and created a heavier maintenance burden.
By the late empire, and this is important, soldiers were paid in coin debased to near worthlessness, so more taxes were extracted by force from a populace that resented them, and the bureaucracy required to hold it all together consumed an ever-larger share of the productive economy. When collapse came, it was catastrophic for most people. In the case of Europe, the fall of the Roman Empire led to the Dark Ages – a 600-year period of stagnation and retrenchment.
The Classic Maya. The Maya followed the same pattern in a different register. Elites competed through monumental construction — temples, plazas, stelae. Each generation outdid the last. Agricultural intensification supported larger populations, which required more labor for the monuments, which required more intensification. When sustained drought hit in the 8th and 9th centuries, the system could not cope. The elites needed the monuments to legitimize their rule; the rule depended on the monuments. The “tax” of complexity over time meant cities were abandoned. Population dispersed into smaller, simpler village societies that persist to this day. The Maya did not vanish. Their complexity did.
The Soviet Union. The modern textbook case. Central planning attempted to coordinate the economic lives of millions of people through committee. Every shortage was met with another committee, another quota, another secondary market. By the 1980s, the shadow economy rivaled the legal one in size. The collapse in 1991 looked sudden, but the dynamics of collapse had been underway for decades.
B. Business examples.
These business-centered collapses, each illustrate a different flavor of Tainter’s mechanism:
1. Sears
For most of the 20th century, Sears was the Amazon of its day. The catalog reached every American household; its stores anchored every mall; its financial arm managed people’ financial needs from Allstate and Discover Card to Dean Witter. Together, Sears had a huge footprint in the American psyche.
By the 1980s, leadership believed its future was a “financial and retail supermarket” — you’d buy a fridge, a mutual fund, and a house in one trip. Deep integration between different corporate cylinders was required to realize this vision. It also meant a deepening crisis of complexity.
Each acquisition solved a real problem (growth was slowing in retail) but added a deep layer of new requirements by adding a whole financial services apparatus on top of its retail operations. Increasing complexity stretched management’s focus across too many priorities at the expense of their core business.
The result? Walmart and Target — simpler, more focused — ate the core business while Sears executives were busy running a conglomerate. By the time Sears spun off the financial businesses in the 1990s to refocus on retail, the retail muscle had atrophied. E-commerce delivered the final blow, but the collapse was structural: the company had become too complex to do any one thing well. Bankruptcy came in 2018. The simpler successor arrangement is what we have now — Walmart for cheap stuff, Amazon for convenience and selection, specialty retailers (brick/ mortar and online) for everything else.
2. The 2008 real estate financial crisis
The cleanest modern Tainter case in business. The original problem was simple: how do you make more mortgage loans without holding the risk? The first solution — securitization — worked. Then the layers came.
Mortgages were bundled into mortgage-backed securities. MBS were bundled into collateralized debt obligations. CDOs were bundled into CDO-squareds. Credit default swaps were written against all of it. Synthetic CDOs were built so investors could bet on mortgages without any actual mortgages being involved. Rating agencies blessed instruments no one could fully model. Each layer was sold as making the system safer through diversification; each layer actually made it more fragile by hiding where the risk lived.
By 2007, the system required more analytical and computational sophistication to understand than the institutions running it actually possessed. When subprime defaults rose — a small, knowable problem — the complexity meant nobody could tell which counterparties were exposed, so everyone assumed everyone was. The system seized. Many people lost everything.
3. General Motors before the 2009 bankruptcy
GM in 2008 sold cars under eight North American brands: Chevrolet, GMC, Buick, Cadillac, Pontiac, Saturn, Hummer, and Saab. Each had its own dealer network, marketing budget, engineering priorities, and union contracts. The brand sprawl was a legacy of Alfred Sloan’s brilliant 1920s strategy (“a car for every purse and purpose”) that had calcified into a structure no longer matched to the market.
Layered on top: a healthcare and pension obligation to retirees that, by the 2000s, added roughly $1,500-2,000 to the cost of every vehicle GM produced. This is how the famous claim that GM was “a healthcare company that happened to make cars” arose. It was closer to the truth than many realized.
Layered on top of that: a dealer franchise system protected by state laws that made it nearly impossible to close underperforming dealers. Layered on top of that: a bureaucracy whose decision cycles were measured in years while Toyota moved in a matter of months.
Each layer had been added to solve a real problem at the time. By 2008, the marginal return on the entire structure was deeply negative — GM was burning cash on every car it sold. Bankruptcy was the only way to shed the complexity that was strangling the company. Some brands; Pontiac, Saturn, Hummer, and Saab were eliminated. Dealer networks were cut. Pension obligations were restructured. The company that emerged was smaller, simpler, and actually profitable. The collapse wasn’t the end of GM; it was the price of getting out from under the complexity.
The adtech parallel
Now let’s look at digital marketing tech, a.k.a. martech and adtech.
The early internet had a simple advertising model: buy a banner on a site whose audience you want to reach based on the topics covered in the content. The bargain was crude but legible. Audiences were not targeted at an individual level but based on content interests.
It was simple and it worked to get marketing message out albeit in a brute way.
Then, soon enough, as promises about digital tech’s targeting and attribution precision proliferated, optimization became a “new” problem; how do you reach the right audience across many sites at scale at the “right” time?
The response was, you guessed it, complexity.
This is when the complexity in the marketing tech complex really went into overdrive. To optimize media buys a buyer goes through a DSP (Demand Side Platform), which connects to multiple SSPs (Supply Side Platforms) which connect to ad exchanges, which run header bidding wrappers calling client-side and server-side endpoints, evaluated by brand safety vendors, verified by viewability vendors, routed through SPO platforms, measured by attribution vendors using identity graphs that depend on cookies, hashed emails, alternative IDs, and clean rooms, gated by consent management platforms enforcing frameworks designed to comply with regulations that change every quarter.
Whew. And we all know this only scratches the surface. We haven’t even talked about integration with CRM systems or AI capabilities or data lakes or first party data requirements. The complexity goes on and on and on and on and on and on.
In this ecosystem, individually, each solution seems simple, even useful. However, their collective layers create significant complexity. In the aggregate, all these companies extracted real fees in hard costs and labor costs to manage the system. In the aggregate, less and less money was deployed to reach a real person in the real world.
We can detect the real cost of complexity in looking at the state of measurement. According to eMarketer, (February 2026), a whopping 75% of marketers think their measurement tech stack is broken.
“This disparity between measurement and reality leads to billions in misallocated spend and strategies that don’t match consumer behavior,” (source: https://www.emarketer.com/content/75–of-marketers-say-measurement-broken-ai-becomes-rebuild-strategy).
Due to complexity, we end up with a system where it is easy to spend far more than is worth in the revenue delivered. It should be obvious that complexity usually works in favor of tech firms at the expense of marketers.
As a result, complexity creates an unsustainable system when more than half of every advertiser’s marketing dollar disappears before reaching a person.
It is also clear that martech and adtech are at an inflection point; approaching the negative marginal return Tainter described.
The system requires more cost, more coordination, and more cognitive overhead to maintain than it delivers in actual outcomes. Buyers know this. Publishers know this. The middle keeps adding layers, because the middle is the main financial beneficiaries of the complex ecosystem.
Enter AI: the late-Roman case study
Remember that the Roman Empire’s immediate collapse was foreshadowed by the fact that its soldiers were being paid in devalued money.
This pattern is presenting itself with AI. The current hyped narrative, however, is that AI will solve all this complexity. AI bidders will optimize more efficiently. AI creative scales infinitely. AI fraud detection catches more bots. AI clean rooms preserve targeting without raw identity.
This is all true but there is a dark side too. AI runs the risk of being yet another complexity layer that further erodes the value of digital marketing tech.
Just as we saw in the late-Roman move — debasing the coin to keep paying the army, AI makes the existing complexity machine slightly more efficient but it will intensify the complexity machinery for brands.
In the same way that AI can optimize a brand’s campaigns, it can be just as powerful to generate the next generation of made-for-arbitrage publishers. It can build more sophisticated bot networks. It can create new optimization stories that belie reality. The gap between what the system measures and what actually happens widens, not narrows.
Brands, again, must reckon with this new level of complexity. The pattern remains consistently consistent; AI delays the complexity reckoning but it does not change the outcome.
Tainter’s framework still stands.
What a simpler digital marketing tech looks like
What survives the collapse? My hope is that two, simpler systems emerge on the other side. It probably looks something like this.
Advertisers will have a real choice between profile targeting and contextual targeting tech stacks.
In the former system, profile data won’t be bought and sold as exists today. In the simplified system, users will control their digital souls – all managed by their personal AI agents. This will significantly reduce the need for ID resolution and intense privacy requirements because people will be part of the ecosystem. Programmatic media will be the broad reach channel better managed by AI to detect fraud.
In the contextual targeting system, targeting is contextual and topical. The concept of adjacency, where cars are advertised on car sites and finance product ads run on finance content, will be powered by many, many outlets. This model won’t be dominated by a few big players but will resemble what worked for magazines a century ago – great content to attract high intent audiences.
This contextual system does not require surveillance complexity to function. It does not require complex scale apparatuses to deliver ads. Topics are stable and more limited in ad opportunities thus delivering more quality to marketers. Since topics are constantly replenishable (unlike people), contextual targeting provides strong intent signals on the part of the user.
The brave new world offer a duality of marketing approaches: the push system of profile targeting and the pull system of topic-centric targeting. Instead of today’s monolithic asymmetrical complex system, we would evolve to a balanced symmetrical ecosystem where the benefits can once again be realized by all those touched.
Both systems can work better and more profitably because the framework is simpler. Outcomes can be measured and ROI can be captured because the link between a marketing campaign and outcomes are now only separated by only a few degrees – not 600 layers of complexity.
The benefits of this dual system become obvious. The middle thins out reducing some of the complexity. Direct relationships return between advertisers and audiences add even more simplicity. Real publishers with real audiences can now command real premium prices improving the content for everyone. Real choices for marketers.
Most important, the simplified landscape means two systems diversify the marketing landscape so the needs of marketers are served first without first serving the masters of complexity.
As rosy a picture as this paints, this is not without a downside.
Most notably, existing stakeholders in the current dominant system will fight any evolution towards simplicity with everything they’ve got. As in every other collapse, the next iteration is smaller, less profitable, and less complex than the one that exists today. To effect change means a culling of the current bloat – something that will be resisted by all sorts of firms – large and small. People will lose jobs.
This won’t be easy, but it will work. Advertisers will reach humans, publishers will get paid for real attention, and the cost of running the system will be a fraction of what it is now. By having two systems available to marketers, this will naturally create a healthier ecosystem that drives better value for brands, better monetization for quality publishers and more usefulness of advertising for people.
Tainter’s framework forces us not to wonder whether this will happen but whether we are simply exacerbating the complexity system that exists now or will we embrace the simpler future is that sure to come.
Steve Jobs once said: “Simple can be harder than complex: You have to work hard to get your thinking clean to make it simple.” Truer words were never spoken.
P.S. – Now you know why our company is named engageSimply. It was a promise to our clients not just a brand name.
