MarTech · · 6 min read

WHY MARTECH STACKS ROT

Nobody buys a bad stack. They buy eight good tools over four years, and end up paying for a mess nobody can explain.

Read aloud by your browser
The short version
  • Marketing technology stacks fail through accumulation, not bad decisions. Each tool made sense on the day it was bought.
  • The usual cost is 20 to 30 per cent of licence spend on software nobody opens, plus the staff hours spent moving data by hand between systems that were meant to talk.
  • Rot shows up first in reporting: two systems give two numbers for the same question and nobody can say which is right.
  • The fix is an ownership audit before a tooling decision. You cannot rationalise a stack you cannot map.

Marketing technology stacks fail slowly, and almost never for the reason people expect. Every stack I have audited was assembled by sensible people making sensible decisions. The email platform was chosen because the old one could not segment. The CRM was chosen because the spreadsheet finally broke. The analytics tool was chosen because someone senior asked a question nobody could answer. Each purchase was defensible on its own day.

The problem is that nobody is responsible for the shape of the whole thing. Tools arrive one at a time, over years, chosen by different people solving different problems. No single decision was wrong. The accumulation is what costs you.

What rot actually looks like

It rarely announces itself. There is no outage, no invoice that suddenly doubles. It shows up as friction that everyone has quietly learned to work around.

  • Two systems answer the same question differently, and the monthly report depends on which one someone opened first.
  • A staff member exports a CSV from one platform every Monday and uploads it to another. This is described as "the process".
  • Nobody can name the owner of at least two tools you are paying for.
  • A field called "status" means three different things in three systems, and everyone knows which is which by memory.
  • Onboarding a new marketer takes six weeks because the knowledge lives in people, not documentation.
If the answer to "where does this number come from" is a person rather than a system, the stack is already rotting.

What it costs

20-30%of licence spend on software nobody meaningfully uses
1 monthof a person per year lost to a single weekly reconciliation
3systems that typically disagree on the same number

The licence waste is the easy number, and the least interesting one. Most organisations I work with are paying for somewhere between 20 and 30 per cent of software that nobody meaningfully uses. That is real money and it is worth reclaiming, but it is not the main cost.

The main cost is the hours. Every manual export, every reconciliation, every meeting spent arguing about whose number is right, is capacity you are paying senior people to spend on plumbing. Multiply a two-hour weekly reconciliation across a small marketing team and you have lost a month of someone every year to a problem a working integration would have solved.

The third cost is the one that actually hurts: decisions get slower and worse. When leadership stops trusting the reporting, they stop using it. They fall back on instinct, which is a fine tool for some questions and a terrible one for budget allocation.

Why the obvious fix does not work

The instinct is to buy something that joins it all up. A customer data platform, an integration layer, a bigger suite from a single vendor. Sometimes that is the right answer. More often it adds a ninth tool to eight you have not mapped, and the rot continues underneath a nicer dashboard.

Where the money actually goesRough shape of what an audit turns up on a mid-sized stack. The licences are the visible cost and the smallest one.

You cannot rationalise a stack you cannot draw. Before any tooling decision, you need to know what you have, who owns it, what it costs, what it is actually used for, and where the data goes when it leaves. That is a boring piece of work and it is the only thing that reliably makes the next decision cheap. The MarTech stack audit worksheet is the version of that map I use with clients.

The order of work

  1. Inventory everything, including the tools bought on a card by one person for one campaign.
  2. Name an owner for each. If nobody will claim it, that is your first answer.
  3. Trace the data: what enters each system, what leaves, and which handoffs are a human with a spreadsheet.
  4. Cut what nobody uses. Do this before integrating anything, so you are not automating waste.
  5. Fix the highest-traffic manual handoff first. It pays for the rest of the work.

None of this is glamorous and none of it needs a new platform. It needs someone to sit down with the invoices and the logins and be honest about what is there. That is usually the part organisations find hardest, which is why it is often the part worth bringing someone in for. If the ownership question is the one that worries you most, who owns your marketing accounts covers it on its own.

COMMON QUESTIONS

How do I know if my marketing stack has a problem?

The clearest signal is disagreement. If two systems report different numbers for the same question and nobody can say which is correct, the stack has a structural problem rather than a reporting one. The second signal is any weekly manual export between two systems that were sold as integrated.

How much does unused marketing software typically cost?

In the audits I run, 20 to 30 per cent of licence spend goes on software nobody meaningfully uses. That includes unused seats, tiers bought for a feature nobody adopted, and tools that were replaced but never cancelled.

Should I buy a customer data platform to fix it?

Not before you have mapped what you already have. A CDP added to an unmapped stack usually becomes a ninth tool rather than a fix. Map ownership and data flow first, cut what is unused, then decide whether the remaining problem needs new software at all.

How long does a marketing technology audit take?

For a mid-sized stack, the inventory and ownership map takes a few days of focused work. The harder part is getting honest answers about what is actually used, which depends on how willing people are to admit a tool they championed is not being opened.

Recognise any of this?