Your Marketing Stack by Team Size: Solo, Five, Twenty
What you actually need at each stage, what to add next, and the signal that tells you it is time.
CTFM Team
Stack advice is usually a diagram with thirty logos on it, produced by someone selling one of them. It tells you nothing about what to buy on Monday.
A simpler way to think about it: your stack should be sized to your team, not your ambition. Here is what each size actually needs, and the specific signal that tells you to add the next thing.
Every tool has three costs. The subscription, the time to run it, and the time to keep its data correct. The third one is the one nobody budgets for and the one that kills stacks.
What the research says about over-buying
Gartner's 2025 Marketing Technology Survey put average martech utilisation at 49%, meaning organisations use around half of what their stacks can do. Gartner's 2026 CMO Spend Survey, based on 401 CMOs, also found the share of marketing budget going to martech at a five-year low of 19.4%, down from 26.6% in 2021, even while most respondents said they planned to invest more in technology.
Read those together and the picture is not "buy less". It is that buying capability has been easy and using it has not. Both figures come from vendor-independent survey research, but they are self-reported by marketing leaders and skew towards larger organisations, so treat them as direction rather than a benchmark for a five-person team.
Now the problem changes. It is no longer your time, it is that two people have different versions of the same information.
Add, in this order:
A real CRM
The signal: two people contacted the same lead, or someone asked what happened with an account and nobody could answer. That is the moment, not before.
A shared work tracker
The signal: you are running campaigns from a chat thread and things are being missed. Any project tool will do. The choice matters far less than everyone actually using one.
Basic marketing automation
The signal: you are sending the same sequence manually more than once a week. Start with the automations you have already proved by hand.
One reporting view
The signal: two people bring different numbers for the same thing to the same meeting. Even a manual monthly sheet, as long as it is the one everyone uses.
The integration question, asked properly
"Does it integrate?" is not a useful question, because everything claims to. The useful question is: which specific fields sync, in which direction, and how often? A tool that syncs contacts one way, nightly, is a very different product from one that syncs the full record both ways in real time, and both are marketed as an integration.
At this size the failure mode flips again. It is not missing tools, it is overlapping ones. Three teams bought three products that each do 70% of the same job, and nobody can see it because nobody owns the list.
Somebody owns the stack
Not as a hobby. An actual named person with authority to say no to a purchase.
Everything is written down
Tool, owner, cost, renewal date, what it is for. A spreadsheet is fine. Not having one is the problem.
New tools need a case
What job, what it replaces, who runs it. Three sentences. It stops most bad purchases without slowing down good ones.
What genuinely gets added at this stage: attribution or analytics beyond the basics, a content and asset library, permissions and approvals, and a customer data layer once you have more than a few systems holding customer records.
What gets bought at this stage and should not: a large suite bought to solve a process problem. If three teams disagree about who owns lead follow-up, no platform will settle that. It will encode the disagreement and charge you annually for it.
First, check the price at your next size, not your current one. Contact-based and seat-based pricing has a way of tripling on the day the tool starts working.
Second, test the export before you commit, not after. The cost of a bad tool is not the subscription, it is the migration.
