The short version
- An agency sees the same failure at more than one company. Eight accounts and eleven brands means the second occurrence gets recognised rather than diagnosed from scratch.
- An in-house hire is present when decisions are made and keeps the memory. Our updates report status and never make the strategic call, which is the boundary and the limit.
- The split model fails at the seam. One blog stayed down for weeks because it sat between two vendors and two internal functions with nobody's name against it.
- If nobody can say where the first hundred customers arrive from, neither answer helps. We shipped a product that worked and earned nothing.
We run eight client accounts across eleven brands and still tell some companies to hire instead. An agency gives you pattern access across many companies and a cadence you can write into a contract. It cannot put somebody in the room when the decision gets made.
What does an agency structurally have that a hire does not?
Repetition across companies. Eight accounts and eleven brands means a failure arrives a second time and gets recognised rather than diagnosed from scratch. An in-house marketer has one company to learn from.
Methods worth building once. We measured two ways of making the same ad creative. Compositing a product screen onto a generated phone produced zero usable images. Generating it into the scene in one pass produced three of three. One company rarely gets enough shots at the same problem to justify measuring it.
A cadence you can contract. Weekly delivery on every account, protected the way a print deadline is. On a monthly cycle the gap between a wrong assumption and finding out runs to about eight weeks: twelve chances a year to correct course, not fifty.
What each side structurally owns:
| Dimension | Agency | In-house |
|---|---|---|
| Pattern access | Many at once. Ours is 8 accounts, 11 brands | One company, in depth |
| Cadence | Contractible. We ship weekly on every account | Competes with whatever else is urgent |
| Capacity change | Up or down without hiring or firing | Only by hiring or firing |
| Context | Constructed and maintained deliberately | Absorbed by being present |
| Memory | Leaves when the contract ends | Stays until the person does |
| Visibility | Has to be built. We run a portal per client | Walks past your desk |
| Specialist production | Built once, shared. One creative set ran in three markets | Justified once it is a full-time job |
What does an in-house hire have that an agency cannot buy?
Presence when a decision is made. Our client updates report status and never make the strategic call, because the call belongs to the client. That is the correct boundary and a ceiling on what an outside partner can be.
Memory that stays. Every engagement ends, and the knowledge of why earlier attempts failed leaves with it.
Handoff is usually the bottleneck rather than skill. We hired engineers into a marketing team, three of them, so the people who write the page can build it. Count the hands a piece of work passes through before it goes live, on either model.
Where does the mixed model break?
At the seam. A blog on one account we inherited stopped serving during a rebuild. Everybody saw it. It sat on status lists. It stayed down for weeks because it lived between two vendors and two internal functions, and nobody’s name was against the restore.
That is the real risk of running an agency alongside a team, and it is not a capability problem. What prevents it is a list of every recurring piece of work with one name on each line, including the redirect nobody wants.
When is the answer neither?
When nobody can say where the first hundred customers arrive from. We built an internal product, integrated payments and fulfilment, shipped it, and it earned nothing across its entire life. Nothing was broken. That question had been deferred to launch, where it turned out to need months of lead time.
An agency will execute around that gap and so will a new hire, because both are paid to produce marketing and producing marketing feels like progress.
The product worked and earned nothing across its entire life. Every question had been answered except how the first hundred people would arrive.
How do you decide?
Start with the constraint. If work ships reliably and nobody is sure which work matters, you are buying direction. If the direction is settled and nothing ships, you are buying capacity, and capacity starting Monday beats capacity starting after a notice period.
Then write the ownership list before signing anything: every recurring piece of work, one name against each line. If one person can hold the list, hire. If not, the comparison is an agency against several hires.
Two posts sit underneath this one: what happens when nobody’s name is on it and how to evaluate an agency once you have decided.
Frequently asked questions
Is it better to hire a marketing agency or build an in-house team?
It depends on the constraint. If work ships reliably and nobody is sure which work matters, buy direction. If the direction is settled and nothing goes out of the door, you are buying capacity.
What can a marketing agency do that an in-house marketer cannot?
See the same failure at more than one company. We run eight accounts across eleven brands, so a pattern we had to diagnose from scratch the first time is recognised on sight the second.
What can an in-house marketer do that an agency cannot?
Be in the room when the decision gets made, and hold the memory afterwards. Our client updates report status and never make the strategic call, because that call belongs to the client.
Does a hybrid of agency plus in-house work?
It works when every piece of work has exactly one name against it. The failure mode is the seam. A blog on one account we inherited stayed down for weeks because nobody owned the restore.
When should a company hire neither an agency nor a marketer?
When nobody can answer where the first hundred customers arrive from. We built an internal product, shipped it, and it earned nothing, because distribution had been deferred to launch.
Sources
- Chua Network delivery data across 8 client accounts (internal fact bank)
- Chua Network engagement records, anonymized (internal experience bank)
