AI Makes Production Easier. Marketing Gets Harder.
ZITE / The 2026 In-House Barometer / Insight 03
AI can create more drafts, variations and assets at extraordinary speed. The harder part is deciding what is good, what is safe, what belongs to the brand and where people, partners and technology fit together.
The short version
What you need to know.
- Among respondents assessing AI’s effect on content quality, 50% say quality has declined and 22% say it has improved.
- 52% expect AI to complement human creativity, while 25% expect new roles and skills to emerge.
- Respondents point to skills, governance, business alignment and infrastructure as requirements for successful AI adoption.
- In-house is already hybrid: 44% of organisations with in-house resources also use external agencies.
01 / Analysis
The button is the easy part
AI has made it remarkably easy to demonstrate what is possible. A headline, an image, a rough film or a hundred variations can be produced in the time it once took to organise the first meeting. That is real progress. But a marketing organisation cannot publish a demonstration. It has to publish work that is accurate, compliant, on-brand and useful.
That gap shows up in the quality data. Among the 778 respondents who assessed AI’s impact on content quality, 50% say quality has declined, 27% report no meaningful change and 22% say it has improved. These are perceptions, not an independent quality test. But they make one point difficult to ignore: more output is not the same thing as better output.
Explore the data
AI and perceived content quality
Respondents assessing AI’s effect on content quality (n = 778). Perceptions, not an independent quality test. Reported rounded figures total 99%.
02 / Analysis
People are not disappearing from the picture
The same survey is much more optimistic about the future role of people. Fifty-two per cent expect AI to complement human creativity. Another 25% expect entirely new roles and skills to emerge. Thirteen per cent expect creative roles to be replaced entirely, while 10% expect no meaningful impact.
Those are expectations, not a forecast of actual employment. But the direction is interesting. The largest group is not imagining creativity without people. It is imagining people working differently with the technology.
Explore the data
What respondents expect AI to do
Expectations about AI, not measured employment outcomes.
03 / Analysis
AI needs an organisation around it
When respondents are asked what is necessary for successful AI adoption, 32% point to AI-skilled personnel, 29% to governance policies, 26% to alignment with business goals and 20% to infrastructure. Multiple answers were possible.
That is a useful antidote to the idea that implementation is primarily a licensing exercise. Someone has to decide which tools are approved, which information may be used, what needs human review, how brand knowledge reaches the tool and who is accountable when the output is wrong.
Explore the data
What successful adoption needs
Separate reported requirements. Responses can overlap and should not be added to 100%.
04 / Analysis
The dashboard and the desk do not always agree
There is also a revealing perception gap. Twenty-five per cent of business leaders say technology has made marketing more efficient, compared with 11% of marketing leaders. The figures are perceptions, not measured time savings, and they do not prove that one group is right. They do suggest that the view of a technology rollout can look different from the dashboard than it does from the desk.
That is why AI measurement should follow the complete assignment. How much usable work reaches approval? How much correction is required? How much review time has moved elsewhere? Did the process become faster, or did generation simply become faster?
Explore the data
One rollout. Different perspectives.
Share saying technology made marketing more efficient. These are perceptions, not measured time savings.
05 / Analysis
In-house is already hybrid
AI is only one contributor in an increasingly mixed marketing system. Forty-four per cent of organisations with in-house resources also use external agency partners. Internal capability has not removed the need for specialist expertise, outside perspective or flexible capacity.
And organisations with in-house resources are much more likely to say they have compared internal and external efficiency: 58% versus 28% among organisations without in-house resources. That does not tell us which model performed better. It tells us who has made the comparison.
Explore the data
A hybrid system
Agency partners
The comparison figures show who compared internal and external efficiency, not which model performed better.
Efficiency comparisons
The comparison figures show who compared internal and external efficiency, not which model performed better.
06 / Analysis
Creative Orchestration is the layer in between
The emerging job is to connect all of those contributors without losing the idea, the brand or accountability. That means deciding what should be made, where it should be made, who should contribute, which tools should be used and how the work moves from brief to approved output.
That is what we mean by Creative Orchestration. It combines three things:
- People: creative leadership, operations, internal teams, specialists and external partners.
- Process: intake, prioritisation, capacity, routing, review and governance.
- Technology: work management, asset management, AI, automation and the data that keeps the system connected.
AI makes this more important, not less. When production capacity becomes abundant, judgement, context and coordination become scarcer. The organisations that learn to orchestrate those things will be able to use AI without turning marketing into a larger pile of outputs waiting for somebody to decide what to do with them.
Explore the data
People. Process. Technology.
The article’s Creative Orchestration framework.
Go a little deeper
Questions worth asking.
Definitions and context for reading the figures.
Has AI objectively reduced content quality?
The survey does not establish that. It records respondents’ perceptions of quality, not an independent assessment of the content.
Do the AI figures show actual job losses?
No. The role figures describe expectations about how AI may affect work. They do not measure employment changes.
Does comparing efficiency show that in-house wins?
No. The 58% versus 28% comparison shows which organisations have compared internal and external efficiency. It does not show the result of those comparisons.
Hello, World!
