In-House vs. Agency Marketing: What to Own and What to Outsource

The in house vs agency marketing question rarely has an answer at the department level, because no business actually decides its whole marketing function in one move. It decides whether to hire a paid media specialist, whether to renew a creative retainer, whether the analytics stack should be run by an employee or a partner. Asked at that level, the question gets tractable. Own the functions where institutional context and decision rights create the value: strategy, brand standards, first-party data governance and the measurement questions the business needs answered. Buy the functions where specialist depth, tooling and bench redundancy are expensive to maintain for one brand. Cost settles the argument only once both sides are fully loaded. The ANA found cost efficiency to be the top perceived benefit of in-house agencies by a wide margin, and that survey did not measure whether the savings were realized. The tests that follow let you check whether they were, function by function.

Most writing on this subject comes from one side of the trade and reads like it. Agencies conclude that outsourcing wins. Production platforms conclude that in-housing wins. Software vendors conclude that automation wins. GPI's view is that the structural question deserves the same treatment as an agency pitch: evidence first, limitations stated. The ANA finding that in-housing is perceived as cheaper is real, and so is its caveat that realized savings went unmeasured. The claim that agencies outperform on programmatic execution comes from an agency. Treat both as hypotheses. Fill in your own fully loaded numbers, decide function by function which business questions need answering, and only then decide who answers them.

TL;DR: How to Split Marketing Between In-House and Agency

• Neither model wins in general. The decision is made per capability, not per department, and most mid-market brands end up with a hybrid marketing model whether they designed it or inherited it.
• Cost comparisons mean something only when in-house headcount is fully loaded (benefits, tooling, management time, recruiting, ramp-up, backfill) and agency fees include the internal oversight time they still consume.
• Own what depends on institutional context and decision rights: strategy, brand standards, first-party data governance, budget allocation and the measurement questions the business needs answered.
• Outsource what depends on specialist depth, expensive tooling and bench redundancy: technical media execution, programmatic, platform-specific channels and surge creative production.
• In-house teams carry a risk that ANA-linked research surfaces: internal decision hierarchies constrain them, so ownership must arrive with real authority.
• Structure the outsourced half with project scopes or outcome-linked fees where the agency controls the outcome, and verify agency claims against evidence before signing.

Worth checking which of your current functions sit on the wrong side of that line. A media buyer who also owns strategy, tag management and creative briefs is a common example of the generalist trap.

The Decision Is Made Per Capability, Not Per Department

Why the binary framing keeps producing bad org charts

Most comparison articles run a pros-and-cons list for the whole marketing department: in-house knows the brand, agencies bring specialists, pick one. The business never faces that choice. It faces a sequence of smaller ones, each with a different profile of context, depth and risk. Treating them as one decision produces org charts where a strong strategist is asked to run programmatic bidding, or where an agency drifts into setting positioning because nobody inside was assigned to own it. The useful unit of analysis is the capability, and the rest of this article works at that level.

Perceived savings versus realized savings

The strongest evidence that brands believe in-housing saves money comes from the ANA, whose respondents ranked cost efficiency as the single primary benefit of an in-house agency, far ahead of any other benefit. Two things about that finding matter. It is a measure of perceived benefit, reported by the people who made the decision. And the survey did not measure whether the savings materialized after payroll, tooling and management overhead were counted. That distinction runs through everything that follows. A belief in savings is a reasonable starting hypothesis, and it stays a hypothesis until the fully loaded numbers are in.

Budget pressure makes the wrong split more expensive

Gartner describes CMOs facing falling budgets while caught in a cycle of more: more technology investment, bigger remits and expanding expectations. Under those conditions, an unexamined in-housing decision converts flexible spend into fixed headcount exactly when the ability to scale down matters most. The reverse error is also real: an agency retainer that quietly covers strategy leaves the brand without the internal judgment it needs when budgets get cut.

Three tests carry the argument from here. The fully loaded cost test asks what each option really costs once hidden line items are counted. The context-versus-depth test asks whether value comes from knowing the business or knowing the craft. The continuity test asks what breaks when one person leaves. A practical starting point is to list every marketing function you currently fund and mark whether its in-house or agency status was chosen deliberately or simply inherited. As you work through that list, GPI's directory criteria are a useful companion: for any function you are considering handing out, ask for the same documented proof of capability, methodology and limitations that GPI looks for when assessing agencies, and treat a partner's inability to produce it as a finding in itself.

Fully Loaded Cost: What In-House Headcount and Agency Retainers Actually Include

Line items competitors leave off the in-house side

Base salary is the visible in-house marketing cost, and it is usually the only one that appears in a comparison against a retainer. The rest of the stack is invisible until finance closes the books. Employer benefits and payroll taxes sit on top of salary. Recruiting adds agency fees or internal recruiter time, plus the months a role stays open. New hires work at partial productivity while they learn the brand, the stack and the approval culture. Software and platform licenses that an agency spreads across many clients land on one brand alone. Training and certification take hours out of the working year. The marketing leader spends management time on one-on-ones, reviews and career conversations. When the person leaves, backfill repeats the recruiting and ramp-up costs. None of these is exotic. They are simply left out because the comparison is usually built by someone arguing for the hire.

Line items buyers leave off the agency side

The agency side has its own omissions. The retainer or project fee is visible, but the internal time to brief, review, approve and chase work is not, and it does not disappear because the execution moved outside. The agency needs onboarding time to learn the brand, during which output is slower and revisions are heavier. Media or tooling may carry markup or pass-through charges, which should be identified in the contract. If the relationship fails, switching cost includes re-briefing a new partner and possibly losing account history and platform structure. A fair comparison loads both columns, not just the one the author dislikes.

Fixed versus variable cost and why it matters under budget pressure

Headcount is largely fixed. Salaries, benefits and licenses continue whether campaign volume is high or low, and reducing them means redundancies with their own cost and delay. Agency fees can typically be scaled, paused or ended at a contract boundary. That difference is invisible in a steady year and decisive in a falling one. Gartner's picture of shrinking budgets alongside expanding remits and technology spend describes exactly the environment in which fixed marketing cost becomes a liability rather than a saving. The rational move under uncertainty is to keep variable capacity variable until utilization is proven.

How outcome-linked fees change the comparison

Mordor Intelligence describes a shift from labor-hour billing toward outcome-linked fee structures tied to lead volume, incremental sales or brand-lift metrics. That is a market forecast overview, not contract data, so treat it as a description of direction rather than prevalence. Where it applies, it changes the cost comparison in two ways. The agency's fee becomes partly variable with results, which improves the fixed-versus-variable position further. It also creates a measurement dispute waiting to happen: if incremental sales define the fee, the brand and agency must agree in advance on how incrementality is measured, what baseline applies and who runs the analysis. An outcome fee without a pre-agreed measurement method is a negotiation deferred to the invoice.

A worksheet you can fill in with your own numbers

The research behind this article contains no approved salary, benefits or retainer benchmarks, so none appear here. The model below is a structure for your own figures.

For the in-house column, published employer compensation data from the U.S. Bureau of Labor Statistics and role-level salary listings such as Glassdoor can seed your estimates, though your own payroll data is the better source.

One error to avoid when the sheet is complete: comparing one salary to one retainer. A retainer often buys partial time from several specialists, so the honest in-house equivalent is fractions of several hires, not one person. Complete the worksheet for a single function you are considering moving in either direction, using your actual payroll, tooling and management-time figures.

Specialist Depth vs. Brand Familiarity: What Each Side Really Controls

Where institutional context creates value

Brand familiarity is the strongest genuine argument for keeping a function inside an in-house marketing team. An employee knows the product roadmap, has the sales director's mobile number, has sat through the compliance review that killed last year's campaign, and can judge in seconds whether a headline is brand-safe. That judgment is expensive to transmit to an outsider and slow to rebuild every time a partner changes. Integration with sales and product is a daily exchange rather than a quarterly briefing. Context is worth most where decisions are frequent, judgment-heavy and consequential for the brand.

Where specialist depth creates value

Specialist depth is the matching argument for external execution. An agency team working across many accounts sees platform changes, auction behaviour and creative fatigue patterns on a scale no single brand generates. Its tooling is paid for across a client base. Its people do one discipline all day and are measured on it. Depth is worth most where the craft changes quickly, where the tooling is expensive and where mistakes show up as wasted spend rather than brand damage.

The table reflects the mechanism described in this section and the agency-authored programmatic claim from Major Tom, discussed below.

The generalist trap: one hire for SEO, paid media and creative

The mechanism is arithmetic, not a judgment on the people involved. Each modern discipline changes at its own pace: search algorithms, ad platform features, creative formats. Staying current in one requires a steady share of working time. A single hire covering three disciplines can allocate perhaps a third of that attention to each, while the platforms keep changing at full speed. The gap between what the role knows and what current practice requires widens in every discipline simultaneously. The result is competent, shallow work everywhere, and it looks fine until it is compared with specialist output.

The hierarchy problem inside in-house teams

Proximity has a cost. Marketing Dive, summarizing ANA-linked work, reports that in-house agencies benefit from institutional knowledge and creative ability but are held back by operating practices and decision-making hierarchies that limit their contribution. That assessment comes from an organizational consultant and no methodology was published alongside it, so read it as informed commentary rather than measurement.

Major Tom, a media agency, argues that advanced programmatic campaigns tend to perform better when an agency owns technical execution even where the brand keeps strategy in-house. That is an agency describing its own model, so it is a hypothesis for you to test with your own performance data, not a proven result.

These two points resolve the trust question. Brand familiarity is a governance asset. A hybrid model keeps brand owners inside and lets them govern external execution through briefs, standards and review rights, rather than asking an outsider to absorb the brand or an insider to absorb every craft. For each function, write one sentence on what an insider knows that an outsider cannot, and one on what a specialist knows that a generalist cannot. The longer sentence tells you where value sits.

Speed and Continuity: Time to Launch and the Single Point of Failure

Recruiting, onboarding and ramp-up versus agency start dates

Building a capability internally runs through a fixed sequence: define the role, recruit, wait out the notice period, onboard, then months of partial productivity while the hire learns the stack and the approval culture. None of those stages can be skipped, and your own recruiting records will show how long each has taken for comparable roles. An agency typically arrives with an existing team, a process and platform accounts already configured. It still needs brand onboarding, and the first weeks will involve heavier revisions, but the start date is a contract date rather than a hiring outcome. Speed to market is therefore an early advantage for external execution, narrowing as an internal team matures.

What happens when your marketing manager resigns

A small internal team concentrates three things in one or two people: platform admin access, historical knowledge of what was tested and why, and relationships with vendors and internal stakeholders. When that person leaves, all three go at once. The hiring gap repeats the recruiting sequence above while campaigns run on autopilot or stop. Marketing team continuity is rarely planned in advance; most teams find the gap only when someone hands in notice. The practical defence is documentation, shared access held by a second named owner, or an external partner who holds process knowledge alongside the internal team.

Agency bench depth and its limits

Agency redundancy is real but bounded. Teams are shared across clients, account staff change jobs too, and the person who learned your brand may move to another account without you being told. The buyer's job is to ask, before signing, how the team is allocated, who the named backup is, and where account documentation lives. An agency that can answer those questions in writing offers genuine continuity. One that gestures at a large staff count does not.

Project-based engagements as a speed lever

Digiday reports that as clients bring more work in-house, agencies structured for project-based work are seeing more requests for shorter, specialized engagements. That is trade reporting rather than a measured trend, but it points to a useful lever: a brand can buy capacity for a launch or a seasonal peak as a defined project, without adding permanent headcount or a long retainer. Project scopes also force clarity on deliverables and exit criteria that open-ended retainers often lack.

Speed is also a governance question. The hierarchy problem reported through Marketing Dive means an internal team can be slower than an external one if every asset passes through more approval layers. Proximity to decision makers only helps when those decision makers act quickly.

Document, for each function, who holds platform admin access and where process knowledge lives. Any function with a single name in both places is a continuity risk to address with documentation, a second owner or an external partner.

The Ownership Map: What to Keep In-House and What to Outsource

Each row below is derived from the three tests: fully loaded cost, context versus depth, and continuity. Applied together, they settle what to outsource in marketing and what to keep inside the business: functions whose value depends on institutional context and decision rights stay in-house by default, while functions whose value depends on specialist depth, tooling and bench redundancy default to an agency. The defaults are starting positions, and the flip conditions tell you when your situation overrides them.

The programmatic row rests partly on the agency-authored view from Major Tom and the fee models on Mordor Intelligence's description of outcome-linked structures.

Own: strategy, positioning and budget allocation

These functions are almost pure decision rights. Their value comes from knowing what the business can deliver, what the board expects and where the next product is going. An agency can inform them with market view and channel expertise, and should, but the decision to shift budget from one channel to another belongs to whoever answers for the outcome internally.

Own: brand standards and approval rights

Brand standards are the mechanism by which an in-house owner governs external execution. Written standards, a clear brief format and defined review rights let a specialist partner move quickly within boundaries. Without them, the brand ends up outsourcing judgment rather than production.

Own: first-party data governance and the measurement questions

Consent records, customer data, CRM and the rules for what any partner may access are compliance and control questions, and they stay internal even when every campaign is run externally. So does the definition of what measurement must answer. A partner can run incrementality tests or build attribution models; the business must decide which decision that measurement is meant to inform, otherwise the measurement answers whatever is easiest to measure.

Outsource: technical media execution and programmatic

This is where depth, tooling and redundancy are hardest to sustain for one brand. Major Tom's claim that programmatic performs better when an agency owns the technical layer while the brand owns strategy is the hypothesis to test here: run the comparison with your own spend and see whether specialist execution beats what your team achieves.

Outsource: platform-specific channels and surge creative production

Creator programs, marketplace advertising and similar channels change fast and are often used unevenly. Creative volume spikes around launches and seasons. Both profiles punish fixed headcount and reward variable capacity. Directory profiles such as AB Marketing Group's show the kind of documented capability record a buyer can inspect when looking for a specialist to sit alongside an internal team; the profile is an example of what to look at, not a recommendation.

Case by case: content, SEO, analytics engineering

Content leans in-house when it depends on product and customer knowledge and volume is steady. Technical SEO leans external because it is craft-heavy and episodic. Analytics engineering often starts external, when building the stack needs depth the brand lacks, and moves inside once data volume and continuity risk justify a full-time owner.

Privacy and measurement complexity: why the technical layer keeps moving outward

The mechanism here is conceptual, and the evidence ledger for this article contains no privacy study, so no figures follow. As browser and platform restrictions reduce directly observed conversions, measurement depends on consent management, server-side tagging, modeled conversions and experimental methods. Each of those is a specialist skill that changes often and benefits from exposure across many implementations. That raises the bar for the technical layer and pushes it toward partners who do this work daily. It does not move ownership of the data or of the measurement questions, which stay internal. The trend explains why the ownership map assigns execution outward while holding governance in.

Fee structure follows the map. Where the agency controls the outcome, outcome-linked fees are defensible once measurement is agreed. Where the brand controls strategy and the agency executes, project or scoped fees fit better. Fill this table for your own function list and compare it with your current org chart. Each mismatch is a candidate change with a named reason.

Matching Structure to Company Stage

Stages here are defined by observable signals rather than revenue bands: how many channels run, whether campaign volume is steady or lumpy, and whether a function is used daily or seasonally. The research ledger contains no thresholds, and none are implied.

The hierarchy risk draws on the ANA-linked commentary reported by Marketing Dive; the budget trigger reflects Gartner's description of falling budgets.

Early stage: one internal owner, everything else bought

The first marketing hire should hold governance: strategy, brand judgment and vendor management. Execution is bought as projects because volume is uneven and no single discipline would fill a full-time role. The common failure is the owner becoming the producer, writing ads and building dashboards while strategy goes unattended.

Growth stage: in-house strategy and brand, agency execution

This is where most hybrid marketing models settle. Internal roles cover strategy, brand, data governance and channel leadership; agencies handle technical execution and specialist channels. The marketing team structure works when briefs and review rights are explicit, and drifts when an agency starts setting priorities because nobody inside is assigned to.

Scaled stage: selective in-housing where volume is steady

In-housing becomes defensible for functions with steady daily volume where a dedicated specialist would be fully utilized and the fully loaded cost holds up. The main risk shifts from capability to authority. An in-house team given production work without decision rights runs into the hierarchy problem described earlier, and the proximity advantage turns into approval queues. Under budget pressure, Gartner's picture argues for keeping variable capacity external until utilization is proven rather than hiring ahead of it.

Signals that your current split has drifted

Three signals are reliable. The same person is running three disciplines. An agency is doing strategy because it filled a vacuum. Tooling licenses are paid for in-house and used at a fraction of capacity. Identify your stage by the signals, then check whether utilization data supports each in-house specialist role you fund.

Making the Transition Without Breaking Live Campaigns

The five steps below apply whether a function is moving out to an agency or back inside. Each has an owner and an exit criterion, and the order matters.

Step 1: Decide the function, not the vendor

Start from the ownership map: name the capability being moved, the reason from the three tests, and what success looks like in cost, quality and continuity terms. Write that rationale down before any agency pitch or job description exists, because it becomes the standard the change is later reviewed against. Exit criterion: a one-paragraph decision record signed by the marketing leader and finance.

Step 2: Set decision rights and access before the handover

Define who approves creative, who holds platform admin, who owns data access and who can change budgets. The hierarchy problem reported through Marketing Dive shows what happens when in-house teams receive work without authority; the same failure hits agencies asked to deliver outcomes without access. Exit criterion: an access and approval matrix with names in every cell.

Step 3: Choose the engagement shape: retainer, project or outcome-linked

Match the fee to the ownership. Defined work with a clear end suits project scopes, the format Digiday reports agencies increasingly building for. Ongoing execution suits a retainer. Outcome-linked fees, as Mordor Intelligence describes them, belong only where the agency controls the outcome and the measurement method is agreed in writing beforehand. Exit criterion: a signed scope with the measurement definition attached.

Step 4: Verify agency capability with evidence

Pitch decks describe what an agency wants to be known for. Evidence shows what it has done: documented case work with stated limitations, named team members and their allocation, platform certifications, references you choose rather than ones supplied. Before moving any function to an agency, work through GPI's marketing agency due diligence checklist to verify the capability you are buying. Exit criterion: every capability claim in the shortlist tied to a document or a reference call.

Step 5: Run a parallel period and review against the original decision

Keep the outgoing owner available while the incoming owner ramps, with a fixed end date. Live campaigns continue under the outgoing owner until the incoming one has passed a defined handover check. At the end, review against the decision record from Step 1: did fully loaded cost, quality and continuity land where the rationale said they would? Reviewing against a new rationale invented after the fact is how bad moves get ratified. Exit criterion: a written review that either confirms the move or triggers a defined rollback.

Write a one-page transition plan for one function with these five steps, named owners and the date the parallel period ends.

How GPI Helps You Evaluate the Outsourced Half

Turning the ownership map into an agency shortlist

Once the map identifies which functions to buy, the remaining risk is choosing a partner whose claims hold up. GPI's methodology and directory are built around documented evidence of capability rather than pitch language, so the shortlist step is to take the outsource rows from your map and look for partners whose recorded capabilities match those specific functions, not the department as a whole. A brand that needs technical media execution and surge creative should be looking at two evidence records, and possibly two partners.

Reading agency claims against evidence and limitations

Apply to prospective partners the same standard this article applied to its sources. The programmatic performance claim from Major Tom is an agency's account of its own model. The outcome-linked fee trend from Mordor Intelligence is a forecast, not contract data. The hierarchy problem from Marketing Dive is consultant commentary without published method. None of that makes them useless; it defines how far each can carry an argument. Ask any agency the same questions of its case studies: what was measured, over what period, against what baseline, and where does the evidence stop.

The move toward project work reported by Digiday and toward outcome fees gives buyers more engagement shapes to choose from. Each still needs measurement agreed before a fee is tied to it.

GPI publishes methodology and directory records; it has not run campaigns for clients and makes no performance claims. For who inside your organization should own each step of the buying decision, see GPI's guide to marketing procurement's role in agency selection.

FAQ

Should an agency ever own marketing strategy, or only execution?

By the decision-rights logic of the ownership map, strategy stays internal because its value comes from knowing what the business can deliver and answering for the result. An agency can and should inform strategy with market and channel insight, and an early-stage brand without a senior marketer may buy strategic input as a defined project. Ownership of the decision, and of budget allocation, should not sit with a party paid to execute it.

How do we compare one internal hire against a retainer that covers several agency specialists' partial time?

Break the retainer into the roles it buys and the share of each. As a hypothetical illustration, a retainer covering part of a media buyer, part of an analyst and part of a creative director compares against fractions of three loaded salaries, not one. Use the fully loaded worksheet on both sides, then compare cost per function rather than headcount against headcount.

When does outcome-linked pricing make sense, and what measurement has to be agreed first?

Outcome-linked fees make sense where the agency controls the levers that drive the outcome and the outcome is measurable within the contract period. Before signing, agree the metric, the baseline, the measurement method (including how incrementality is established), who runs the analysis, and how disputes are settled. Mordor Intelligence describes these structures as a market direction; your contract has to make them concrete.

What should we ask an agency about team allocation and documentation to test its continuity claims?

Ask for named team members and the share of their time allocated to your account, the named backup for each, the notice you receive when staff change, and where account documentation, platform structure and test history are stored and whether you retain access. Written answers to each of these are the real test; a large staff count on its own proves nothing about continuity.

How do we give an in-house team enough authority to avoid the decision-hierarchy problem?

Assign decision rights explicitly in the access and approval matrix from the transition process: who can approve creative within brand standards, who can reallocate budget within limits, and who can change platform settings without escalation. Limit the number of mandatory approvers per asset and give the team a direct line to the marketing leader. Writing that authority down is what counters the hierarchy drag reported in the ANA-linked commentary.

What is the trigger for bringing a function back in-house after outsourcing it?

Three conditions together: steady daily volume that would fully utilize a dedicated specialist, a fully loaded cost comparison that favours the hire after tooling and management time are counted, and a continuity plan with a second owner and documentation. If any one is missing, the function usually belongs outside for now, revisited when the signal changes.