CMO playbook & advanced tactics for agency management

The decision that costs the most is not which shop wins the pitch. It is the work you quietly decide never to buy from outside again.

Every planning cycle shaves a little more off the external budget: social content first, then production, then media planning, then, eventually, the brand platform itself. Each move survives its own business case. Together they change what your marketing organisation can do, and that change is far easier to make than to undo. This lesson is about the arbitration at company scale: what in-housing costs once you price capability rather than deliverables, what a pitch costs on both sides of the table, and what it takes to reverse an over-rotation after the numbers have stopped moving.

The arbitration only you can make

Take the structural options as given (the three the frameworks lesson lays out) and the fee mechanics as given too. What lands on the CMO's desk is narrower and harder: where the boundary line goes, how often you are permitted to move it, and who is honest enough to tell you when you have moved it too far.

Four things make that judgement unreliable:

• The saving shows up in the current fiscal year. The cost shows up two or three years out, in work that is competent and forgettable.
• Everyone who prepares the analysis has a directional interest. Procurement is measured on fee reduction. Your in-house studio lead is measured on volume absorbed. Your agency is measured on scope retained.
• The counterfactual is invisible. You never see the campaign the outside team would have made, so the comparison is always between a real cost and an imagined one.
• Capability degrades silently. Nothing breaks on the day it stops working.

The unit cost comparison that is always wrong

The standard analysis divides agency fees by assets produced and compares that to an internal cost per asset. It flatters in-housing every time, because it prices salaries and ignores everything wrapped around them.

A fully loaded internal head runs roughly 1.3 to 1.5 times base salary once you add employer contributions, benefits, recruitment, space, and the licence stack a modern studio needs. A twenty-five person studio is therefore a multi-million dollar fixed cost that exists in the months when demand does not. An agency prices its own bench into a blended rate and carries the idle time on its balance sheet, not yours, in exchange for a margin that sits somewhere in the mid-teens on most fee-based arrangements. You are not comparing cost to cost. You are comparing a fixed cost to a variable one, and paying a premium for the option to stop.

That makes in-housing a bet on demand stability. Below roughly seventy percent utilisation you are funding idle capacity. And in-house studios rarely fail at the average, they fail at the peak: a product launch, a Q4, a sporting calendar. The peak is met with freelancers hired at panic rates, which is the exact arbitrage the studio was built to remove.

Capability loss is a ratchet

Standing up an internal team takes six to nine months to hire and another year or more to get good. Unwinding the decision is slower, because what you lost was not headcount, it was a relationship carrying years of context about your category, your legal constraints and your CEO's tolerance for risk.

Two second-order effects do most of the damage. The first is A-team drift. When you halve a scope, your account stops being one of the shop's largest revenue lines, and the senior people who made the work good get reassigned to whoever is now paying more. You keep the name on the roster and lose the reason you hired it. Nobody sends a memo. You notice three quarters later.

The second is measurement gravity. An in-house team is judged on the metrics it controls, and the metrics it controls are almost all lower funnelfunnelThe customer journey from awareness to purchase, typically Awareness, Interest, Consideration, Decision, Action, with prospects narrowing at each stage.View full definition →, so budget migrates towards what the team can prove. Nike ran that experiment at full scale: through the Consumer Direct Acceleration years it expanded in-house digital and performance capability and leaned hard into measurable demand creationdemand creationCreating and stimulating demand for your offer, often upstream of the buying process to generate interest and awareness before prospects are ready to buy.View full definition →. In 2024 the company's own leadership said publicly that it had over-invested in performance marketing and under-invested in brand storytelling. The in-house build was not the mistake. The mistake was that the measurement system attached to it had no way to argue for the spend it was displacing.

What a pitch actually costs, on both sides

Reviews are the instrument CMOs reachreachThe number of unique people exposed to your message in a given period. Unlike impressions, reach counts each person once, no matter how often they see it.View full definition → for when a relationship is disappointing, and they are the most expensive tool on the shelf.

On your side, a large creative or media review commonly takes six to nine months end to end, with search consultant fees in six figures for a global account, before you count the senior internal hours: brand leads, procurement, legal, and your own. On the agency side, pitching a major account can absorb hundreds of thousands in unbilled time and speculative production. That cost does not vanish. It is recovered in the fees of the accounts that shop wins, including yours.

Then there is churn. Average client-agency tenure now sits around three years, a fraction of what it was a generation ago, and every reset spends the first two quarters of a new relationship rebuilding knowledge the outgoing team already had. Worse, the incumbent's behaviour changes the day the review is announced: nobody invests in an account they are being asked to defend.

The practical rule is unglamorous. Do not call a review to solve a problem you have never named to the incumbent in writing, with a remediation window and a date. Renegotiation, scope resets and a documented performance period cost a fraction of a pitch and preserve the context you have already paid for.

Where the line sits: volume against the marginal idea

The usable split is not by discipline, it is by what you are buying. Work whose value comes from speed and quantity belongs inside. Work whose value comes from being different from what everyone else made belongs outside, because difference is what you cannot manufacture from a team that only ever sees one brand.

AB InBev has run that split deliberately since launching draftLine, its in-house agency network, in 2018 out of Brazil and then across a long list of markets. Local, high-volume, fast-turnaround content sits inside, close to the market and the data. The global brand platforms still go to outside shops, because a beer brand competing for cultural attention needs an idea that did not come from the same room as last year's.

Duolingo is the counter-example that stops this being a formula. Its social work is made by a very small internal team, and the asset is reaction time measured in hours: a five-second Super Bowl spot in 2024, the owl's on-brand "death" stunt in February 2025. You cannot brief a 48-hour cultural reaction through a review cycle. But notice why it generalises poorly. The brand voice is one character, the risk appetite is unusually high, and approval sits with a very short chain of people. Remove any of those and the same structure produces slow, committee-safe content at internal cost.

Reversing an over-rotation

Reversal is the part nobody budgets for. Nike's correction did not begin with a campaign, it began with an admission in 2024, a change of CEO in October of that year, and a restored commitment to brand spend, with major brand work like the Paris Olympics campaign returning through Wieden+Kennedy, a partner it had never fully let go of. That last detail is the lesson: reversal is far cheaper if the relationship still exists, even thinned.

Three costs land when you turn the ratchet back. You pay for four to six quarters of brand spend before the effect is legible in anything your CFO tracks. You pay to reinstate senior external talent, at current rates rather than the rates you locked years ago. And you carry an internal team whose skill mix was built for a different job, which means retraining, redeployment, or a reduction that everyone in your market will read about.

CMO action items

• Rebuild your in-house versus external comparison on fully loaded cost, including idle capacity at your actual seasonal demand curve, not the annual average. If your studio's utilisation drops below seventy percent for a full quarter, that quarter is the real number.
• Name the two or three capabilities you will never bring inside, write them down, and defend them at the next budget cycle. Undefended capabilities disappear one scope reduction at a time.
• Before authorising any review, put the specific failure in writing to the incumbent with a remediation window of at least a quarter. Reviews you can avoid are worth six figures and six months.
• Add one brand-level metric that your in-house team is accountable for but cannot fully control. Without it, measurement gravity will move your budget for you.

Common mistakes that kill results

• Comparing cost per asset instead of cost per outcome. Internal teams win the first comparison almost automatically, which is why it gets used.
• Cutting scope without checking who leaves the account. Ask, in writing, which named senior people remain assigned and at what percentage of their time.
• Treating a review as a negotiating tactic. The incumbent stops investing the day it is announced, and you pay the transition cost whether or not you switch.
• Building an internal studio around a peak year. Demand that arrived once tends not to justify a permanent bench.
• Assuming reversal is a campaign decision. It is a multi-year funding decision with a headcount consequence attached.

Key takeaways

• The in-house versus external line is a bet on demand stability and on whether you are buying volume or difference. Price it as fixed cost against variable cost, not as salary against fee.
• Capability moves inside faster than it moves back out. A-team drift and measurement gravity do the damage quietly, over quarters.
• Pitching is expensive on both sides and the cost returns to you in fees. Fix relationships with named problems and remediation windows before reaching for a review.
• Nike's public 2024 correction shows the shape of over-rotation: measurable spend crowds out brand spend until growth stalls, and the repair takes years and a partner you did not fully cut.
• Duolingo's in-house social works because of speed, a single voice and a short approval chain. Copy the conditions or do not copy the model.