The Gartner stat is the one every agency owner who reads it tries to forget.
39%
Of mid-market CMOs plan to cut agency budgets this year. (Gartner 2025 CMO Spend Survey)
The follow-up question, the one Gartner asked next, is the more important one: what tactic will you use to cut? The top answer was "eliminating unproductive agency relationships." Not reducing scope. Not renegotiating fees. Eliminating. Firing.
This piece is about the four reasons agencies get fired in the mid-market and what the proactive layer looks like that gets you renewed instead. Not opinion. Pattern recognition from running a 50-person agency that watches client books churn and doesn't churn ours.
What "unproductive" actually means in the CMO's head
The Gartner question used the word "unproductive." When we asked CMOs what they meant, the answers fell into four shapes. None of them were about deliverable quality.
I can't see what you're doing.
The most common answer. The CMO can't tell, between calls, whether work is happening. The deck arrives Friday. Nothing is visible Tuesday. The CMO doesn't know if the agency is heads-down or stalled out.
This is not a quality problem. The work is fine. The visibility is the problem. The CMO is paying $7,500 a month and from her seat, the engagement looks like a mystery box that opens once a week.
I can't defend you to my CFO.
The agency's monthly report is in marketing dialect. The CFO is in finance dialect. The CMO has to translate, and she has to translate without preparation because the agency report didn't pre-translate it for her.
When the CFO asks "what are we getting for $90,000 a year on this agency," the CMO opens five tabs and starts cobbling together an answer. By the time she has a clean version, the conversation has moved on. The CFO concludes "she couldn't defend it" and the agency line item gets flagged for next quarter's cuts.
The agency didn't fail. The translation layer failed. We wrote about this in detail here.
You're reactive when I need you proactive.
The CMO finds out about the competitor's pricing change from her own sales team. The agency hears about it on Tuesday's call, when the CMO mentions it. The agency is supposed to be the early warning system. Instead, it's a recap of things the CMO already knew.
This is the most expensive failure mode because it removes the strategic value of the relationship. If the agency is just executing on what the CMO tells it to do, the CMO has hired an agency to do execution, which is half of what she wanted. She wanted execution plus signal. She got execution.
You feel like a vendor, not a partner.
The vaguest reason. Also the deepest. When the CMO says this, she means: the relationship has no shared workspace, no shared narrative, no shared decision-making cadence. The agency sends invoices and decks. The CMO files them. The agency moves on. There is no "we" in the engagement.
This is the reason that gets cited last and acts on first. When budgets tighten, the relationships that feel like partnerships survive. The relationships that feel transactional get cut.
The proactive layer: what un-cuttable agencies do differently
We watched our own client book over the last 18 months and pulled apart what the un-cut agencies were doing differently. The pattern is consistent. We call it the proactive layer.
The proactive layer has three jobs.
Job one: surface signal before the client has to ask
The competitor's pricing change shouldn't reach the CMO through her sales team. It should reach her through the agency's morning brief, with context, two days before the sales team notices. This is the watch tower function. It's also the function that justifies the retainer.
You build the proactive layer by making signal-watching a system, not a strategist's habit. The signals to watch are knowable: competitor pricing pages, key SERP movements, paid spend velocity in your category, audience signals from social listening, search trend changes. None of these are opinions. They're observable. The strategist's job is curation and framing, not observation.
Job two: pre-translate the math into the buyer's language
The agency report should arrive in two dialects: operational (for the marketing team) and financial (for when the CFO walks in). Same numbers. Different framing.
The CMO who can hand her CFO the financial-language version without re-doing the work is the CMO who keeps her seat at the executive table. The agency that ships in two dialects by default makes itself indispensable to that CMO. Eliminating it would mean she goes back to translating by hand.
Job three: turn meetings into decisions, not reports
Every standing call should walk in with two or three pre-curated decisions waiting on the CMO's accept/decline. Not "here's what we did last week." Not "here's what we plan to do." Specifically: "Here's the thing we want to ship. Here's why. Here's what it costs. Here's what it'll return. Yes or no."
This is the Approve flow. It's the difference between a meeting that recaps and a meeting that decides. The CMO who walks out of every standing call having made three forward decisions is the CMO whose agency stays.
What this looks like in product form
You can run the proactive layer manually. We did, for years. It eats hours. The strategist becomes a translator and a watch tower simultaneously, and the strategist's hourly rate gets consumed by work that should be a system.
Or you build it into a surface. Next Best Action is the surface we built around our own proactive layer.
The Status surface is the watch tower. Signals show up in the daily brief before the CMO has to ask. The Ask surface lets the CMO question the data and the strategy in her own language, with citations, with the strategist reviewing high-stakes answers before they hit her. The Approve surface turns standing calls into decision moments. Each card carries the recommendation, the cost, the expected return, and a one-tap accept or decline.
Three surfaces. One job each. The strategist's hours move from translation and watch-tower duty to judgment work. The CMO's experience of the agency moves from mystery box to partner. The 39% becomes the 0% for the agencies that ship the proactive layer in product form.
"An agency that produces decks gets cut. An agency that produces daily insight you didn't know to ask for gets renewed."
The renewal math
Mid-market agency retainer median is $7,500 per month, or $90,000 per year. The cost of building the proactive layer manually, in strategist hours, is roughly 8-12 hours per month per client. At a $200 strategist rate, that's $1,600-$2,400 per month, or 21-32% of the retainer.
Most agencies don't run this math out loud. They feel the cost as "our strategists are burned out" and they cope by either not doing the proactive layer (and getting cut) or doing it and watching margin erode (and laying off strategists).
The product version of the proactive layer collapses the cost. The watch tower runs 24/7. The translation happens automatically. The Approve cards generate from agent flagging plus strategist curation. Strategist hours per client drop to 3-5 per month, mostly on judgment and review. Margin recovers. Renewal rate goes up because the surface itself is the deliverable the CMO can't replace.
Software-wrapped service businesses get acquired at 3-8x revenue. Pure services get 0.8-1.5x. We wrote about that math here. The renewal rate gain stacks on top of the multiple gain. Both compound.
The honest read
If you're an agency owner, the 39% stat is not abstract. It's the conversation your VP of Account Management is going to have on three of your top accounts in the next 90 days. The accounts that don't churn will be the ones where the proactive layer is built, the surface is shared, and the CMO's relationship with you feels like a partnership instead of a vendor relationship.
If you're a marketing leader, the 39% stat is the implicit pressure that's been building in every QBR. You're not unique in feeling like your agency is opaque. You're not unfair for considering a cut. The fix is to require the proactive layer as a condition of the next renewal, or to find an agency that already ships it.
Either way, the next 18 months separate the agencies that ship the proactive layer in product form from the agencies that don't. The 39% is the gravity that does the separation.