A diagnosis from inside the industry: what agencies are commonly paid to produce, what clients need, and where the accountability gap appears.
Recommendation Marketing is emerging because buyers do not experience marketing as a collection of channels. They experience one business and decide whether it is understandable, credible, relevant, and worth recommending. That raises an uncomfortable question for my own industry: what happens when the agency is accountable for completing the work, but nobody is accountable for whether the business becomes easier to choose?
I have founded and run agencies for more than thirty years. I know the value agencies create. I also know how often a team can deliver everything it promised while the client still feels the outcome slipping away. This is not an accusation from outside the room. It is a diagnosis from inside it.
TL;DR
- The traditional agency model is not one model, and it is already changing. But many agency relationships still attach payment and reporting most directly to hours, scope, deliverables, and channel outputs.
- The accountability gap appears in four places: output can become the optimization target; reporting favors what the agency controls; attribution remains difficult; and channel ownership fragments a buyer decision that is experienced as one whole evaluation.
- AI-assisted buying did not create those problems. It makes a limited part of evaluation more observable by showing how businesses appear, are represented, and are supported in answers to defined buyer questions.
- Recommendation Marketing provides the emerging outcome discipline: organize the work around whether the business becomes easier to understand, verify, and recommend. This article does not yet prescribe the replacement agency contract or promise that recommendation visibility proves causation, trust, or revenue impact.
Start With Fairness
Agencies are not a monolith. Some work on retainers. Some price projects or deliverables. Some use commissions, performance incentives, value-based arrangements, or hybrids. Many are already changing how they work and how they are paid.
The World Federation of Advertisers reported in September 2026 that labor-based payment was the standard model for 19% of the multinational companies in its study, down from 54% in 2011. Fixed-fee or output-based models rose from 20% to 33% over the same period. Performance-based compensation is expected to grow, but across most disciplines, 67% to 80% of respondents said it still accounted for less than 20% of total agency compensation.
Source: World Federation of Advertisers, “Global brands are shifting to paying for outputs not time, WFA research,” September 29, 2026.
That evidence complicates the easy critique. The industry is moving away from a pure hours model. But moving from hours to fixed deliverables does not automatically move accountability to the buyer outcome. It may simply change the unit of production.
For much of marketing’s history, the buyer’s deliberation could rarely be observed directly or comprehensively. Agencies and clients therefore built contracts and reporting around things they could define and inspect: people, time, scope, campaigns, media, content, traffic, inquiries, and other measurable activity. Those measures were not meaningless. They were practical instruments for governing work with many contributors and uncertain causation.
The breakdown begins when the instrument becomes the outcome.
Four Places the Accountability Breaks
1. Output becomes the optimization target
When revenue is attached to hours, scope, or deliverables, completing those units matters. That is not misconduct. It is the rational response to the commercial agreement.
The risk is that the monthly conversation shifts from what the buyer needs in order to understand and choose the business to what the agency can show it completed. Cadences fill. Reports thicken. Everything may be on time while the client remains difficult to explain, verify, or recommend.
2. Reporting favors what the agency controls
Channel metrics describe real behavior on specific surfaces: impressions, rankings, clicks, engagement, traffic, conversions, and leads. They are useful operating instruments. But they do not, by themselves, reconstruct the buyer’s evaluation or prove why a buyer chose one business over another.
An agency naturally reports the surfaces it can operate and inspect. The gap appears when accurate channel reporting is treated as a complete account of the buying decision.
3. Attribution ambiguity weakens accountability
Marketing outcomes rarely belong to one intervention. Product, price, reputation, sales, timing, distribution, competition, customer experience, and multiple channels can all contribute. That makes clean attribution difficult and sometimes impossible.
Ambiguity can protect weak reasoning on both sides. Agencies may claim influence when results rise and narrow their responsibility when results fall. Clients may demand an outcome while withholding the access, consistency, budget, or organizational change required to support it. The problem is not that uncertainty exists. The problem is failing to state what can be known, what is inferred, and who owns the response.
4. Channel ownership fragments a whole decision
A search partner, content team, social agency, PR firm, website vendor, and sales organization may each own a legitimate part of the work. The buyer does not experience those parts as separate contracts. The buyer encounters a business and tries to decide whether it makes sense, can be trusted, and belongs on the shortlist.
Gartner reported in May 2026 that 645 surveyed B2B buyers used an average of seven information sources during a recent purchase. Forty-five percent also used generative AI, primarily to gather information about vendors and products.
Source: Gartner, “Gartner Survey Finds 69% of B2B Buyers Turn to Sales Reps to Validate AI-Generated Insights,” May 20, 2026.
That does not mean an AI system owns the decision. It means the buyer is synthesizing across sources while responsibility often remains divided by channel. Each specialist can improve its surface while nobody owns whether the total business is coherent enough to understand, credible enough to verify, and relevant enough to recommend.
What AI Assisted Buying Changes
AI-assisted buying does not reveal the buyer’s whole decision. It does make one layer of evaluation easier to inspect.
Ask a defined buyer question and you may be able to observe who appears, how businesses are represented, what alternatives are named, and which sources are cited. Freeze the question set, record a baseline, preserve the conditions, and repeat the observation. Recommendation Measurement can then describe whether that answer-level standing changed within the measured frame.
That is partial evidence. It is variable. It is not a causal record, a measure of human trust, or proof that a sale occurred because an answer changed. It is still useful because it gives the business a new observation layer above isolated channel activity.
AI did not break the agency model. It made part of the accountability gap easier to see.
Why Recommendation Marketing Is Emerging
The traditional model begins with the work the organization has divided: search, content, media, social, PR, web, brand, sales enablement. Recommendation Marketing begins with the decision the buyer is trying to make.
That change in starting point matters. The governing question is no longer only whether each channel performed its assigned task. It is whether the business, taken as a whole, is becoming easier to understand, verify, and recommend for the buyer questions that matter.
Recommendation Marketing does not eliminate channels or specialists. It gives them a shared destination. Search can improve discoverability. Content can answer questions. PR can create independent corroboration. The website can establish clarity and proof. Sales can add context and confidence. The work remains specialized, but the outcome is no longer fragmented.
Recommendation Measurement supports that discipline by observing selected signals of standing, appearance, representation, citations, alternatives, and stability, under declared conditions. Those signals do not replace business results. They create a more honest connection between channel activity and the recommendation environment the buyer increasingly encounters.
This is the emerging solution Article 9 can name without pretending the operating model is already settled: marketing needs an outcome discipline above the channels. Recommendation Marketing is that discipline. How an agency should accept, price, and govern responsibility for it is the next question, not this article’s shortcut.
What This Diagnosis Does Not Say
- It does not say channels are dead. Search, content, social, PR, websites, paid media, and sales enablement still perform necessary work.
- It does not say specialists are obsolete. Depth matters. The issue is fragmented ownership of the whole outcome, not specialist expertise.
- It does not say all agencies use the same compensation model. Current research shows meaningful movement toward fixed, output, performance, value-based, and hybrid approaches.
- It does not say agencies act in bad faith. Clients and agencies both operate inside commercial arrangements that shape what gets produced, measured, and discussed.
- It does not say Recommendation Measurement solves attribution. It adds bounded observations of answer-level standing. It does not prove causation or replace financial, sales, brand, customer, or qualitative evidence.
- It does say the work needs a governing discipline above the channels. Recommendation Marketing provides that shared outcome: becoming easier to understand, verify, and recommend.
- And it does not yet say what agency operating or commercial model should carry that discipline. That question deserves its own issue: what would have to be true, in measurement, incentives, authority, and shared responsibility, before an agency could accept greater accountability for results?
Five Questions That Locate Your Position
Whether your marketing runs through agencies, an internal team, or both, write down direct answers to these questions:
- What outcome is each party contractually accountable for—not aspirationally, but contractually?
- Which reported measures are direct outcomes, which are proxies, and which are activities?
- Can your current reporting show how the business appears and is represented when buyers ask its most consequential questions across AI-assisted and human research?
- If that standing changed next month, who would notice, who would investigate, and who would be authorized to coordinate the response across channels?
- Would the current arrangement welcome a governed measurement that could produce an unfavorable result, or would it treat that measurement as a threat to the existing scorecard?
Score it plainly. If contractual accountability stops at activity or delivery, the proxies have not been tested, and nobody owns the cross-channel response, the arrangement may be producing exactly what it was designed to produce. The question is whether that design still matches what the business needs.
A Mirror Not a Window
I have spent my career inside agencies. I have sold retainers, managed teams, built scopes, defended budgets, reported activity, and tried to connect work to results when the causal line was anything but clean. I am not interested in pretending that agencies caused every gap or that a new dashboard removes uncertainty.
I am interested in a harder standard: say what we are accountable for, expose the proxies, preserve the limits, and organize the work around whether the business becomes easier to understand, verify, and recommend. That is the work of Recommendation Marketing.
If you run an agency, tell me where this diagnosis is wrong. If you lead a client organization, tell me where it lets the client off too easily. This is a mirror, not a window, and it becomes more useful when people inside the room challenge it.
Coming next: what would have to be true before an agency could share greater accountability for results—and why most claims of “pay for performance” do not settle the question.
Frequently Asked Questions
Is Recommendation Marketing anti agency?
No. Recommendation Marketing is the emerging discipline agencies, internal teams, and advisors can practice together. It gives specialized channel work a shared outcome: making the business easier to understand, verify, and recommend. The diagnosis concerns accountability, not the worth of agencies or specialists.
Are agencies still mainly paid for hours?
Not universally. WFA research published in September 2026 found labor-based models had declined among the multinationals surveyed, while fixed-fee, output-based, hybrid, value-based, and performance-linked approaches were growing. The remaining question is whether a different payment unit creates accountability for a different outcome.
Do channel metrics still matter?
Yes. They help teams operate and improve particular surfaces. The mistake is treating them as a complete account of the buyer’s integrated decision or as proof of causation.
Can Recommendation Measurement prove that marketing caused a sale?
No. It can observe defined aspects of answer-level standing under declared conditions. Sales causation, human trust, preference, and financial impact require other evidence and may remain only partially attributable.
Does AI assisted buying make agencies less necessary?
Not necessarily. It increases the need for coherent evidence, specialist execution, monitoring, and coordination. It does make fragmented responsibility easier to question when nobody owns how the whole business is understood and recommended.
How does Recommendation Marketing address the breakdown?
It moves the organizing question above the channel level. Instead of treating search, content, media, PR, web, and sales as separate destinations, it coordinates them toward one buyer-centered outcome: becoming easier to understand, verify, and recommend. It does not remove attribution limits, but it makes the intended outcome explicit.
What should replace the traditional agency model?
Recommendation Marketing supplies the outcome discipline, but this article does not yet prescribe the agency operating or commercial model. Any credible model would need clear definitions of the outcome, authority over the variables that affect it, governed measurement, shared obligations, stated limits, and commercial terms that do not reward distortion. That argument belongs to the next gated issue.
B Randall Willis
Founder, Right Angle
Visibility + Engagement = Opportunity
