What Agency Reporting Must Contain, and Where It Stops
Three things a report should let you verify. The third is what the agency was paid, and the industry's own literature explains why that one is missing.
A good agency report lets you verify three things: what was spent, what it produced, and what the agency was paid. Almost every report handles the first two. The third is where reporting reliably stops, and the industry’s own literature explains why.
This is not a claim that your agency is doing something wrong. It is that the standard reporting format was never designed to answer the third question, and the documented practices that make it worth asking sit at a level your contract may not reach.
This page covers what to require, what the transparency literature actually found, and what both sides of that argument said.
The three questions a report should answer
Structure the requirement this way and most format arguments disappear.
What was spent. By platform, by campaign, by period. Reconciled to the platform’s own interface, not retyped into a slide. If you cannot match the report to what the ad account says, the report is a narrative.
What it produced. Not impressions and clicks. Leads that entered your CRM, meetings held, opportunities created. The metric should belong to your business, not to the platform.
What the agency was paid. The fee, the basis on which it was calculated, and any other consideration received in connection with your spend. This is the question that distinguishes a report from a scorecard.
Why the third one matters even in a clean relationship. Because you cannot judge whether a media plan was built for your outcomes or for the supplier’s economics unless you know how the supplier is compensated. That is true of any intermediary and it is not an accusation.
And a fourth item that is not a question but a hygiene requirement. The definitions. Which attribution window, which conversion counting setting, which attribution model, and which days are still provisional. Without them, month-over-month comparisons are not comparisons.
What the transparency study found
The reference document in this field, with its sample and its caveats.
The study. Conducted 20 October 2015 to 31 May 2016, commissioned by the US advertiser association, carried out by an investigations firm.
The sample, verbatim. Based on information compiled from interviews with 150 individual sources, including marketers, media suppliers, ad tech vendors, current and former agency professionals, trade association executives, consultants, attorneys and others, all granted anonymity.
The response rate, which is itself a finding. An additional 131 interviews were requested: 61 declined, 70 failed to respond. Five of the six major agency holding companies declined formal requests to make any current executives available.
Who was actually in scope. Of the 150, 117 were directly involved in US media buying. Of those, 59 reported direct experience of non-transparent practices: 34 on rebates, 33 on principal transactions.
On rebates, verbatim. Cash rebates from media companies were provided to agencies with payments based on the amount spent on media, and advertisers interviewed indicated they did not receive rebates or were unaware of any rebates being returned.
The scale. Of 41 sources reporting that rebate deals occur, 34 reported indications they were not disclosed, not passed through, or demanded by agencies, in amounts ranging from 1.67% to approximately 20% of aggregate media spending.
And the mechanism that hides them. Rebates structured as service agreements, in which suppliers paid agencies for non-media services often tied to spend volume, which sources described as being used to obscure what was essentially a rebate.
Citing one side of a contested study and calling it settled would be exactly the failure this series is supposed to avoid.
The agency association’s response, verbatim. That the immense shortcomings of the report, anonymous, inconclusive, and one-sided, undercut the integrity of its findings.
On process, verbatim. That a healthy and constructive debate about media buying can only happen with a bipartisan, engaged, industry-wide approach, and that this is precisely the opposite of what the advertiser association pursued.
Their central request, verbatim. That the association make available to specific agencies, on a confidential basis, all of the materials related to them, because without an opportunity to assess and address the veracity of the information, sweeping allegations will continue to drive attention-grabbing headlines.
Individual holding company responses. One called it an unfair and unwarranted attack on the entire industry. Another called it disappointing, stating that its outside counsel had requested specific evidence and that none was provided.
How to weigh both sides fairly. The anonymity objection is genuine: anonymous sourcing cannot be independently checked. The refusal to participate is also genuine: five of six holding companies declined to make executives available, which is one reason the sourcing ended up anonymous.
What survives the argument. That practices of this kind exist and are documented in enough detail to be describable. Not that any specific agency engaged in them, which the study never claimed and explicitly disclaimed.
And what that means for you. The right response is not suspicion of your agency. It is a contract that makes the question answerable, so that neither of you has to rely on trust for something that can simply be written down.
What an audit can and cannot reach
The British advertiser association publishes audit guidance, and it is unusually candid about its own limits.
What an audit covers, verbatim. That it is an audit to ensure the agency is complying with the agreed financial and legal terms of the agreement, particularly relating to fees and billings, and that any discounts, agency trading benefits and balances are being returned to the client.
Recommended frequency. On average, annually for larger spending markets, every two years for smaller ones, plus an exit audit at the end of any contract.
On auditor independence, verbatim. That most agency groups now do not allow auditors to perform their services if their remuneration is based on a percentage of recovery, as this questions the auditor’s ability to be impartial and objective.
The limit, stated plainly, verbatim. That access to vendor deal terms will be severely restricted due to confidentiality, as agencies will not share potentially commercially sensitive information directly with advertisers.
And the moving-target problem, verbatim. That persistent barriers have included the use of proprietary or inventory media, restricting data access, the changing form of trading benefits and absence of audit rights, and that as advertisers amended contracts to mitigate risks, agencies changed their trading terms with vendors, created new terminologies for practices and moved the goal-posts.
The structural point underneath all of it. The practices described sit at holding company level. An audit right against the local entity you contracted with cannot see them, which is why the audit-rights dispute between the two associations was specifically about extending audits to the holding company.
One study tried to trace the money end to end under the best possible conditions. Its result sets the realistic ceiling on any reporting requirement.
The setup. Fifteen months of work, more than fifty participating organisations including advertisers, agencies, demand and supply platforms and publishers, all consenting, with an accounting firm as the trusted intermediary.
The matching result, verbatim. From 267 million impressions served from study advertisers to study publishers, 31 million, 12%, were successfully matched. The rest could not be mapped due to low data quality.
The money, verbatim. Publishers received half of advertiser spend. 15% of advertiser spend, the unknown delta, representing around a third of supply chain costs, could not be attributed.
What that establishes. Under the most favourable conditions anyone has assembled, with everyone cooperating and a professional intermediary, 88% of impressions could not be traced end to end.
Why that matters for your report. It bounds what any reporting requirement can honestly promise. A contract clause demanding full traceability of programmatic spend is asking for something that has not been achieved in a controlled study.
What to require instead. Traceability where it is achievable, meaning platform-direct buying reconciled to platform invoices, and disclosure where it is not, meaning a stated declaration of any principal or inventory media in the plan.
And the honest framing for your own reporting, if you are the agency. Say which parts are verifiable and which are not. A supplier who volunteers the limit is more credible than one who implies there is none.
A small, specific correction that comes up constantly.
What it is. Optimization score, presented in many agency reports as a headline number, often with a target attached.
What the platform says it is, verbatim. An estimate of how well your Google Ads account is set to perform, on a scale of 0 to 100%, with 100% meaning the account can perform at its full potential.
What that makes it. A prediction about the account’s configuration, derived from the platform’s own pending recommendations. It is not a measurement of results.
What the platform claims for acting on it. That regularly reviewing and applying recommendations can help improve performance. Not that it will.
The circularity worth noticing. The score rises when you apply the recommendations that the score is calculated from. Moving it is not evidence that anything improved for your business.
Where it belongs. In the diagnostic section, as a list of pending recommendations to accept or decline with reasons. Not as a headline, and never as a contractual target.
Six clauses, none of which is adversarial, all of which make the questions answerable.
An audit right that names the holding company, not only the entity you contracted with. That is the specific gap the association dispute was about.
A stated remuneration model. Fee basis, and a declaration of any other consideration received in connection with your spend.
Disclosure of principal or inventory media. Whether any media is being resold to you rather than bought on your behalf, stated per plan.
Platform access in your own name, so that reported spend can be reconciled to the platform’s own interface without going through anybody.
Reporting definitions on the face of the report. Attribution window, counting setting, model, and provisional period.
And an exit audit, contracted in advance. It is recommended practice, it is far easier to agree at signature than at termination, and it is the moment when it matters most.
A report should let you verify three things: what was spent, what it produced in business terms, and what the agency was paid. The third is usually absent.
The 2016 study interviewed 150 sources; of the 117 directly in US media buying, 59 reported direct experience of non-transparent practices.
34 of 41 sources reporting rebate deals said they were undisclosed, not passed through, or demanded, at 1.67% to approximately 20% of aggregate spend.
Principal transaction markups ranged from approximately 30% to 90%.
Five of six major holding companies declined to make executives available, which is both a finding and the reason the sourcing is anonymous.
The agency association rejected it as anonymous, inconclusive and one-sided, and asked for the materials to respond to. Both objections have force.
Audits cannot reach vendor deal terms. The British guidance says access will be severely restricted due to confidentiality.
And full traceability is not achievable: 12% of impressions matched end to end in a study where everyone cooperated.
Three things: what was spent, what it produced in business terms, and what the agency itself was paid. Most reports do the first two well and stop before the third.
Why is agency remuneration usually missing?
Because the arrangements it would reveal can sit above the contract you signed. The documented practices include rebates from media suppliers and principal transactions at the holding company level rather than the local agency.
How large are those markups?
In the 2016 study, principal transaction markups ranged from approximately 30% to 90%, and rebates ranged from 1.67% to approximately 20% of aggregate media spending.
Is that study reliable?
It interviewed 150 sources over seven months, and 34 of 41 sources who reported rebate arrangements said they were undisclosed, not passed through, or demanded. It is also anonymous, commissioned by the advertiser side, and was rejected by the agency association.
What did the agencies say?
The agency association called it anonymous, inconclusive and one-sided, said those shortcomings undercut the integrity of its findings, and asked that the underlying materials be shared confidentially with named agencies so they could respond.
Can I audit my agency?
You can audit contract compliance if your contract grants the right. The British guidance is explicit that even then, access to vendor deal terms will be severely restricted because agencies will not share commercially sensitive information with advertisers.
Can anyone trace the money end to end?
Barely. A study across fully cooperating participants matched 12% of impressions from advertiser to publisher, and could not attribute 15% of advertiser spend. That was under the most favourable conditions available.
So what do I actually put in the contract?
An audit right that reaches the holding company, not just the local entity; a stated remuneration model; disclosure of any principal or inventory media; and an exit audit at termination.