Media Agency RFP Template: Scope, Questions, and Evaluation Criteria

A media agency RFP template is only useful if it makes agencies comparable. Most templates ranking for this search are dated, member-gated or written for general marketing work, and none ask about modeled conversions, programmatic margin or who will run the account after the pitch. This guide supplies a nine-field structure, scenario-based media agency RFP questions, a mandatory fee grid and an RFP scoring rubric you publish before responses arrive. The Write Direction states the job plainly: every agency answers the same questions in the same order, so the committee compares approach and price (The Write Direction). If you have not yet built a shortlist, GPI's guide on how to choose a paid media agency covers the vetting work that should come before any tender.

GPI’s view. GPI reads an RFP as a measurement instrument pointed at agencies rather than at campaigns, and it should be designed the same way: decide first what business decision the answers must support, then ask only questions whose answers can be checked. A response that describes a method is worth less than one that attaches the method's artifacts and states where it breaks down. The same standard applies to the buyer. If the RFP does not disclose the real data environment, name the creative boundary and publish the weights, it cannot fairly separate a documented methodology from a polished pitch, and the committee ends up scoring presentation skill by default.

When a media agency RFP is the right instrument, and when it is not

RFI, RFP and RFQ: what each document can and cannot decide

An RFI asks agencies to describe capabilities, clients, tools and team so you can build a shortlist; it decides who deserves a full process. An RFP asks for a strategic approach to your stated problem plus commercial terms; it decides who you appoint. An RFQ prices a fixed specification, which suits a defined buy but not a planning relationship where the plan is what you are buying. PriceWeber recommends deciding whether you are hiring for a project or a long-term relationship before picking the document (PriceWeber).

Elapsed times are planning assumptions, not measured averages; definitions follow PriceWeber and Responsive's marketing RFP guide (Responsive).

Spend, timeline and relationship depth as the three go/no-go tests

Spend: if agency fees would be a large fraction of monthly media, a tender costs more than the decision is worth, and an RFI plus two working conversations will serve you better. Decision date: if a plan must be live inside six weeks, there is no room for a Q&A window, written responses, independent scoring and a working session. Whichever instrument you choose, cap the list: Responsive recommends six or fewer vendors so every proposal gets a close read (Responsive).

Why buyers are moving to fewer, deeper media relationships

Digiday reported in 2018 that media buyers were pursuing fewer but deeper relationships as programmatic, Google and Facebook became the workhorses for reach (Digiday).

The media agency RFP template, field by field

Copy this media planning and buying RFP into a working document, complete the buyer columns, and delete any field you cannot fill before issuing; an empty objectives or measurement field means you are not ready to tender. The structure below adapts the BUILD logic The Write Direction applies to web procurement, which baselines what exists and settles integrations and decision criteria before price (The Write Direction).

1. Cover, process rules and confidentiality

The buyer states NDA status and whether a signed NDA is a condition of receipt, names a single point of contact, fixes the Q&A dates, states that contacting other stakeholders is grounds for exclusion, and confirms that no unpaid speculative media plan or creative is required. The agency returns a signed acknowledgment and its own contact. One page.

2. Company and business context

The buyer describes category, markets, revenue band or growth stage, customer segments, the current agency arrangement and, in plain terms, why the account is out to tender. Agencies return nothing here except questions during the Q&A window. Two pages.

3. Objectives and the decision the media plan must support

The buyer names the business decision rather than a media metric: reallocate budget across channels next fiscal year, hold a blended CAC ceiling while doubling new-customer volume, enter two new markets, or replace a bonus paid on platform-attributed conversions. The agency returns its reading of the decision and the two or three metrics it would use to judge progress. One page.

4. Scope of work and channel boundaries

The buyer lists channels in scope, channels explicitly out of scope, geographies, the creative ownership arrangement (three options appear in the scoping section), approval cadence, and the data and account access the agency will receive. The agency returns a scope confirmation and any exceptions it wants to negotiate. Two pages.

5. Measurement, data and tracking environment

The buyer attaches a one-page data fact sheet: current attribution setup, consent mode status, the share of reported conversions that are modeled, any existing MMM or experiment history, first-party data volumes and clean-room access. The agency returns a measurement approach that starts from what exists rather than from what it would like to exist. TrinityP3 advises buyers to require detail from media agencies on proprietary planning tools, clean-room integration and independent trading audits (TrinityP3). Three pages.

6. Capability and approach questionnaire

This field points to the numbered question set in the questionnaire section. The agency answers in the buyer's order, respects the stated length limits, and labels every attachment by question number.

7. Commercial proposal and fee sheet

The buyer inserts the mandatory fee grid and states that free-form pricing sections will not be scored. The agency completes the grid at the three spend scenarios the buyer specifies. The fee sheet section covers the grid design and the total cost calculation.

8. Team, operating model and continuity commitments

The buyer asks for the named day-to-day leads, their allocation percentage, tenure, number of other accounts and the replacement notice period, plus a continuity clause: anyone who presents during the process holds the role shown in the proposed org chart for a stated minimum period after award. The agency returns a named org chart with percentages and confirms the pitch team's post-award roles.

9. Timeline, submission format and disclosed evaluation criteria

The buyer publishes dates, file format, page limits, and, in the same field, the scorecard weights and the questions the committee will ask itself while scoring. iPullRank describes receiving a buyer rubric that disclosed the committee's own internal questions and treats that transparency as a marker of a serious process (iPullRank). The agency confirms compliance.

Scoping the media brief: channels, creative boundaries and the data environment

Drawing the line between media buying and creative production

Undefined creative ownership causes scope disputes because media performance depends on creative volume and refresh rate; when results lag, the buyer blames the buying and the agency blames the assets. Write one of three arrangements into field 4:

1. Agency buys only. The client supplies approved assets in platform-ready specifications by a fixed date each month. The agency is responsible for placement, bidding, pacing and reporting, and is explicitly not accountable for asset-level performance.
2. Agency buys and adapts. The agency may resize, re-cut and re-caption client master assets within brand guidelines and report which adaptations performed. New concepts remain client production.
3. Agency owns performance creative volume. The agency produces a stated number of variants per channel per month, runs a stated testing cadence, and reports creative-level results alongside media results. Approval turnaround on the client side is written in as a reciprocal obligation.

Documenting what tracking you actually have after ATT and consent changes

The data fact sheet should state, in one page: the pixel and server-side setup by platform; consent mode status and the regions it covers; the approximate share of reported conversions that are modeled rather than observed; any MMM, geo-holdout or conversion-lift history with dates; first-party data volumes and where they live; and whether you have clean-room access through a retailer, platform or independent provider.

Clean rooms, modeled conversions and what the agency needs access to

At RFP stage, describe the environment; do not grant access. At finalist stage, under NDA, read-only ad account and analytics access lets shortlisted agencies validate assumptions. State whether clean-room integration is expected in year one; TrinityP3 lists it among the capabilities a media agency RFP should probe (TrinityP3), and an agency without that experience should say so before award.

Budget disclosure: ranges, phasing and what to hold back

Disclose a spend range and the phasing across the first year so agencies can size teams and propose a realistic channel mix. Hold the internal ceiling for finalist negotiation. State that the range is genuine and that the three fee-grid scenarios sit inside it.

The questionnaire: questions that separate diagnosis from checklists

Select no more than 15 to 20 questions from the five categories below, weight the categories in the scorecard before sending, and require numbered answers with attachments. Omni Lab notes that questions such as describe your reporting process are easy for any competent shop to answer well, so they separate agencies with a good proposal writer from agencies with a bad one (Omni Lab).

Planning and diagnostic reasoning

1. Our marketing-qualified leads are up but pipeline is flat. Walk us through your first four weeks.
2. Platform-reported ROAS has risen for two quarters while blended CAC has worsened. What do you check, in what order, and what would make you cut spend?
3. A channel's reported conversions doubled in a month with no change in spend or creative, shortly after a platform modeling update. How do you tell your client what happened?
4. What would you need from us in the first 90 days that most clients fail to provide?

Programmatic supply chain and commercial transparency

1. List every fee that sits between our budget and working media: agency fee, DSP fee, SSP fee, data fees, verification, and any other. State which are disclosed at invoice level.
2. Do you, your holding company or any affiliate buy inventory as principal and resell it to clients? If so, disclose how it is priced and how a client can opt out.
3. Are volume discounts, rebates or free inventory earned on our spend passed through to us? Describe the mechanism and provide the contract clause.
4. Will you accept an independent trading audit and provide log-level data on request?

Measurement, incrementality and modeled conversions

1. Which incrementality methods have you run in the past 24 months (geo holdouts, conversion lift studies, MMM)? Attach one anonymized readout showing design, holdout logic, confidence intervals and what changed as a result.
2. How do you reconcile modeled conversions with observed conversions in monthly reporting?
3. When an experiment contradicts platform attribution, which one do you act on, and what do you tell the client?

These questions matter because platform attribution is a weak proxy for incremental outcomes. GPI's analysis of a 2026 manuscript on historical Meta experiments found that raw seven-day last-click conversions per dollar achieved an out-of-sample R² of 0.19 when treated as a measure of incrementality, against 0.88 for the full experiment-calibrated model, in a sample of 2,226 experiment and conversion-event pairs. The sample is Meta-specific and drawn from large US advertisers, so treat it as evidence about the fragility of last-click rather than a universal constant. For a closer look at why platform-reported conversions make a fragile fee trigger and what to check instead, see GPI's analysis of performance marketing agency measurement and fees.

Search and social execution discipline

1. Describe your default account structure for search and for paid social, and give an example of a client where you deviated from it and why.
2. What creative testing cadence and asset volume do you assume for paid social at our spend range, and what happens to your plan if we cannot supply it?
3. How is automated bidding governed: who can change targets, what guardrails exist, and how are changes logged and reviewed?

Team continuity and operating maturity

1. Name the day-to-day leads for our account, their allocation percentage, their tenure with you, and the number of other accounts each supports.
2. What is your replacement process and notice period if a named lead leaves, and who approves the replacement?
3. Describe your first 90 days on a new account, including what you would stop doing from the incumbent plan and when you would first present an experiment design.
4. Under what circumstances have you resigned an account?

How to format questions so answers are comparable

Number questions, set length limits, request supporting attachments, and reject reordered or merged answers.

The fee sheet: normalizing retainer, percent-of-spend and performance models

Why a mandatory fee grid beats free-form pricing

Require the same fee grid from every bidder. Include the retainer, percent-of-spend rate and base, performance trigger, pass-through fees, minimum term and exit notice. Compare total cost at three spend scenarios before scoring.

Fee triggers: what the performance component is actually paid on

A bonus triggered on platform-reported ROAS pays the agency for attribution noise as much as for results, given how weakly last-click tracks incremental outcomes in the evidence cited in the questionnaire section. An experiment-calibrated or incremental measure is a firmer trigger, though it costs more to administer and produces readings less often. Require each bidder to state the trigger metric and its measurement basis in the grid, and score a last-click trigger below an incremental one regardless of the percentage attached.

Hidden margin: tech fees, principal buys and unrebated discounts

Require disclosure of retained DSP or platform margins, and a yes or no on whether volume discounts and rebates are returned. TrinityP3's guidance that media-agency tenders should require independent trading audits (TrinityP3) exists because these layers are invisible without one.

Modeling total cost at three spend levels

The table below is a hypothetical illustration, not a measured or quoted result.

TrinityP3 suggests weighting commercial transparency at 35 percent when pricing and efficiency are paramount (TrinityP3).

Weighted scorecard: how to score strategic rigor, transparency and team maturity

Setting weights before the RFP goes out, and publishing them

Fix weights and anchors before distribution, publish them in field 9, score independently, then reconcile in committee with a written rationale. The starter ranges below are GPI editorial recommendations, not measured norms; move weight toward commercial transparency when efficiency is the priority, and toward measurement methodology when you are replacing an attribution-dependent bonus or entering a market with no baseline. Publishing your agency RFP evaluation criteria and the committee's questions improves answer alignment; iPullRank's account of receiving such a rubric shows agencies read it as a marker of a serious buyer (iPullRank).

Scoring anchors: what a 1, 3 and 5 look like for each criterion

The 35 percent upper bound on commercial transparency follows TrinityP3's example for efficiency-led media tenders (TrinityP3).

Scoring boutique specialists against holding-company networks

Specialists offer senior attention and depth in a few channels; networks offer buying scale, tooling and breadth. The consolidation Digiday described (Digiday) narrows the scale advantage for many advertisers without removing it for large multi-market budgets.

Chemistry meetings with the day-to-day team, not the pitch team

Replace presentation-heavy finals with a working session run by the named day-to-day leads on a real but bounded brief, with no deliverable produced for free and no expectation of a finished plan. Anyone in the room who is not on the org chart from field 8 should be identified as such.

Reconciling committee scores and documenting the decision

Score independently first. Then discuss only the variances above a set threshold, for example any criterion where two scorers differ by two points or more. Keep the final scores and rationale for bidder feedback and your next review.

High-stakes fields: guardrails, failure modes and how to check the finished RFP

NDA, data access and account permissions before distribution

If the RFP contains spend, CAC targets, margin data or platform account structure, get NDAs signed before distribution, not with it. PriceWeber places the NDA first in its transparent-process sequence for the same reason (PriceWeber). At RFP stage, agencies get the data fact sheet, not access. At finalist stage, grant read-only ad account and analytics access with a stated end date and revoke it on award.

Speculative work: what to ask for and what to pay for

Strategic approach and worked scenario answers are fair to request. A bespoke media plan with channel budgets, or creative concepts, is not, unless you pay for it. PriceWeber reports that unpaid spec work is an automatic deal killer with good agencies, who generally pull out rather than risk giving their work away, and notes that many brands now offer a nominal fee for high-value pitch work (PriceWeber). Decide your policy before issue and write it into field 1.

Simultaneous distribution, Q&A discipline and incumbent handling

Run the process in this order:

1. Sign NDAs and confirm the shortlist.
2. Distribute the RFP to every bidder on the same day.
3. Open a single written Q&A window; circulate every question and answer to all bidders without attribution.
4. Receive written responses and score independently.
5. Reconcile scores, shortlist finalists, and grant read-only access.
6. Hold working sessions with the named day-to-day teams.
7. Award, and give written feedback to the others.

Decide in advance whether the incumbent competes. If it does, it receives the same document and answers the questionnaire as if new, without referencing account history the other bidders cannot see.

A pre-issue checklist for evaluating your own RFP

• The business decision is named in one sentence.
• The data environment fact sheet is attached.
• The creative boundary uses one of the three arrangements.
• The fee grid is mandatory and the three spend scenarios are stated.
• Questions are numbered, limited and paired with attachment requests.
• Weights and committee questions are published in field 9.
• The committee is named and has agreed to score independently.
• The timeline allows a Q&A window and a working session.
• NDAs precede distribution.
• No free deliverable is requested at any stage.

The cold-reader test: hand the draft to one colleague unfamiliar with the project and ask whether, using only the scorecard, they could tell two agencies' answers apart. If they cannot, the questions are too generic or the anchors too vague.

Where GPI's methodology fits in your RFP evaluation

The template in this article works for one reason: it forces identical structure, attached evidence and pre-published weights, so the committee scores documented methodology rather than presentation skill. That is the same standard Growth Partner Index applies when it evaluates agencies for its directory. The public Confidence Score methodology scores agencies on documented public evidence across seven pillars, including proof of measurable outcomes, creative-media integration and operating maturity, and buyers can lift that evidence standard directly into their scorecard anchors. GPI does not run campaigns, pitch consulting or agency searches; it publishes methodology and directory criteria that buyers can apply, as described on its about and ownership disclosure page. To align your scorecard anchors with a published evidence standard, review the Growth Partner Index methodology and adapt its criteria to the weights you set in this template.

Frequently asked questions about media agency RFPs

How many agencies should receive a media agency RFP, and how do you build the shortlist before issuing it?

Keep the list to six or fewer, in line with Responsive's recommendation, so the committee can read every response closely (Responsive). Build the shortlist from an RFI or from documented evidence such as public case work, methodology disclosures and reference checks rather than from inbound sales interest, and record why each agency made the list so you can defend the exclusions later.

Should the budget be disclosed in a media agency RFP?

Disclose a range and the phasing across the first year so agencies can size teams and propose realistic channel mixes. Hold the internal ceiling for finalist negotiation, but state that the range is real; agencies that suspect a hidden number tend to price defensively or decline.

What should you do if a strong agency declines to participate?

Ask why before moving on. If the reason is a request for unpaid bespoke work, an unrealistic timeline or a process that appears predetermined toward an incumbent, treat it as a review of your own document. PriceWeber's observation that good agencies withdraw rather than give work away, and that some brands now pay a nominal fee for finalist work (PriceWeber), suggests the fix is usually on the buyer's side.

How can a committee evaluate incrementality claims when the company has never run its own lift tests?

Ask each agency to attach one anonymized experiment readout with design, holdout logic, confidence intervals and the decision it changed. Score the presence and quality of the artifact, not the headline lift number, and treat claims supported only by platform-reported ROAS as weak given the gap between last-click and incremental outcomes documented in the questionnaire section.

Should the incumbent agency compete in the RFP?

Only if you would genuinely re-appoint it. If so, give it the same document, the same Q&A and the same scoring, and instruct it to answer as if new rather than referencing account history the other bidders cannot see. If you have already decided to leave, tell the incumbent directly and do not include it as a benchmark.

How do you compare proposals that quote different fee models and different spend assumptions?

Require every bidder to complete the same fee grid at the same three spend scenarios you specify, then compare total cost of engagement rather than headline rate. Where a performance component exists, compare the trigger metric and its measurement basis before comparing the percentage; a bonus paid on platform-attributed conversions and one paid on experiment-measured lift are not the same price, even when the numbers match.