The cold email agency RFP scorecard: compare the operating system, not the pitch.

Scope and ownership: 20 points

1. What does the agency own from account selection to calendar booking? Ask for the complete sequence and every handoff.
2. What must our team provide every week? Hidden client workload changes the true cost.
3. Who handles interested replies, objections, qualification, and scheduling? A lead list or mailbox alert is not appointment setting.
4. What remains ours after the engagement? Clarify data, domains, inboxes, copy, CRM history, and reporting access.

Targeting and data: 20 points

1. Show the written account, buyer, geography, use-case, and exclusion model. “We use Sales Navigator” is not a targeting strategy.
2. How are signals verified and used? Public context can support relevance, but it should not be presented as private buyer intent.
3. How are customers, active opportunities, competitors, and sensitive accounts suppressed?
4. How does the system prevent the same account from receiving conflicting campaigns?

Infrastructure and deliverability: 15 points

1. Are campaign domains and inboxes separated from the primary corporate domain?
2. Who owns authentication, warmup decisions, sending limits, bounce handling, and provider monitoring?
3. What conditions pause sending automatically or trigger manual review?

Qualification and sales handoff: 15 points

1. Define a qualified meeting in contract language. Role, company, use case, geography, exclusions, and expressed interest should be explicit.
2. Show what context the AE receives before the meeting.
3. How are no-shows, timing objections, and relevant-but-not-ready accounts handled?

Proof and measurement: 15 points

1. Can each result be traced to a named client, period, metric definition, and source?
2. Does reporting separate replies, interested replies, qualified conversations, meetings booked, meetings held, opportunities, and pipeline?
3. Can we inspect live operations rather than wait for a monthly slide deck?

Governance and incentives: 15 points

1. What behavior does the pricing model reward? Per-meeting pricing can reward loose qualification. Pure retainers can reward activity without outcomes. The control system matters more than the label.
2. Who approves targeting, claims, exclusions, and material campaign changes?
3. What are the stop, correction, and exit procedures?

Test the agency's judgment before you sign

A written ICP can look precise while two teams still select different accounts. Before treating a proposal as operationally ready, ask each shortlisted agency to work through the same small, anonymized account-and-reply sample. This is a proposed buyer evaluation exercise, not a statistically validated benchmark or a free service included in an agency engagement. Agree on the scope and any fee first.

1. Build the answer key internally. Have sales and RevOps agree which examples are eligible, excluded, or unresolved under your current rules. Include good-fit accounts, obvious exclusions, and difficult cases. Remove personal details and confidential customer information before sharing. Use fictional examples where real records cannot be shared.
2. Give vendors the rules, not the answer key. Provide the same product scope, account criteria, exclusion list, and evidence for every vendor. Ask for an independent decision, the evidence behind it, missing information, and the next action. Do not let a polished presentation substitute for a completed exercise.
3. Review disagreements together. Count hard exclusions accepted, eligible accounts rejected, and unknowns incorrectly treated as facts. Show the counts and sample size, not just one accuracy score. Your internal answer key may also need correction: resolve disagreements against the written rules, not seniority.
4. Repeat only the disputed cases after clarification. Check whether the revised rule changes the decision. Record the version, approver, and escalation owner before live outreach. A small exercise tests shared judgment; it does not establish future meeting volume or sales performance.

Include cases a meeting quota can hide

• Relevant enterprise champion, budget unknown: distinguish an exploratory problem-validation call from a funded evaluation. Do not invent purchasing authority.
• Target account with an open opportunity: route to the existing account owner before outreach, even if the contact looks ideal.
• Enthusiastic reply from an excluded business: record the exclusion instead of accepting it merely because the person will take a call.
• “Send information” from a good-fit buyer: answer the request; do not report an information request as a booked or held meeting.

These are illustrative cases, not customer results. Use the qualified meeting criteria and handoff template to write your own acceptance rules.

If the agency runs alongside your internal SDR team

Assign account ownership before contact, including subsidiaries and shared buying committees. Keep one suppression process for customers, active opportunities, and opt-outs. Agree who handles replies when a buyer crosses territories, and preserve the original source rather than letting the last scheduling link take all the credit. A new agency meeting with an existing opportunity is not automatically new pipeline.

Use this exercise as evidence for the existing targeting, qualification, and governance questions above. Do not add bonus points to the 100-point workbook. If an unresolved hard exclusion or ownership conflict remains, settle it before launch regardless of the total.

Interpret the total

How Snipe should be scored

Snipe publishes the operating sequence, written scope, client-specific case studies, qualification approach, and limitations. The service covers approved targeting, prospect research, cold-email copy, dedicated sending infrastructure, reply handling, qualification, and calendar booking. Clients own sales calls, proposals, and closing.

Those are two specific engagements. They are evidence of prior work, not forecasts for another company. Use the same standard with every agency: named scope, precise definitions, and no guarantee disguised as proof.

Pair this scorecard with the qualified-demo cost calculator so each vendor is compared on the same meeting stage and cost basis.