Agency Sales: A Governed System From Market to Handoff

Define the agency commercial model

An agency sells a promise of future work performed by a specific team under constraints. That makes the offer different from a repeatable software SKU. The buyer must evaluate whether the agency understands the situation, can staff the work, can produce acceptable evidence, and can control changes after the agreement begins.

Start by naming the commercial model:

Hybrid offers are valid, but every component still needs its own unit, owner, evidence, and acceptance rule. Do not present “strategy plus execution” as one undifferentiated promise. A buyer cannot compare it, and delivery cannot govern it.

Choose the market and buying situation

The U.S. Small Business Administration’s current market-research guidance separates demand, market size, location, saturation, and pricing, and recommends examining competitors by service and market segment. Apply that discipline to an agency service line before building an account list.

Write one market record with the service, buyer, triggering situation, current alternative, decision constraints, geography, expected delivery capability, and clear exclusions. Separate facts from hypotheses. “New marketing leader” may be an observable event; “will replace the incumbent agency” is an inference until the buyer confirms it.

A useful ideal-client definition answers four questions: can the buyer use the service, can the agency produce the required evidence, can both sides support the delivery dependencies, and is the engagement economically viable under a realistic capacity plan? The account-selection guide owns scoring mechanics. This page owns the connection between market choice and the service promise.

Qualify for evidence and delivery fit

Qualification should protect delivery capacity, not merely confirm interest. Record the problem, current state, desired result, affected stakeholders, decision process, timing driver, budget authority, required proof, procurement path, security or legal constraints, buyer inputs, and disqualifying conditions.

Use evidence states rather than optimistic checkboxes:

• confirmed: a named buyer or governed artifact supports the assertion;
• inferred: plausible but not yet confirmed;
• contradicted: credible evidence conflicts with the assertion;
• missing: the decision cannot yet be made;
• expired: the assertion is too old for the current decision.

Do not invent a universal qualification score. Define hard stops for work the agency cannot legally, ethically, technically, or economically deliver. A seller can escalate a pricing, privacy, intellectual-property, or contracting question; they should not improvise an answer outside their authority.

Build stages with buyer-verifiable exits

HubSpot describes pipelines as stages representing steps in a process and allows administrators to create and govern them. Salesforce likewise treats opportunity stages as the basis for pipeline movement and stage analysis. Those products do not define the agency’s process; the agency must define stage-entry and stage-exit evidence.

HubSpot’s stage-calculated properties and Salesforce’s stage-analysis documentation show how configured systems can track entry, exit, time, progression, and conversion. Preserve stage history and test reopened, skipped, backdated, merged, and owner-transferred deals. The agency sales-process guide covers deeper implementation.

Connect discovery, proposal, and agreement

Discovery should produce a decision packet, not a transcript dump. Capture the buyer’s current state, desired result, baseline, constraints, stakeholder views, required evidence, dependencies, risks, and open questions. Label each important statement with source, speaker, date, and confidence.

Translate the packet into a proposal containing situation, objectives, scope, exclusions, approach, named responsibilities, milestones, deliverables, acceptance, assumptions, dependencies, commercial terms, change process, and signature path. AIGA’s design-services agreement reference uses a modular structure and identifies proposal elements such as objectives, process, milestones, fees, expenses, work schedule, billing schedule, and signatures. It also explicitly recommends qualified legal advice rather than treating the model as universal.

Walk the proposal with the buyer, but do not treat presentation behavior as acceptance. Capture each requested change, owner, deadline, commercial effect, and approval. Contract language, intellectual property, employment classification, privacy, tax, and regulated claims belong to qualified owners.

Price from capacity, dependencies, and risk

Price from a capacity and risk model, not a copied industry range. One transparent project estimate is:

minimum viable fee = direct labor + outside cost + allocated delivery overhead + risk reserve + required contribution.

Suppose a fictional engagement has $42,000 direct labor, $8,000 outside cost, $10,000 allocated overhead, a $6,000 risk reserve, and a $14,000 required contribution. The minimum viable fee is $80,000. That is arithmetic, not a market benchmark or recommended margin.

Retainers also need a demand model: included capacity, priority rights, service levels, rollover, buyer inputs, overage or change rules, and termination effects. Outcome-linked fees require a frozen baseline, observable metric, data authority, attribution rule, excluded events, dispute path, and ceiling or floor. Use the agency-pricing guide for model design, then obtain finance and legal approval.

Manage acquisition as a source portfolio

Manage acquisition as a portfolio of governed sources: referrals, partner introductions, account-based outreach, published expertise, events, directories, paid demand, and formal procurement. Do not call one channel “best” without a comparable cohort.

For each source, record source event, cost basis, eligible opportunity rule, first meaningful interaction, influenced versus sourced status, and final outcome. Suppress people and accounts when consent, contract, policy, or jurisdiction requires it. Never turn a trigger event into a claim about the buyer’s private motive.

Run one controlled improvement at a time. Compare the same service, segment, eligibility rule, and observation window, and keep a holdout or baseline where practical. The agency-prospecting guide owns channel execution; the follow-up system owns response-state handling.

Measure cohorts without universal benchmarks

Publish definitions beside metrics. At minimum, review accepted opportunities, stage progression, stage dwell, win/loss by mature cohort, signed value, delivery-capacity commitment, forecast error, source cost, scope changes, and handoff defects. Counts must accompany percentages.

• stage progression = eligible deals advancing from stage ÷ eligible deals entering stage;
• mature-cohort win rate = won deals ÷ all won and lost deals in the frozen cohort;
• forecast error = absolute forecast minus actual ÷ actual, with a declared zero-actual rule;
• handoff defect rate = accepted handoffs with a defined defect ÷ all accepted handoffs.

Do not infer causality from a small before-and-after chart. Segment by service, source, size, region, and period; preserve exclusions; and distinguish projected fee, signed fee, invoiced amount, collected amount, and recognized revenue. The CRM should support decisions, not rewrite finance definitions.

Run the agency sales operating system

Run a weekly opportunity review for stage evidence and blockers, a monthly source-and-capacity review, and a quarterly service-line review. Every meeting should end with decisions, owners, deadlines, and a change log—not a second set of undocumented CRM notes.

Use this printable control record:

A mature agency-sales system is reproducible and reversible. Another qualified person can inspect why an opportunity advanced, what was promised, who approved it, whether delivery can accept it, and how the decision would be corrected. That control is more valuable than an unsupported claim that one script, price, or channel always wins.