Buyer Intent Data Service Proposal and Scope of Work Template

Who this is for

• Agency owners productizing buyer intent, audience, prospecting, paid-media, or RevOps services.
• Client services and operations leaders who need a reusable intent data proposal template.
• GTM consultants translating topic research, visitor identification, enrichment, and activation into deliverables.
• Buyers reviewing an intent-data managed service and trying to separate useful scope from attractive promises.

This guide is a commercial and operational template, not legal advice. Adapt it with counsel for the data sources, channels, clients, and jurisdictions involved. Intent and identity are probabilistic. A well-written SOW cannot guarantee that a company will buy, reply, enter pipeline, or produce incremental revenue.

The five-part proposal framework

The strongest proposal can be summarized in five numbered decisions. Put these before detailed legal language so an executive buyer can understand what is being purchased.

1. Define the business decision

Name the decision the service will improve. Examples include which accounts enter weekly seller research, which topics influence a content plan, which qualified accounts enter an ad audience, or which existing customers receive an expansion-risk review. “Provide intent data” is not an outcome.

State the baseline and the decision owner. If the current process is a monthly cold list, record list size, acceptance rate, time to action, and downstream results. If there is no baseline, the first delivery period should establish one rather than promising a lift that cannot be measured.

2. Define an eligible signal

Specify the signal source class, topic or behavior, observed time, geography, account or person identity level, freshness window, and minimum fit requirements. Keep observation separate from interpretation. Research activity around a topic is evidence of activity, not proof of budget, authority, timing, or purchase readiness.

The proposal should say what happens to weak, stale, duplicated, suppressed, or unresolved records. An exception queue is a deliverable. Silently dropping difficult records makes the service look cleaner than it is and prevents the agency from improving it.

3. Define the action and approval boundary

For each accepted signal, state the permitted next action: report inclusion, CRM task, account research, audience creation, nurture branch, approved outreach draft, or another bounded play. Name who approves a new audience, message, budget, field write, or external communication.

Low-consequence, reversible work may be automated. Person-level outreach, new claims, budget changes, sensitive-data use, and external publishing should stay behind explicit human approval. The boundary should apply whether a team executes a workflow manually, with Claude or ChatGPT, or in the browser through Moxby.

4. Define evidence and acceptance

Attach acceptance criteria to every deliverable. A report can be accepted when required fields reconcile to the source, reason codes are present, duplicate rules pass, and the client owner can take the named action. An audience can be accepted when eligibility, suppression, destination policy, and matched-size checks pass.

Do not let “number of signals delivered” stand alone. Record usable records, accepted records, completed actions, time to action, exceptions, outcome evidence, and the method used to connect activity with pipeline or revenue.

5. Define commercials and change control

Separate the wholesale data/platform cost, setup, direct agency labor, client support, connected-tool cost, media, and risk reserve. Then price a defined allowance for topics, clients, records, reports, reviews, destinations, and revisions. Overages and custom work need an approval step before the work begins.

This is the margin-protection layer. “Reasonable requests included” is not a scope. A request is either included, an exchange for another included item, or a priced change.

Copy-and-adapt intent data scope of work template

Use this structure as the drafting spine for an agency proposal. Replace brackets with verified client-specific terms.

1. Purpose and outcome

Purpose: Agency will operate a buyer-intent service to help [client team] make [named decision] for [market/use case]. The target operational outcome is [measurable action], not a guaranteed sales or revenue result.

Baseline: During [baseline window], client currently produces [volume], accepts [rate or count], acts within [time], and records [outcome fields]. If a reliable baseline is unavailable, the first period establishes it.

2. Protected and covered scope

• Market, territory, and client entity: [define].
• Topics/categories/competitors/products/services: [list or attach taxonomy].
• Use case: [seller prioritization, audience, reporting, customer risk, content, or other].
• Exclusions and suppression: [industries, accounts, people, geographies, sensitive contexts].
• Term and availability: [start/end, hold, activation, renewal, release].
• Delivery surface: [standard portal, API-only, custom portal, export, dashboard, or combination].

If topic protection is part of the offer, the signed order must state the precise topic, territory, use case, exclusions, and term. BrandWell’s current terms describe Protected Topics as availability-dependent and scoped in writing; that is materially different from a claim of universal exclusivity.

3. Inputs and prerequisites

Client supplies a current ICP, account exclusions, suppression lists, approved domains, destination access, field definitions, campaign or seller owners, brand assets, approval contacts, and lawful-use instructions. Agency supplies the topic map, event schema, qualification rules, report or portal configuration, play cards, QA plan, and operating calendar.

No production activation begins until required access, permissions, data roles, and approvals are documented. Missing prerequisites extend the schedule under the change process rather than forcing the agency to absorb recovery work.

4. Signal and qualification rules

Define required fields such as client ID, event ID, source, observed time, topic, company domain, geography, identity level, confidence, fit fields, freshness, consent or use flags, suppression status, and reason code. Define accept, reject, hold, investigate, and expire states.

Version qualification rules. A threshold change must state the owner, rationale, effective time, records affected, rollback method, and evaluation window. An AI agent may recommend a change but should not rewrite and deploy eligibility rules without review.

5. Deliverables and cadence

List each artifact, format, owner, due time, acceptance check, and revision allowance. Possible deliverables include:

• weekly qualified-account report with source timestamps and reason codes;
• branded topic report for client presentation;
• portal workspace and named client users;
• CRM research tasks or approved field updates;
• audience file or destination sync after approval;
• exception and suppression report;
• monthly decision review and quarterly commercial review;
• agent-ready workflow instructions with inputs, allowed tools, output schema, evidence, and escalation conditions.

Avoid the phrase “real time” unless the contract defines latency from observation to availability and from availability to action.

6. Roles and service levels

The service level must include dependencies. The agency cannot meet an action deadline when the client approver or destination access is unavailable. Define pause, escalation, and expiry behavior.

7. Measurement and evidence

Report four layers: time-to-value, data quality, adoption, and business outcome. Label each outcome as descriptive, platform-attributed, modeled, or experimental. When an incrementality test is feasible, define the hypothesis, treatment, comparison group, duration, conversion lag, decision threshold, and limitations before launch.

The FTC’s substantiation policy is a useful reminder that objective and implied performance claims need a reasonable basis before they are made. Put that discipline into proposals, case studies, and client reports.

8. Data use, security, and compliance

Attach a data-flow diagram and responsibility schedule. Address permitted purpose, roles, instructions, confidentiality, security, subprocessors, retention/deletion, access, audit evidence, individual rights and opt-outs, incident handling, and end-of-service return or deletion.

The UK ICO’s controller/processor contract checklist is jurisdiction-specific but useful for spotting missing contract fields. The NIST supplier due-diligence guide can inform the security appendix. Use appropriate counsel rather than treating either source as a compliance certificate.

For U.S. email, the FTC says CAN-SPAM applies to B2B commercial messages and responsibility cannot simply be contracted away. Put sender identity, subject-line approval, postal address, opt-out, suppression ownership, and monitoring in the workflow.

9. Fees, usage, and change orders

State setup fee, recurring fee, included modules, topic and client capacity, usage allowance, agency review hours, reports, destinations, support window, taxes, overages, third-party spend, and payment terms. Exclude custom creative, CRM cleanup, bespoke integrations, legal review, media, and unlimited revisions unless priced.

Use a change request containing: request, reason, dependency, impact on fee, impact on schedule, risk, acceptance criteria, approvers, and effective date. Work begins after written approval.

10. Pilot, renewal, and exit

Define pilot inputs, branded output, expiry, evaluation meeting, success criteria, data disposition, and the separate acceptance that creates paid service. Define renewal notice, review date, topic continuation or release, export/deletion, destination disconnection, credential revocation, and transition support.

The most useful proposal tools and resources

A reusable proposal system needs five artifacts, not an oversized slide deck:

1. Discovery worksheet: outcome, baseline, market, topics, sources, destinations, owners, data restrictions, and disqualifiers.
2. Scope configurator: modules, capacity, usage, cadence, service levels, dependencies, exclusions, and price impact.
3. Signal dictionary: every field, source, timestamp, meaning, confidence, retention, and permitted use.
4. Play-card library: trigger, qualification, evidence, proposed action, owner, approval, expiry, and outcome fields.
5. Evidence ledger: delivery, adoption, exceptions, costs, outcomes, attribution method, and claim-support status.

Government performance-work-statement guidance offers a useful design analogue. FAR 37.602 emphasizes results and measurable standards rather than dictating labor methods, while FAR 8.405-2 identifies fields such as work description, location, period, deliverables, standards, and special requirements. Private agency engagements are not automatically governed by those rules, but the structure is practical.

Compare four delivery models before you scope

Do not choose a model because it sounds advanced. Choose the least complex model that can move the client’s decision and return evidence. A report-only engagement can be better than automation when ownership is unclear. A white-label model becomes attractive when the agency can reuse topic setup, qualification, reporting, and client onboarding without mixing client data.

Price for contribution margin, not software markup

Model monthly delivery cost as:

platform and usage + direct labor + integration maintenance + client support + media or tools + risk reserve

Then calculate contribution margin:

(client recurring fee − monthly delivery cost) ÷ client recurring fee

Setup should fund market mapping, taxonomy, data and access review, portal or report configuration, play design, destination setup, QA, baseline, and training. Recurring fees should fund enabled capacity, monitoring, reports, optimization, and the agreed support window.

Margin protection comes from operational controls: record or audience bands, topic and client limits, revision caps, named destinations, exception allowances, response windows, and priced changes. Track manual minutes per accepted action. A low platform fee paired with endless data repair is not an affordable service.

For package tiers and retail pricing mechanics, keep this SOW focused and refer readers to BrandWell’s guides to buyer intent service package design and what agencies should charge.

Metrics that show whether the scope is working

Use a balanced scorecard:

• Delivery: on-time outputs, schema completeness, duplicate rate, source latency, destination success, exception backlog.
• Quality: accepted-signal rate, sampled false positives, identity/enrichment yield, suppression accuracy, freshness distribution.
• Adoption: records reviewed, actions accepted, action completion, time to action, report/portal usage, rejection reasons.
• Outcome: qualified responses, meetings, opportunities, pipeline, won revenue, retained revenue, or media outcomes – with denominator and evidence method.
• Economics: setup recovery, platform utilization, direct labor, support burden, contribution margin, expansion revenue, and churn risk.

A proposal improves revenue quality when fewer unqualified records consume time and more eligible signals receive the intended action. It improves profitability when that happens within the agreed labor and capacity model.

Best-fit clients and disqualifiers

The best fit has a defined market, meaningful economic value per opportunity, observable topic or website behavior, a reachable owner, a timely action, destination readiness, and enough volume to learn. It also has permission to use the data for the planned purpose and a CRM or evidence process capable of recording outcomes.

Pause or narrow the scope when the client requires a promise of a fixed lead outcome, has no ICP, cannot provide suppressions, lacks an action owner, expects automatic person-level outreach, treats identity as certain, wants sensitive consequential use, or cannot agree on a baseline. A small manual discovery project is safer than a broad recurring promise.

Signal, identity, activation, and outcome must stay separate

Build a chain of evidence:

1. Observed signal: what happened, where, and when.
2. Fit: why the organization belongs in the client’s market.
3. Identity: what account or person could be resolved and with what confidence.
4. Freshness: whether the evidence is still relevant to the planned action.
5. Activation: what the agency or client actually did.
6. Outcome: what happened later and how it was measured.

The SOW should never allow a downstream report to erase gaps in that chain. “Pipeline influenced” should point to the signal cohort, accepted action, opportunity record, time window, and attribution rule.

Where BrandWell fits in the proposal

BrandWell agency plans are $2,500–$5,000 per month, depending on topic count, contract term, and any contractually scoped topic exclusivity that is available. Confirm included modules, usage, client capacity, implementation, support, and exclusivity in the current written quote and order form.

Contractually scoped topic exclusivity may be available, subject to topic and market availability, use case, exclusions, term, and the Order Form. Treat that as a written commercial boundary, not a universal ownership claim.

When offered, a $70 seven-day reseller pilot lets an agency generate branded topic reports before accepting a paid order. BrandWell also delivers agent-ready workflow instructions that can be carried out with Claude, ChatGPT, or directly in the browser through Moxby. Claude and ChatGPT are execution choices rather than endorsements or implied native integrations; Moxby is a separate browser-first product. Every instruction should still name allowed tools, forbidden actions, output format, evidence, and human approvals.

These capabilities reduce packaging and delivery work only when the scope is standardized. They do not guarantee coverage, identity accuracy, pipeline, revenue, ad performance, or compliance.

Final scope review before signature

Before an agency sends the proposal, ask ten questions:

1. Is the client buying a named decision and action rather than a feed?
2. Are signal, identity, freshness, fit, and suppression rules explicit?
3. Does every deliverable have an owner, cadence, acceptance check, and revision limit?
4. Are automated, reviewed, and prohibited actions separated?
5. Are delivery dependencies and client delays reflected in service levels?
6. Are costs, included capacity, overages, and custom work visible?
7. Can the agency measure adoption, outcomes, and gross margin with denominators?
8. Are privacy, security, permitted use, retention, and opt-out duties assigned?
9. Does the pilot have a defined end and a separate paid-conversion act?
10. Can both sides explain renewal, topic release, data exit, and transition?

If any answer is “we will figure it out later,” convert it into a discovery deliverable, an assumption, or an exclusion. That is how an intent data SOW protects client trust and agency margin at the same time.

Frequently asked questions

How should an agency approach an intent-service proposal?

Start with the client decision, baseline, and named action. Then define eligible signals, approval boundaries, evidence, costs, and change control. The proposal should make value understandable while the SOW makes delivery testable. Avoid selling an undefined quantity of “high-intent leads.”

What inputs and review cadence are required?

Require the ICP, market and topic map, exclusions, suppressions, destinations, field dictionary, owners, permitted uses, baseline, and outcome fields. Review exceptions and delivery weekly during launch, decision quality monthly, and commercial scope on a scheduled quarterly or renewal cadence. Change frequency according to signal expiry and client capacity.

Which template or system is most useful?

Use a linked set: discovery worksheet, scope configurator, signal dictionary, play cards, and evidence ledger. A generic proposal template can format the document, but these operational artifacts prevent ambiguity. Keep the accepted Order Form, data-flow diagram, responsibility schedule, and rule versions with the engagement record.

How do report, managed, client-owned, and white-label approaches compare?

Report-only delivery is simplest but relies on the client to act. Managed delivery improves follow-through but adds exception labor. Client-owned platforms maximize control and implementation burden. A white-label service can be most repeatable for an agency, provided tenant separation, capacities, permissions, support, and customization limits are explicit.

What cost drivers belong in an intent-data proposal?

Include platform and usage, implementation, topic/client capacity, identity and enrichment, destinations, direct labor, support, media, third-party tools, security or legal work, and a risk reserve. Price custom integrations, CRM cleanup, creative, data remediation, extra revisions, and faster service windows separately.

Which KPIs show revenue quality and profitability?

Track usable and accepted signal rate, time to action, action completion, exceptions, qualified responses and opportunities, pipeline or revenue with the evidence method stated, direct labor, utilization, support time, and contribution margin. Always include denominators and distinguish attributed from incremental results.

Which clients and contract types fit best?

The best fit has a defined market, sufficient economic value, observable behavior, an owner who can act quickly, and systems that record outcomes. A recurring order works when the workflow repeats. Use paid discovery or a bounded pilot when the topics, volume, action, permissions, or baseline remain uncertain.

Which signal and identity evidence matters most?

Preserve source, timestamp, topic or behavior, account resolution, person resolution if permitted, confidence, fit, freshness, suppression status, action, and outcome. Do not merge account-level research with person-level identity. The SOW should state which level is required for each activation.

What risks most often damage margin or trust?

Unlimited revisions, vague “real-time” promises, missing client dependencies, bad CRM data, unsupported match-rate expectations, automatic outreach, unclear data roles, unmanaged overages, platform-policy violations, and inflated ROI claims are common failures. Convert each risk into a prerequisite, acceptance check, cap, approval, or exclusion.

How should the SOW change for a recurring white-label service?

Add client and topic capacity, tenant separation, brand and portal responsibilities, report cadence, wholesale-versus-retail responsibility, standardized onboarding, included support, renewal and topic-release mechanics, and cross-client data controls. Keep the core service standardized and route unusual client requests through change control.