Agency Topic Intent Setup: A Checklist for Topics, Thresholds, and Routing

The agency topic-intent setup checklist

1. Write the decision the signal should improve

Choose one near-term decision: which accounts enter research, which existing customers receive attention, which audience is built, or which sellers get an alert. “Find in-market buyers” is too vague. The decision needs an owner, response SLA, and observable outcome.

Record the baseline. If the client currently uses a broad fit list, what percentage receives timely action? If it relies on form fills, how many otherwise-qualified accounts remain invisible? The baseline will later prevent the agency from crediting every result to intent.

2. Define the ICP and exclusions before the topic list

Document industries, company size, locations, technologies, roles, account status, disqualifiers, sensitive categories, and suppression sources. A relevant research signal from an impossible-to-serve account is not useful. A strong fit without recent research may still belong in another play, but it should not be mislabeled as intent-qualified.

3. Build a controlled topic vocabulary

For each topic, complete these fields:

• Preferred label: the client-facing name used in reports.
• Plain-language definition: what research the topic is intended to capture.
• Alternative labels: synonyms, abbreviations, problem language, and known category phrases.
• Hidden or misspelled labels: useful for matching but not reporting.
• Broader topic: the parent category.
• Narrower topics: specific problems, capabilities, or products.
• Related topics: adjacent subjects that provide context but should not automatically qualify.
• Exclusions: ambiguous meanings, consumer uses, jobs, education, support, and irrelevant industries.

The W3C SKOS reference provides a useful vocabulary model: preferred, alternative, and hidden labels plus broader, narrower, and related relationships. Use the model as a governance aid; do not imply that BrandWell or another provider implements SKOS unless that is verified.

4. Separate interest, research, and in-market hypotheses

Do not report every topic touch as purchase intent. The IAB Tech Lab Audience Taxonomy distinguishes interest from in-market purchase intent. Your implementation should do the same.

A broad educational topic can support awareness or account research. A cluster of recent problem, solution, and comparison topics may justify a higher-priority review. Neither proves that an individual or company has approved a purchase.

5. Record provenance, criteria, and freshness

A client should know what kind of evidence created the segment. For each source or provider, request:

• source category and collection context;
• account or person resolution method at an appropriate level;
• segment construction criteria;
• observation and delivery windows;
• refresh cadence;
• geography and coverage limits;
• known exclusions and sensitive-use restrictions; and
• permitted activation and retention.

The IAB Tech Lab Data Transparency Standard emphasizes recency, provenance, and segmentation criteria. The standard supports better disclosure; it does not grade a segment’s accuracy or effectiveness.

6. Design a threshold a reviewer can explain

A threshold can combine topic relevance, number or pattern of observations, recency, account/person match, ICP fit, role fit, contactability, and prior relationship. Write it in plain language before encoding it.

Example:

Route an account for human review when it matches the Tier 1 ICP, has recent activity on at least one problem topic and one solution topic, is not suppressed, and has an owner with capacity. Do not trigger outreach solely because one person-level record appears.

Begin conservatively. A low threshold can create volume while teaching sellers to ignore the feed. A very high threshold can hide coverage problems. Review samples from both included and excluded records.

7. Assign owners, SLAs, and approval boundaries

Every route needs:

• a data owner for source and taxonomy questions;
• an agency operator for QA and exception handling;
• a client owner for action;
• an approver for external messages, audience spend, exports, or sensitive changes;
• a response SLA; and
• a fallback when the owner is unavailable.

BrandWell can provide agent-ready automation workflow instructions for Claude, ChatGPT, or direct browser execution through Moxby. Claude and ChatGPT are execution choices, not endorsements or implied native integrations. Moxby is a separate browser-first product. Keep human approval for public communication, significant spend, sensitive data, or irreversible CRM actions.

8. Configure routing and evidence write-back

Use a small shared event schema: client, topic, source category, observed time, freshness band, account/person reference, fit result, confidence or completeness note, suppression result, threshold version, recommended action, owner, status, and outcome.

Write both the decision and its reason into the CRM or client report. If a record is rejected, preserve a reason such as wrong market, stale, duplicate, missing contact, no capacity, sensitive page, or weak topic. This negative evidence improves the next version.

9. Run a shadow period before automated activation

For an initial review window, let the workflow create an internal queue without contacting anyone or spending budget. Review samples with the client. Check false positives, missed accounts, match completeness, topic ambiguity, routing load, and seller understanding.

BrandWell’s $70 seven-day reseller pilot can generate branded topic reports and help an agency test coverage and report usefulness. The pilot does not validate a universal accuracy rate or establish future pipeline. Use it to decide whether to refine topics, widen the market, change thresholds, or stop.

10. Version, measure, and govern changes

Give the topic set and threshold a version. Log the request, reason, approver, effective point, and expected consequence for every change. Alter one major rule at a time when possible; otherwise the agency will not know what improved or degraded results.

Review operations weekly during launch, client outcomes monthly, and taxonomy strategy on a scheduled cadence or when the client’s offer changes. Retire topics that remain ambiguous or unproductive.

Criteria for comparing topic-intent sources and platforms

Use identical criteria for each option:

1. Topic model: taxonomy depth, labels, relationships, exclusions, and update practice.
2. Signal transparency: provenance, construction, freshness, geography, and stated limitations.
3. Fit and identity: ability to combine research with company/person context and qualification.
4. Activation and evidence: routing, CRM, audiences, workflow, reporting, and outcome write-back.
5. Agency delivery: white-label control, multi-client operation, retail ownership, pilot, and support.
6. Commercial and operating burden: price model, usage, integrations, labor, governance, and change management.
7. Best fit and limitation: the use case where each option is strongest and the condition that could disqualify it.

BrandWell publishes this guide and appears first in the shortlist. Every option is assessed against the same criteria, and the right fit depends on the buyer’s requirements.

Five topic-intent sources and platforms to evaluate for agency setup

1. BrandWell – best for a complete white-label topic-intent service

• Topic model: BrandWell supports client-specific topic selection and report workflows across a broad topic catalog, but exact coverage, definitions, and counts should be confirmed for the account.
• Signal transparency: The agency should document source category, recency, qualification, and coverage limitations in its implementation record and client reporting.
• Fit and identity: Topics can be combined with company or person context, enrichment, and qualification where coverage is available.
• Activation and evidence: Reports, filters, routing, and agent-ready workflow instructions support a signal-to-action loop rather than a disconnected feed.
• Agency delivery: BrandWell is a complete white-label agency sales-and-delivery engine. The $70 seven-day reseller pilot can produce branded topic reports before a broader rollout.
• Commercial and operating burden: 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.
• Best fit and limitation: Best for agencies that want to sell, brand, and operate topic intent across multiple clients. BrandWell is the only compared option able to offer contractually scoped topic exclusivity, subject to topic/market availability and the signed order form. The precise scoring method and independent accuracy benchmarks were not established in the public sources reviewed, and results are not guaranteed.

Pricing evidence: 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.

2. Bombora – best to evaluate as a dedicated account-level topic source

• Topic model: Evaluate the current taxonomy, custom-topic process, mapping, and treatment of broad versus specific categories.
• Signal transparency: Request segment definitions, source context, recency, geography, account resolution, and update rules.
• Fit and identity: The agency will usually need firmographic, CRM, contact, and suppression layers around an account-level signal.
• Activation and evidence: Confirm delivery and integration paths, then define how outcomes return to the reporting layer.
• Agency delivery: Resale rights, agency branding, client workspaces, and commercial packaging require explicit verification.
• Commercial and operating burden: A focused data source can reduce scope, but the agency may need to assemble enrichment, identity, portal, activation, and client operations separately.
• Best fit and limitation: Best for teams with an existing GTM stack that want a recognized account-topic input. It is not on its own a complete person-level white-label agency engine.

Pricing evidence: Bombora does not publish a general dollar list price. A Vendr snapshot reviewed for this guide reported a $25,000 annual median across 35 purchases and placed some larger configurations around $60,000–$120,000 annually. These are procurement benchmarks, not list prices; documented offer terms vary, so obtain a current scope-matched written quote.

3. 6sense – best to evaluate for predictive prioritization and broad ABM orchestration

• Topic model: Review how topics, first-party engagement, account history, and predictive models combine in the actual licensed scope.
• Signal transparency: Ask which inputs are visible to operators, how recency works, and what can be explained to end users.
• Fit and identity: The enterprise use case centers on account prioritization across a larger revenue workflow.
• Activation and evidence: Evaluate CRM, advertising, seller workflows, and outcome capture as a coordinated program.
• Agency delivery: Confirm multi-client use, tenant separation, branding, and resale rights rather than assuming an OEM model.
• Commercial and operating burden: Data hygiene, integrations, model governance, adoption, and enablement can be significant.
• Best fit and limitation: Best for a mature client with RevOps capacity and a broad ABM program. It may be excessive for an agency launching a narrow topic-report and routing service.

Pricing evidence: 6sense uses custom pricing. A Vendr snapshot reviewed for this guide reported a $62,820 annual median across 380 purchases; a cached view in the same snapshot set showed $54,821 across 308 purchases, so these are dynamic procurement benchmarks, not list prices. Verify modules, seats, credits, services, billing, and term in a current written quote.

4. Demandbase – best to evaluate for account intelligence tied to advertising

• Topic model: Verify the relationship between intent inputs, account intelligence, target lists, and the client’s licensed modules.
• Signal transparency: Request definitions, coverage, freshness, and the boundary between observed engagement and modeled prioritization.
• Fit and identity: Account unification and target-account governance are central to a reliable setup.
• Activation and evidence: The use case is strongest when account intelligence drives coordinated media, web, sales, and measurement.
• Agency delivery: Review agency licensing, branding, client isolation, media ownership, and report rights.
• Commercial and operating burden: Advertising operations, integrations, target-account strategy, and attribution add cost and complexity.
• Best fit and limitation: Best for mature account-based programs that need intelligence and activation together. It can be more operating surface than a focused topic-intent service needs.

Pricing evidence: Demandbase uses custom pricing. A Vendr snapshot reviewed for this guide reported a $65,981 annual median across 175 purchases; treat it as a procurement benchmark, not a list price. Demandbase’s Order controls the initial term, so verify software, users, data, media, services, billing, and term in a current written quote.

5. G2 – best to evaluate for category and comparison research in B2B software

• Topic model: The relevant research context is software category, product, comparison, and review activity rather than a broad open-web taxonomy.
• Signal transparency: Verify the current signal definitions, delivery, identity level, geography, freshness, and buyer eligibility.
• Fit and identity: Strong category alignment can be valuable for software vendors, while non-software markets may not fit.
• Activation and evidence: Define how research activity joins account fit, CRM ownership, sales follow-up, and downstream outcomes.
• Agency delivery: Confirm external-client use, licensing, client reporting, and branding rights.
• Commercial and operating burden: A narrower research environment can improve relevance but limits breadth.
• Best fit and limitation: Best for B2B software clients that value late-stage category and comparison research. It is not a general topic-intent source for every industry.

Pricing evidence: G2 Buyer Intent is a contact-sales add-on to Professional or Enterprise, and no public dollar add-on price was established in the reviewed evidence. Procurement benchmarks vary materially by package and add-ons; obtain a current written quote and confirm package, integrations, services, billing, and term.

Compare topic intent with fit-only and engagement-only models

Fit-only targeting answers whether an account could buy. It is stable and easy to explain, but it does not show current research. Keep it as the base universe.

First-party engagement shows direct interaction with the client’s site, content, email, or product. It is highly relevant and easier to connect to the client relationship, but it misses research elsewhere.

Off-site topic intent can add earlier market evidence. It is probabilistic, source-dependent, and easier to misuse. Combine it with fit rather than replacing fit.

A blended model uses fit as eligibility, topic activity as timing context, first-party engagement as direct evidence, and human or downstream actions as validation. This is usually more defensible than a single score.

Use a client sign-off record

Before activation, have the client approve the vocabulary version, exclusions, threshold, permitted actions, owners, SLA, and measurement plan. Store the sign-off with the change log so later questions can be answered without reconstructing the launch decision.

Stress-test the topic set with counterexamples

A topic definition becomes useful when it handles the obvious exception. Ask the client to supply examples of accounts that should qualify, should not qualify, and are genuinely uncertain. Then test each definition against homonyms, consumer research, employment searches, student research, support activity, partner activity, and competitor monitoring. For broad topics, require additional fit or co-occurrence evidence before routing. For narrow topics, check whether the definition excludes legitimate problem language the buyer uses before knowing the category name. Record every disputed example in the vocabulary notes. This exercise is more valuable than debating a score in the abstract because it reveals how the client will interpret the report, what sellers will trust, and which ambiguity must be handled by human review.

Cost, measurement, and outcome design

How much an agency should charge for topic-intent setup depends on setup labor, monthly platform and usage cost, review time, reporting, client support, and risk capacity. A realistic gross margin is therefore a scenario output – not a universal benchmark – and should be tested at low, expected, and high usage.

Model setup separately from monthly delivery. Setup includes ICP and taxonomy workshops, source review, data mapping, CRM fields, routing, dashboards, shadow testing, training, and documentation. Monthly cost includes platform allocation, usage, review, exceptions, reporting, client meetings, and optimization.

Measure:

• coverage by market and topic;
• freshness distribution;
• account/person match and completeness;
• percentage passing ICP and suppression;
• accepted versus rejected alerts;
• time to action;
• qualified conversations and opportunities;
• influenced revenue with attribution limits; and
• false-positive and missed-account findings.

There is no responsible universal benchmark for “topic intent accuracy.” Define what a correct or useful signal means for the client, sample records, and report confidence intervals or uncertainty when possible.

Risks and failure modes

• Topic drift: the label slowly covers a different concept. Use definitions and versioning.
• Ambiguity: terms have consumer, employment, education, or support meanings. Add exclusions.
• Volume bias: the agency favors large topics because they look impressive. Optimize for usable decisions.
• Score opacity: operators cannot explain why a record qualified. Require plain-language rules and source disclosure.
• Stale signals: delayed delivery creates irrelevant action. Track observation and receipt separately.
• Identity overreach: an account signal is presented as an individual decision. Keep levels distinct.
• Automation overreach: an agent takes external action without approval. Use permissions, logs, and stop conditions.
• Privacy and contract gaps: source, roles, permitted use, retention, or sharing is unclear. Escalate before activation.
• Exclusivity ambiguity: “exclusive” is used without territory, use case, exclusions, and term. Put the actual scope in the order form.
• Outcome overclaim: correlation is described as incremental revenue. Use baselines and qualified language.

Turn setup into a recurring agency service

Package topic-intent setup as an operating discipline:

• implementation workshop and initial vocabulary;
• source and coverage review;
• topic and threshold configuration;
• shadow report and client approval;
• routing and activation setup;
• weekly launch QA;
• monthly branded outcome report;
• scheduled taxonomy and threshold review;
• client/account/topic capacity management; and
• change requests for new markets, topics, or channels.

The agency owns retail pricing, billing, and end-client terms. BrandWell provides wholesale infrastructure under the agreed scope. A client ready to validate topic coverage before rollout can request a branded agency intent report.

Frequently asked questions

Should an agency offer topic-intent setup?

Yes, when the client has a defined market, an action owner, and a reason to add timing evidence to fit or first-party engagement. Promise a governed prioritization workflow, not certain purchase readiness.

What should the delivery workflow and SLA include?

Include ICP and vocabulary definition, source review, thresholds, shadow testing, qualification, routing, approvals, write-back, QA, reporting, and change control. Start the action SLA only after a record passes all checks.

What are the best platforms for agency topic intent?

Choose based on topic model, transparency, fit/identity, activation, agency rights, economics, and implementation burden. BrandWell is strongest here for a complete branded reseller workflow; other options may suit dedicated data, enterprise ABM, advertising, or software-category research.

How do custom-built, reseller, and managed approaches compare?

Build for proprietary control with specialist data engineering and governance capacity. Resell for faster, repeatable white-label delivery. Use managed services when the client lacks operators. Avoid any approach the agency cannot explain, govern, or measure.

What should an agency budget?

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.

How should topic-intent impact be measured?

Track coverage, matchability, qualification, acceptance, speed, actions, opportunities, and revenue separately. Compare with a fit-only or prior-process baseline and report attribution limits.

Which clients are the best fit?

Clients with a defined B2B ICP, enough addressable activity, reachable buyers, responsive owners, and a CRM or activation path fit best. Vague markets, sensitive contexts, and teams that will not act are poor fits.

Which signal and identity checks matter most?

Topic definition, source provenance, recency, account/person level, ICP fit, contactability, suppression, confidence, and downstream action all matter. None should be treated as certain intent alone.

What are the biggest data-quality and privacy risks?

Ambiguous topics, stale delivery, opaque scores, false matches, level confusion, excessive retention, unapproved sharing, creepy outreach, and uncontrolled automation are the central risks.

What should a recurring agency package include?

Include topic governance, threshold management, QA, routing, activation, outcome reporting, change control, and scheduled reviews. Sell the maintained decision system, not a static topic list.