How to Run Multi-Round Company Outreach Without Repeating Yourself

Who this playbook is for

This playbook is for teams that may revisit an eligible B2B company after one bounded outreach period ends without a human conversation. It is appropriate when the company still fits the active ICP and a materially different, useful reason for contact exists. It is not a method for bypassing unsubscribe, company-stop, existing-customer, suppression, or contact-invalidity controls. It also does not define how many rounds any company should receive.

Contents

1 Define a round at company level 2 Pass the next-round eligibility gate 3 Write a material change brief 4 Reuse contacts by default 5 Replace contacts only from evidence 6 Control round count and fatigue 7 Close the learning loop 8 Illustrative example 9 Common mistakes 10 What this does not mean 11 Common questions 12 Sources and methodology

1. Define a round at company level

A round is one bounded company-level attempt to test a defined outreach hypothesis during a configured follow-up window. It is not one email, one person, or one sequence step. A round can involve several eligible contacts, but all actions share a company scope, purpose, offer, learning objective, owner, and stop-state register. Use a stable record: Field Meaning Company and round ID The account scope and ordinal round Active ICP version The criteria that made the company eligible Application hypothesis The problem/use case being tested Contact pool Eligible people, status, and role evidence Window Opening, expiry, and configured duration Learning objective What uncertainty this round should reduce Material message basis What useful reason supports contact now Stop controls Company, person, address, channel, and campaign states Owner Who can open, pause, close, or approve an exception The company-level definition prevents a common failure: separate team members treating contact A and contact B as unrelated campaigns even though both belong to one account. Company fatigue and contradictory claims can accumulate across people and channels.

2. Pass the next-round eligibility gate

Under the SellersGPT product rule, a company may enter another round only when it has no active lead and no human interaction that should govern the next action. This is a hard routing condition, not a weighted factor. Then test the remaining gates: 1 The company remains inside the active ICP and current application scope. 2 No Existing Customer state blocks outreach and billing. 3 No Suppression or company-stop state blocks outreach. 4 Each proposed contact and channel remains eligible. 5 The prior round is closed with outcomes and learning recorded. 6 The configured maximum round count has not been reached. 7 A material, evidence-backed change supports another attempt. Write pass, fail, or unknown for each gate. Unknown is not a weak pass. If a critical control is unknown, hold the round until the owner resolves it. A scoring system must never average away active-lead status, human interaction, or a stop state.

3. Write a material change brief

Before opening round two or later, complete the SellersGPT Round Change Brief: Question Required answer What did the prior round test? One falsifiable account/contact/message hypothesis What was observed? Delivered, bounced, replied, referred, visited, unknown - kept separate What was learned? Evidence and confidence, not a story about silence What changed? Account trigger, application, proof, role rationale, timing, format, or offer Why is the change useful? Buyer-relevant reason, not internal desire for activity What remains unchanged? Facts and claims that must stay consistent What would stop this round? Explicit control states and human events What decision is due at expiry? Advance, another eligible round, or close Material change can come from a verified facility expansion, a newly relevant application, improved technical evidence, a contact referral, a changed procurement period, or a better-defined role. Replacing "checking in" with "following up" is not material. Changing a subject line while keeping the same claim, proof, ask, and contact logic is not material. The GAO Green Book describes principles for control activities, information, communication, and monitoring in U.S. federal internal control. It is not an outreach manual. It is useful here as a primary control reference for why repeatable work needs objectives, quality information, documented responsibilities, and monitoring rather than unrecorded operator memory.

4. Reuse contacts by default

SellersGPT reuses the existing contact pool across rounds unless evidence supports a change. This preserves company learning and prevents a system from constantly discovering new people simply because earlier contacts did not reply. Contact-pool change is appropriate for defined cases such as:

• role or application mismatch;

• contact cannot be reached through the governed channel;

• unsubscribe or another applicable contact-level stop;

• hard bounce or invalid address;

• person left the company or retired;

• wrong-contact confirmation;

• company-stop signal;

• authorized manual or workflow skip;

• follow-up evidence shows mismatch;

• configured timeout triggers replacement review. These states do not all mean the same thing. A hard bounce invalidates an address route. A wrong-contact reply can provide useful role evidence. An unsubscribe controls further messaging at the applicable scope. A company stop may end all account outreach. A departed or retired person must never re-enter a new SellersGPT round. The product default maximum is five contacts per company. That is a control ceiling, not an instruction to contact five people. An account may need one eligible contact, or it may be ineligible altogether.

5. Replace contacts only from evidence

When replacement is permitted, carry account learning forward. The replacement record should include:

• predecessor contact and status;

• exact replacement trigger;

• evidence source and date;

• account/application hypothesis retained or changed;

• role sought and why;

• stop states that must propagate;

• messages, claims, and proof already used;

• next owner and review date. Never reset round history because the person changed. The company has already experienced prior contact. The new person should not receive a message pretending no prior account work occurred if that would be misleading, nor should private remarks from another person be repeated without an appropriate purpose and boundary.

6. Control round count and fatigue

SellersGPT defaults to three rounds and permits an administrator-configured range from 1 to 100. The upper bound exists for configuration flexibility; it is not a benchmark, target, or recommendation. A campaign can close after one round even when three are allowed. Use a round continuation gate, not a quota: Continue = eligibility gates pass AND material change exists AND useful action exists AND owner approves If any hard term is false, the output is close or hold, not "continue with a lower score." Consider accumulated company exposure across contacts and channels. Google's email sender guidelines and Yahoo's sender best practices place importance on authentication, complaints, unsubscribe, and responsible sender behavior for relevant senders. They do not certify a number of rounds. Provider requirements and reputation signals are constraints to monitor, not evidence of buyer interest. Applicable communication rules also remain in force across rounds. The FTC CAN-SPAM guide and CRTC CASL FAQ should be reviewed for their actual scope; neither allows a team to erase prior requests by creating a new campaign label.

7. Close the learning loop

At round expiry, write a short, evidence-bounded conclusion:

• What observable events occurred?

• Which assumptions were supported, contradicted, or left unknown?

• Did any human interaction create a new owner or next step?

• Which contacts remain eligible, changed, or stopped?

• Did the company/application fit change?

• What material change could justify another round?

• Is the best decision to close? The NIST Cybersecurity Framework 2.0 is designed for cybersecurity risk, not sales. Its Govern and continuous-improvement concepts illustrate a broader discipline: policies, roles, information, and feedback should be reviewed as conditions change. In this playbook, the analogy is limited to control design. It does not turn sales activity into a security process.

8. Illustrative example

Illustrative example - fictional. VectorSeal supplies fictional food-processing plants. Round one at Harbor Canning tested a broad maintenance-cost hypothesis with two verified operations contacts. Both messages delivered; there was no reply, human interaction, active lead, click relied upon, or stop signal. The team learned only that the channel delivered. Silence did not prove lack of need. Before round two, a first-party facility notice shows a new high-temperature line. VectorSeal has a documented material compatibility guide relevant to that application. The Round Change Brief records three proposed deltas: application specificity, evidence offered, and role rationale. It keeps the same two contacts because neither is invalid or mismatched. The team audits five required delta fields:

• verified changed context: complete;

• changed application hypothesis: complete;

• new useful evidence: complete;

• differentiated ask: complete;

• stop-state review: complete. material-delta completeness = 5 / 5 = 100% This result means the brief is complete. It does not mean the buyer will respond. During round two, one contact refers the team to process engineering. That is human interaction, so the company exits ordinary round automation. A human owner reviews the referral and the next action. Round three never opens, even though the product default allows three rounds.

9. Common mistakes

• Calling a resend a new round without changing the decision hypothesis.

• Treating silence as proof that the offer, company, or contact is wrong.

• Replacing the entire contact pool after every non-response.

• Ignoring prior unsubscribe or company-stop states under a new campaign name.

• Opening a new round while a human referral or active lead is unresolved.

• Running the configured maximum because capacity exists.

• Counting contacts rather than company-level exposure.

• Failing to close a round and record what remains unknown.

10. What this does not mean

Multi-round outreach is not a promise that persistence produces results. It is not permission to contact every role, use every channel, or run 100 rounds. It does not turn public account news into personal intent. It does not make a replaced contact a fresh account. The purpose of a round structure is to constrain execution and preserve learning. In some cases the highest-quality decision is to close after the first attempt.

11. Common questions

How different must the next round be? Different enough to change the buyer-relevant reason for contact or the hypothesis being tested. Cosmetic copy changes are insufficient; document the evidence and expected learning. Can we use the same contacts in all three default rounds? Yes, if they remain relevant and eligible and no human interaction or stop state changes the path. Contact replacement is evidence-triggered, not mandatory rotation. Does no reply mean there was no human interaction? Usually no reply means human interaction is unknown, not absent in a metaphysical sense. Operationally, use only observable governed events. Do not infer intent from silence. When should a company close before the configured maximum? Close when fit fails, a stop state applies, no useful material change exists, the contact pool is exhausted or inappropriate, or the expected learning no longer justifies another controlled attempt.

12. Sources and methodology

The official sources constrain communication, sender operation, and control design; they do not prescribe SellersGPT round counts. The three-round default, 1-100 configurable range, no-active-lead/no-human-interaction gate, contact reuse rule, five-contact default, Round Change Brief, and formulas are SellersGPT rules or editorial tools as labeled. 1 U.S. Federal Trade Commission, CAN-SPAM Act: A Compliance Guide for Business. https://www.ftc.gov/business-guidance/resources/can-spam-act-compliance-guide-business. Accessed August 25, 2026. 2 Canadian Radio-television and Telecommunications Commission, Frequently Asked Questions about Canada's Anti-Spam Legislation. https://crtc.gc.ca/eng/com500/faq500.htm. Accessed August 25, 2026. 3 Google, Email sender guidelines. https://support.google.com/a/answer/81126. Accessed August 25, 2026. 4 Yahoo, Sender Best Practices. https://senders.yahooinc.com/best-practices/. Accessed August 25, 2026. 5 U.S. Government Accountability Office, Standards for Internal Control in the Federal Government (Green Book). https://www.gao.gov/greenbook. Accessed August 25, 2026. 6 National Institute of Standards and Technology, Cybersecurity Framework 2.0. https://www.nist.gov/cyberframework. Accessed August 25, 2026.