A busy follow-up queue is not proof that estimates are moving—or that follow-up produced a sale. The useful question is narrower: can the business show which estimates are eligible, who owns the next step, when follow-up must stop, and what evidence supports the final status?
Control: a documented check, rule, owner, or evidence field that helps prevent or reveal a workflow failure.
Drafted, queued, and sent are different events. None should silently become proof that the estimate reached the intended recipient.
Record the estimate-delivery event, channel, time, and approved supporting record separately. If delivery cannot be confirmed to the business's standard, label it unknown instead of starting follow-up by assumption.
“Open” is a workflow status. It does not by itself establish that another message is appropriate.
Have a person review the sending basis, intended recipient, active replies, opt-outs, suppression, acceptance, loss, disqualification, service issues, disputes, wrong-recipient records, and other business-specific stop states. Record who reviewed it, when, and the resulting status. Never infer eligibility from missing data.
A shared inbox or task list can make work visible without making anyone responsible for it.
Give each eligible estimate one accountable human owner before scheduling a next step. That person must be able to correct facts, stop queued messages, route replies, and record the actual decision.
Statuses such as “follow up,” “pending,” or “waiting” do not tell the team what must happen or when.
Record a dated next action with a clear purpose, responsible owner, approved channel, and current message version. If the team is waiting on the homeowner, record the next human review date rather than leaving the estimate unowned.
A sequence can keep running after the underlying facts or decision state have changed.
Treat cadence as a purpose map, not standing authorization. Before each scheduled email, recheck eligibility, replies, suppression, opt-outs, decision status, factual accuracy, approved version, and ownership. Pause when the current state is unknown.
Templates can drift into claims about price, financing, availability, warranties, insurance, scope, or timing that the business has not confirmed.
Tie every draft to an approved fact source and version. Resolve placeholders, confirm the sending identity, and require business approval before use. Automation should not invent facts, interpret a negotiation, or make a new offer.
A reply, opt-out, wrong-recipient notice, dispute, acceptance, loss, or service issue can arrive while later messages remain queued.
Make each stop event cancel the remaining sequence before the record moves to a person. Record the stop time, reason, and new owner. A reply is a handoff event—not permission for automation to infer sentiment or outcome.
Scheduled tasks, sent messages, delivery events, opens, and clicks can make a dashboard look active without proving a reply, decision, job, or payment.
Report activity, response, decision, financial, and attribution evidence as separate layers. A message sent is one event. It is not an accepted job, collected revenue, verified payment, or supported attribution.
A job accepted or paid after follow-up does not, by timing alone, prove that the sequence caused it.
Preserve decision evidence, payment evidence, and attribution evidence as separate records. Use “unknown” when the evidence does not establish cause. Keep costs separate as well; revenue is not contribution or profit.
Using one consistent period, enter aggregate counts for delivered estimates, human-confirmed follow-up-eligible estimates, and eligible estimates with both an accountable owner and dated next action. The result is a documented operational control gap and coverage—not likely revenue, qualified sales, recoverable jobs, replies, payments, or profit.
Check the documented control gap
Calculator entries stay in the browser. They are not transmitted to Arbor, stored, placed in the URL, or carried into the audit-request form. Use aggregate business figures only; do not enter names, addresses, contact details, claim or insurance information, estimate documents, credentials, payment details, or sensitive notes.
Each step needs its own evidence. A later step should never be inferred merely because an earlier one occurred.
Arbor's current one-time founding-pilot offer is $350 for the first three verified pilots. “Verified” means Arbor confirms the business identity, one eligible supported email workflow, an accountable owner, and a written scope; it does not mean verified performance.
The initial audit documents and prepares the workflow; it does not contact homeowners or send messages. Platform, sending, messaging, CRM, advertising, and media-spend fees are excluded. The 14-day measurement review begins only after the business approves and starts operating the workflow. No reply, accepted job, revenue, payment, contribution, attribution, or profit is guaranteed.
A request starts a fit review only. It does not authorize work, sending, platform access, customer-record access, or payment; it does not reserve a pilot place or confirm the $350 price. Availability, price, written scope, delivery window, responsibilities, exclusions, permissions, and written guarantee language are confirmed before payment.
Submit public business details and an aggregate monthly estimate range only. Do not include customer records. Requests are reviewed during U.S. Eastern business hours, and calculator entries are not transferred. Review the full audit scope and purchase boundary.
No. They are operational checks to inspect. The business must approve its own timing, facts, permissions, tools, sending basis, and applicable requirements.
No. It reports an aggregate owner-and-next-action control gap from the figures entered. It is not a forecast, sales qualification, or attribution model.
No. It starts a fit review only. Any scope, payment, access, configuration, or later sending requires separate written confirmation and appropriate business approval.