Skip to content
← All resources

Founders, sales leads, and agency account owners

Run a weekly outreach review you can trust

Explain what was recorded, what is still unknown, and the one next test worth reviewing.

Guide dated September 30, 2026. Your actual report has its own evidence time and scope.

The work to complete

  1. Lock the scope and dates

    Write the Workspace, Seat, campaign filter, exact start/end dates, timezone, and report execution time. Compare like scopes. Campaign cohorts can overlap, so adding every campaign's reply total can double-count people.

    Keep this artifact

    A scope header that another authorized reviewer can reproduce.

  2. Separate messages, people, and meetings

    Sent counts qualifying outbound message rows. Contacted and replied count distinct people in the report cohort. Meetings counts bookings in the window. Reply rate divides replied people by contacted people; meeting rate divides contacted people with a meeting by contacted people. Those rates count replies or meetings after the first qualifying send in the window. Off-platform meetings can count as bookings without counting in the contact conversion rate.

    Keep this artifact

    Counts with their units and denominators. A booked meeting is not automatically a qualified Klose or a sale.

  3. Keep missing evidence visible

    Sent means recorded provider-accepted sending, not inbox placement. Delivery and bounce values depend on explicit outbound provider evidence. When sending exists but no delivery outcome is tracked, the value is unavailable, not zero. Tracking coverage is the share of sent messages with a recorded delivered-or-bounced outcome; it measures evidence coverage, not inbox placement. Reply and meeting rates remain unavailable below five contacted people.

    Keep this artifact

    A short explanation of missing evidence, sample size, and freshness.

  4. Read the conversations behind the count

    Review actual questions, declines, referrals, and interest within your permitted scope. Record themes without exposing message bodies or contact details. A reply count includes replies that are not sales interest.

    Keep this artifact

    Two supported observations, one unknown, and the human follow-up owner.

  5. Choose one next test

    Change one audience or message assumption only after evidence review. Name an owner and next review date. Delivery uncertainty needs investigation; never resend a provider action with an unknown outcome just to improve a count.

    Keep this artifact

    One proposed test with a reason, an approval owner, and a dated review. Observed changes do not prove causation.

Inspectable walkthrough

How we use Klosent: the workflow

Fictional arithmetic exercise: a report records 24 sent messages to 10 contacted people, 2 people replied, and 1 contacted person booked 2 meetings. It has no explicit delivery receipts. These are invented counts for learning, not Klosent or customer performance.

An example to adapt, with no customer result or live execution. Access and actions depend on authorized scope and your Workspace's actual availability.

  1. 01

    Outkomes → Reporting

    Start with
    One authorized scope and fixed dates; 24 messages, 10 people contacted, 2 people replied.
    Reviewable output
    Reply rate: 2 ÷ 10 = 20%. Delivery: unavailable because no explicit delivery outcome is tracked.
    Human decision
    Do not calculate 2 ÷ 24 as the canonical reply rate. Record the execution time; provider events may arrive later.
  2. 02

    Outkomes → Meetings

    Start with
    Two booking records, both for one contacted person.
    Reviewable output
    Meetings: 2. Contact-grain meeting rate: 1 ÷ 10 = 10%.
    Human decision
    Check held and qualification evidence separately. Neither number proves two qualified opportunities or revenue.
  3. 03

    Konvos → human review

    Start with
    The permitted reply threads and the audience brief.
    Reviewable output
    An observation such as ‘one person asked for a resource’; a named owner for the answer.
    Human decision
    Use the actual request to decide the next step. Leave the uncertain delivery outcome for investigation and propose only one reviewed change.

Your working checklist

Checkboxes are for your reading and are not saved. Download the worksheet for your review. Keep client data in its authorized space.