Overview shows a checklist based on your saved workspace: actual offer, Fathom connection, imported calls, calibration, Slack test, and automation.
Owners confirm whether approved knowledge sources have been reviewed or the playbook alone will be used. Mark three real sales audits reviewed under Manager review. They must use your current playbook revision to count.
A successful Send test message updates the Slack step. Reconnect a channel and test again. Slack and automatic analysis are optional when you review calls manually.
In Connections, enter optional comma-separated title phrases to include or skip and a minimum call duration. Matching ignores case; any included phrase qualifies, and excluded phrases always win. Save settings to apply them.
These rules apply to both manual and automatic analyses before a credit is reserved. They do not remove imported calls from your library or guarantee that matching calls are sales calls.
For individual recordings, use Exclude this call from analysis under Manager review. Queued analyses are canceled, but a review already running may finish. Removing the exclusion does not automatically queue a new review.
Open an imported call → Manager review. Choose Agree, Flag an inaccurate assessment, or Record a corrected overall score. Add an explanation, ideally with transcript timestamps.
Corrections apply only to scored audits and use a whole-number score from 0 to 100. The original AI score stays visible; the manager score and explanation are stored separately.
Feedback history records who submitted each assessment and when. Re-analysis keeps earlier feedback in history, but does not treat it as feedback on the new audit. Feedback does not retrain the AI, edit category scores, or resend Slack messages.
Open Coaching trends and select a rep, the last 30 or 90 days, and a playbook revision. The comparison uses the equally long preceding period and the date each call took place.
Category averages use original AI grades out of five. Sample calls, unscored calls, and other playbook revisions are excluded. Each row shows its sample count.
A change appears only when both periods contain at least three calls. Small samples and different prospect mixes can affect results; changes do not establish why performance changed. Manager corrections appear separately below the category table.
Owners can enable failure-alert emails in Connections and save settings. Emails go to current company owners, with at most one grouped notice per company per hour.
Alerts cover blocked jobs and failures that exhausted automatic retries. Emails contain no call titles, transcripts, or raw provider errors; open Activity & retries for details. Previously reported jobs are not repeatedly emailed.
Slack failure alerts use email, so they can still arrive when Slack delivery is broken. Email delivery and the background scheduler must be configured. Connections shows the last sent alert and pending notices. If the scheduler itself is unavailable, it cannot send alerts.
Open Team playbook and replace the example offer, objections, and grading criteria with your company’s actual standards. Make the rubric weights total 100%, then save.
If you have a sales handbook, upload it under Client setup & references. Review the extracted text and approve it before it can inform audits.
Connect Fathom, choose the right team and recording owners, and import a small batch of historical calls. Leave automatic analysis off while you check the setup.
Open a real sales call from Call library and select Analyze call. Compare the scorecard and full audit with the transcript. We recommend calibrating on three representative sales calls.
Connect Slack, save your notification settings, and send a test message. Once you trust the results, enable automatic sync and, separately, automatic analysis.
In Connections, save your Fathom API key. Your key’s access determines which recordings CloseKitty can retrieve. Never share the key in support messages.
Choose team and recorded-by filters to focus on the sales conversations you want to review. Historical imports can request up to 25, 50, or 100 matching calls at a time, with optional UTC date boundaries.
Importing saves the call and transcript; it does not necessarily analyze the call. Automatic analysis is a separate setting. Already imported calls are not duplicated.
If an import returns zero, check the date range and filters, whether those calls were already imported, and whether the connected Fathom account can access them. A newly recorded call may still be processing in Fathom.
Team playbook holds your grading rubric, objection handling, offer details, and client setup. Owners can edit these settings; managers can review and analyze calls.
Describe the actual offer, audience, price, and approved payment guidance. Include additional offers when relevant. Do not rely on the AI to invent discounts, financing eligibility, or promises.
Use coaching instructions to explain your sales approach. Save the playbook after editing; rubric weights must total 100%.
Saved changes apply to future analyses. Existing audits retain their original context. Re-analyzing a call uses another analysis credit. You can export and import a playbook template without sharing calls or credentials.
In Team playbook → Client setup & references, upload PDF, DOCX, TXT, or Markdown files. Each file can be up to 3 MB; PDFs can have up to 150 pages.
Uploads begin as drafts. Inspect the extracted text, fix or replace unreadable documents, and approve only the content you want the auditor to use. Image-only scanned PDFs may need a searchable text version.
The limits are 40,000 extracted characters per file and 10 active references with 100,000 combined characters. CloseKitty retains extracted text, not the original document file.
New versions must be reviewed and approved. Audits retain reference version information so you can identify the guidance used. Approved reference text is included with the call when sent for AI analysis.
Open a call in Call library. Scorecard summarizes performance; Full audit adds coaching detail. Use Transcript to check evidence and Manager review to save your own notes and review status.
Audits assess opening, discovery, value, objections, and closing against your saved standards. Separate transcript observations from hypotheses about the prospect’s perspective. AI coaching should be checked by a person.
A call can be marked as non-sales or unclear and remain unscored. Coaching sessions and support calls should not receive a sales-performance score.
An outcome such as Committed is not proof of payment. Follow-up suggestions are coaching guidance; CloseKitty does not automatically contact the prospect.
An owner can connect Slack from Connections and select the destination. For a private channel, add the CloseKitty app to the channel as well.
Choose every completed review, scores below a threshold, daily digest, or off. Save settings and use Send test message to verify the destination.
The app connection sends individual review summaries with the full audit in a thread. Daily digests link to audits. An existing incoming-webhook connection supports summaries only; reconnect through the app for threaded audits.
People in the channel can read posted summaries and audit excerpts. Opening a linked call still requires access to your CloseKitty company.
Check activity if a message fails. If delivery is marked uncertain, inspect Slack before retrying to avoid posting a duplicate.
Automatic sync checks for new calls every 5, 15, or 60 minutes and runs while CloseKitty is closed. It starts with calls from the time you enable it; use a historical import for older calls.
Analyze imported calls automatically is a separate switch. When enabled, historical imports can queue analyses too. Save your playbook and check your call filters before enabling it.
Your analysis cap applies to manual and automatic reviews together. The plan allowance also applies. At the cap, jobs wait until you raise the cap or renew your allowance and retry. There are no automatic paid overages.
Calls must have a transcript and be no longer than 120 minutes. Very large transcripts can also exceed processing limits. Re-analysis uses another credit.
Use Company & billing to check usage and your access period. Owners with a paid subscription can open Manage billing. A complimentary pilot does not automatically become a paid subscription.
Use the email address associated with your account or invitation. Open the newest sign-in link in the same browser where you requested it. Links expire after 15 minutes and can be used once.
If the email is missing, check spam and confirm the address before requesting another link. If you are joining an existing company, ask its owner to confirm your invitation.
In Company & billing, accept an invitation or switch between your companies. Owners can invite and remove managers within their seat allowance.
Managers can view, analyze, and coach all calls in their company. Owners also manage the playbook, integrations, billing, and membership. Invite only people who should have access to company calls.
Awaiting AI analysis: the transcript was imported but no audit is ready. Open the call and select Analyze call, or check automatic analysis and activity in Connections.
Evidence could not be verified: an audit could not be grounded in the transcript. Check that the transcript is readable and retry. If it repeats, contact support with the call ID and error text.
No Slack message: confirm the channel, saved notification mode, and completed analysis. Threshold mode does not send every review. Daily digest follows the configured hour and time zone. Check delivery activity and try a test message.
Analysis paused: check both your configured cap and your plan allowance. Also check access status, call length, and whether the transcript is available.
For support, include the page, error message, approximate time, and steps that triggered the issue. Do not email API keys, webhook URLs, or full transcripts.
CloseKittyOpen workspace