Start Here / Step 7: Track, Improve & Scale
Seller Campaign Troubleshooting Guide
Verify common failure causes before changing the campaign.
Ad delivery
Symptoms: no spend, approval problem, very limited delivery. Inspect status and reason, payment, schedule/timezone, correct assets, audience controls, destination, and dataset eligibility. Fix the stated problem. Request review through the official route when appropriate; do not bypass a restriction. Recheck the destination after an edit.
Poor CTR or rising CPM
Compare the same metric definition, time grain, and placement scope. Link CTR uses link clicks, not all clicks. Inspect the actual mobile creative and whether the offer is understandable. If the hook is the issue, change one hook. If cost rises but qualified outcomes improve, judge the economics rather than the isolated CPM. Creative fatigue is a hypothesis to investigate, not a diagnosis from a single bad day.
High CPL and page problems
Separate link clicks, landing-page views if available, and successful submits. Test redirects, page availability, loading, field errors, and submission response on a phone. Check whether the ad promise is actually delivered. Repair defects immediately; otherwise use one headline or one field test. Do not remove required privacy or identity disclosures for conversion.
Weak qualification or seller fit
Sample real CRM records and exclusion reasons. Are they outside scope because the permitted delivery area is wider than expected? Is the property question missing? Did the offer attract information seekers without seller intent? Verify the agent can serve the compliant region and that the offer honestly connects to selling. Keep qualification consistent. Do not filter by protected traits or proxies, or redefine qualified after the fact to improve reporting.
Poor booking
Check that qualified sellers requesting a conversation receive a human response and a valid calendar route. Inspect open appointments, mobile calendar, required questions, timezone, reply monitoring, and promised conversation value. Ask an actual willing seller whether the page worked. Correct the bottleneck you can document; do not automatically send more messages to everyone.
Poor attendance
Use resolved appointment attempts from the scheduled-date window. Inspect reminders and reschedule jobs, invite time and location, agent availability, seller replies, and scheduling delay. A cancelled or rescheduled attempt is not a no-show. Test one message or availability change after repairing errors. After an actual no-show, use Appointment Confirmation and Reminder Pack's two respectful recovery steps and close the task if the seller does not respond.
Held appointments with weak listing conversion
Review notes from substantive conversations. Was the seller qualified before attendance? What information or decision was missing? Was a next step agreed and completed? Were expectations about timing, condition, price, or representation different from the offer? Have the brokerage or an appropriate coach review the conversation process. Keep private recordings subject to consent and applicable law. Track a signed listing separately from a closing.
Email or SMS problems
Check verified sender setup, form permission, suppression, bounced or rejected messages, and provider logs. Confirm the inbox receives replies. For email, inspect domain authentication with the sender's official instructions and fix reported failures; do not claim that one DNS edit guarantees inbox placement. For SMS, pause it until the reviewed consent, sender registration, opt-out, quiet hours, and state requirements are satisfied. Email/human fallback can continue where permitted.
Automation failure
Find the first missing event: accepted form, CRM create/update, resource send, task, qualification update, booking, reminder, or outcome. Inspect IDs and timestamps from CRM Automation and Connection SOP. Retry safely by the event's unique ID; do not blast duplicates. Preserve contact suppression during recovery. Backfill permitted pending tasks manually and verify the repair with one new test before reactivation.
Duplicate events or inconsistent counts
Repeated submit, page refresh, imported leads, browser/server events, co-owners, and reschedules can inflate different stages. Merge duplicate opportunities by seller/property in the CRM. Preserve attribution history. Assign a stable logical event ID. Reconcile raw platform events with accepted deduplicated records and document differences instead of forcing totals to match.
Uncertain experiment result
Inspect the baseline, unchanged variables, spend exposure, outcome maturity, and denominator. Unequal ad delivery is not random assignment. Mark Inconclusive if you lack sufficient evidence or practical difference. Stop at the prewritten cap. New ideas can wait until the important active test is resolved.
Recovery note
Issue [DESCRIPTION] First affected time [TIMESTAMP] Contacts affected [IDS ONLY] Verified cause [LOG / TEST] Fix [ONE CHANGE] Owner [NAME] Backfill [TASKS] Suppression respected [CHECK] New test [RESULT] Reactivated [DATE / BY WHOM] Next review [DATE]
You’re done when
The root issue is verified, the smallest useful fix is applied, and its result is logged.
Support reference R19