01Caller ID reputation: your numbers may be flagged before anyone hears a ring
If your connect rate declined gradually over weeks rather than dropping overnight, start here. The major carriers each run analytics engines (First Orion for T-Mobile, Hiya for AT&T, TNS for Verizon) that label or block numbers based on dialing behavior, and consumer apps add another layer of spam databases on top. A number that dials too much, too fast, with too few answers gets labeled “Spam Likely,” and answer rates on that number collapse regardless of anything else you fix.
<strong>How to check:</strong> scan your ANI pool against carrier databases with a reputation tool, or the zero-cost version — call your own numbers from phones on each major carrier and see what displays. In one finance client audit, roughly half of all blocked outbound calls were eliminated after a full caller ID review and remediation. That was not a script problem or an agent problem; the calls were dying before they rang.
The durable fix is behavioral, not just remediation: rotate numbers so no single ANI carries too much volume, register your numbers with the carrier analytics engines, and retire numbers that are burned.
02Answering machine detection: agents talking to voicemail
Five9's answering machine detection profile decides what happens when the dialer reaches a machine. A misconfigured profile routes voicemails to agents, and every one of those is talk time spent on nobody. This one is invisible in most dashboards because the call “connected” — it just connected to a recording.
<strong>How to check:</strong> pull disposition data and look at the share of connects dispositioned as voicemail or no-contact. In one healthcare advisory audit, adjusting a single AMD profile cut voicemail calls routed to agents from 19.2% to 12.4% of calls, and conversion rose from 7.7% to 10.1% within a month. Same agents, same leads, one profile change.
03Dialer pacing and dialing mode: too slow or too fast, both cost you
Pacing that is too conservative leaves agents idle between calls; pacing that is too aggressive drives abandonment and, indirectly, spam labeling from all the short-duration calls. The wrong dialing mode for the campaign type (predictive where power fits, or the reverse) produces the same symptoms. The tell is agents with high ready time next to lists that are nowhere near exhausted.
<strong>How to check:</strong> compare agent occupancy against list penetration per campaign. Low occupancy next to low penetration means the dialer is underfeeding the floor: agents sit idle while the list barely moves. High abandonment next to high occupancy points the other way: the dialer is overpacing, placing more calls than agents can pick up, which drops connects and burns caller ID reputation on every abandoned call. One home services client grew weekly outbound dials roughly 2.5x, from about 12,000 to the low 30,000s, with the same staff, after fixing an underpaced configuration. A different fix applies when the problem runs the other direction: pull the pacing ratio down until abandonment sits under your compliance threshold, then re-check occupancy.
04Disposition sets and recycling rules: winnable leads exiting too early
Your disposition set and redial rules decide how many attempts a lead gets and when. Overly aggressive finalization (treating a no-answer or busy as done) drains your lists of exactly the leads most likely to answer on attempt three or four. The symptom looks like “our lists are bad,” when the truth is the rules never gave the lists a chance.
<strong>How to check:</strong> audit which dispositions finalize a record versus recycle it, and at what intervals. Compare attempts-per-lead against connect-by-attempt data. If most connects happen on attempts two through four but your average lead gets 1.5 attempts, the rules are the leak.
05List segmentation and time windows: right person, wrong moment
Dialing a list flat, without timezone discipline or segmentation by lead age and source, lands calls outside the windows where your contacts actually answer. This shows up as connect rates that vary wildly by hour while the team blames the leads.
<strong>How to check:</strong> break connect rate out by hour of day per timezone. If the spread between your best and worst hours is large and your dialing volume is not concentrated in the good ones, list logic is costing you connects that configuration alone will not recover.
06Reporting gaps: you cannot fix what you cannot see
The most common finding in audits is not any single misconfiguration; it is that nobody can see connect-rate decay, per-number answer rates, or disposition trends well enough to catch problems early. Five9's native reporting can produce this data, but it rarely surfaces it by default, so degradation runs for months before anyone notices.
The fix is a reporting layer that tracks connect rate, answer rate per ANI, abandonment, and disposition mix over time. When those trend lines exist, every problem above announces itself within days instead of quarters.
How to work through this in your own domain
Run the checks in the order above; it is roughly the order of frequency and impact. Most domains have two or three of these issues stacked, and they compound: burned caller IDs lower answer rates, which makes pacing look wrong, which drives redial behavior that burns more numbers.
If you would rather have it done for you, this checklist is the skeleton of the DialNode Five9 Performance Audit: a fixed-fee, read-only review of all six areas with every finding quantified and a prioritized fix list, delivered in 10 business days. If it does not surface at least five concrete improvement opportunities, you do not pay.
See the Five9 Performance Audit