For outbound managers and dialler admins who suspect their lists are being churned rather than worked: how to segment, measure penetration honestly, set recycling rules by disposition, and keep the data clean enough to keep regulators bored.
What is an outbound calling list strategy?
An outbound calling list strategy is the set of rules that decides which records get dialled, in what order, how often, and when they stop. It covers segmentation by source, age, outcome and value, disposition-based recycling timers, attempt caps, penetration measurement, and hygiene: dedupe, suppression and regulatory screening. Done well, it is the main difference between professional outbound and expensive churn.
Four cuts that every later rule depends on.
Most lists arrive as one flat CSV, and one flat CSV is a pile of decisions someone has quietly delegated to the dialler. Before a record loads, cut the file four ways: where it came from, how old it is, what happened last time anyone called it, and what it is worth if it answers. Every rule later in this article keys off those cuts, which is why they come first.
A web form completed twenty minutes ago, a vendor file bought last week and a lapsed file from last year are three different products. They deserve different cadences, different openings and different expectations, and blending them destroys your ability to see which one is paying.
Fresh enquiries decay in minutes and belong at the front of the queue with their own service level. Aged records reward patience and cheap capacity. The worst of both worlds is a mixed list where fresh leads queue behind a thousand fourth attempts.
Never attempted, attempted but never reached, reached but unresolved, and promised a callback are four separate populations. Treating them as one list means the dialler cannot honour the difference, and your best prospects get the same treatment as your dead ones.
A record worth ten times another should not compete with it for the same agent minute on equal terms. Value banding does not need to be clever: top tier, standard and remnant is enough to route capacity where the return is.
The floors we audit almost always have this segmentation sitting neatly in the CRM and one undifferentiated list sitting in the dialler. The dialler is the place it matters, because the dialler is where the money gets spent.
Attempts per record flatters lazy campaigns; pair it with unique-contact depth.
Attempts per record is the penetration number everyone quotes: total dials divided by records loaded. On its own it lies. A list can show three attempts per record while a third of the file has never been dialled at all, because the dialler kept returning to the segments that answered. Pair it with unique-contact depth: the share of records attempted at least once, and the share that has reached a live conversation. One tells you how hard the list was hit, the other tells you how much of it was actually explored.
| Measure | Calculation | What it tells you |
|---|---|---|
| Attempts per record | Total dials divided by records loaded | Effort. Says nothing about coverage on its own. |
| Unique reach | Records attempted at least once, as a share of records loaded | Coverage. Low reach with high attempts means the dialler is circling its favourites. |
| Contact depth | Records with at least one live conversation, as a share of records loaded | Yield. The number the list was bought for. |
Those three lines settle the question of when a list is done. Done means the records still standing have been worked to your attempt cap, across at least two dayparts and more than one day, and contact depth has stopped moving. Lazily abandoned means answer rates sagged on day two and someone loaded a fresh file, because fresh files feel like progress. The gap between those two states is where a surprising amount of paid-for data quietly dies.
One refinement worth the effort: track contact depth to the right person, not just to a human. A list can look well penetrated while mostly reaching gatekeepers and spouses, which is a different problem with different fixes, covered in our piece on right-party contact rate.
Seventeen questions, about five minutes, no sales call required. The free Five9 Health Check scores your list setup, recycling rules and reporting habits against how we configure the floors we run, and flags what to fix first.
Each outcome tells you something different, so each gets a different retry rule.
Every disposition is information, and the retry rule should match what the outcome actually told you. A no-answer says almost nothing: wrong moment, try a different one. A busy tone says someone is on that line right now, which makes it the strongest case for a short-interval retry in the whole scheme. A voicemail greeting confirms the number is live and attended, so the timing was wrong rather than the data. A promised callback is a commitment with a clock on it. And hard outcomes are answers: a firm no, a wrong number, a do-not-call request. Answers leave the pool.
| Outcome | What it tells you | Typical operator treatment |
|---|---|---|
| No answer | Nothing about the number, only the moment | Cycle back after a meaningful delay and land the next attempt in a different daypart, not the same hour tomorrow |
| Busy | The line is live and occupied right now | Short-interval retry with a tight attempt limit. Five9 ships its stock Busy disposition as redial after 5 minutes, up to 5 attempts, which is the platform drawing the same conclusion |
| Voicemail | Live, attended number, wrong timing | Shift daypart on the next pass, and decide deliberately whether a message is left. Two identical voicemails in a day read as spam to the person who plays them back to back |
| Callback agreed | A commitment with a timestamp | Honour it to the minute and route it to the right agent where possible. Five9 lets agents set the redial time on the disposition for exactly this |
| Not interested | An answer | Exit this campaign. Re-approach months later only with a genuinely different proposition, and only where the consent position still supports it |
| Wrong number or DNC request | A data defect or a legal obligation | Suppress immediately, at the moment of disposition rather than in an end-of-day batch. DNC requests wait for nobody's batch job |
Built this way, the dialler stops treating the list as a queue and starts treating it as a set of promises and probabilities. That is also the practical difference between an aggressive cadence and a witless one: both make a similar number of calls, only one makes them at moments that had a chance of answering.
The honest answer is that no universal attempt cap exists, and anyone selling you one has fitted a curve to their own data and called it physics. What exists is a shape we see on every floor we run: the early attempts on a record carry most of its contact probability, and every pass after that buys less contact for the same labour. The operator response follows from the shape, not from a magic integer.
Vary the daypart before you add attempts. Three calls at two in the afternoon on consecutive days are closer to one attempt made three times than to three genuine attempts, because they only ever sample one slice of the prospect's week. Set two caps, one per number and one per record, since a three-number record can rack up nine dials while looking like three from the record view. And remember that callers have reputations now: carrier analytics watch dialling behaviour, and high-frequency repeat calls to unresponsive numbers are part of the pattern that earns a spam-likely label, which we cover in fixing spam-likely caller ID. A cap is cheap insurance on your calling identity as well as your payroll.
Then let the numbers move the cap. Review contact depth by attempt band monthly; if the final band has produced nothing measurable for a month, the cap is telling you where it wants to sit. Diminishing returns are not a slogan here, they are a visible line in a report, and the discipline is simply agreeing in advance to believe it.
The unglamorous work that decides whether the rest of the strategy is legal and true.
The unglamorous work is where the risk lives, and it slips precisely because nobody gets congratulated for a clean load file. Four disciplines, none optional.
Buy from two vendors for long enough and the same person arrives twice with two different lead IDs. Dedupe on normalised phone number first and person second, at load time, keeping whichever copy carries the better consent trail. Duplicates across campaigns are nastier than duplicates within a list: two campaigns each behaving reasonably can combine into a contact frequency nobody signed off, and the complaint that follows lands on the brand, not on either campaign.
Converted customers, complainants, people who told you to stop, records flagged deceased or vulnerable: all of it belongs on a standing suppression file applied at every load and every recycle, not once at purchase. Lists get reloaded, re-imported and restored from backups, and a suppression check that ran only on day one leaks on day forty. Reconcile the counts every time: records in, records suppressed and records loaded should balance exactly, because a suppression step that fails silently is the most expensive kind of bug in outbound.
Regulation 21 of PECR prohibits unsolicited live marketing calls to any number listed on the Telephone Preference Service register unless the subscriber has notified the caller that they want the calls, and a number is treated as protected once it has been registered for 28 days (the regulation text is short and worth reading). The ICO's guidance draws the practical conclusion: screen call lists against the TPS, screen business calling against the CTPS as well, and keep and screen your own do-not-call list on top. The stakes changed recently. Provisions of the Data (Use and Access) Act 2025 that commenced in February 2026 lifted the ceiling on PECR penalties from £500,000 to £17.5 million or 4 percent of global annual turnover (Clifford Chance has a clear summary), which rather changes the economics of casual screening. Operators typically screen at load and rescreen the standing pool on a short cycle, since a list that was clean at purchase does not stay clean while it sits.
The FTC's Telemarketing Sales Rule requires covered sellers and telemarketers to synchronise calling lists with an updated version of the National Do Not Call Registry at least every 31 days (the FTC's own Q&A states it plainly), and the rule's safe harbour depends on using a registry version obtained no more than 31 days before any call is made (16 CFR 310.4). Entity-specific requests sit alongside the registry: a person who tells one seller to stop is honoured regardless of whether their number is registered. The established business relationship exemptions, 18 months from a transaction and three months from an enquiry, are narrower than most sales teams assume, and they end the moment the person asks not to be called. Rules in this space have moved repeatedly and enforcement moves with them, so verify your cadence against the current text and take advice from qualified counsel where it matters, rather than from consultants' articles, this one included.
Calling windows attach to the record, not the campaign, so every record needs a defensible timezone. Area code is the standard guess and it fails in predictable places: ported mobile numbers, people who moved and kept the number, and shared exchanges near boundaries. Five9 assigns timezone from area code and prefix by default and can use the contact's postal code or state instead where the record carries them (timezone dialling options). The operational fix is upstream: capture state and postcode at acquisition, because asking costs nothing and guessing costs a breakfast-time call to someone three timezones west of their area code.
Lists tell you when they are tired; most reporting just is not listening. Six signals earn a standing weekly check:
When several move together, rest the segment rather than pushing it, re-cut what remains using the outcomes gathered so far, and rotate in genuinely new data instead of raising pace to chase yesterday's numbers. The fix is rarely in the pacing algorithm. Rebuilt recycling rules and honest list resets are the largest single lever we find in tuning work: across our optimisation portfolio the aggregate connect rate improvement is plus 40 percent, and more of it has come from list discipline than from anything done to dialling pace. For the surrounding numbers, our outbound benchmarks piece publishes the bands we actually see in the market.
Cost per lead is the number on the invoice. Cost per contact is the number that matters, and it is a division your reporting can already do: spend by source over unique live conversations by source, weekly. Two vendors at identical cost per lead routinely land far apart on cost per contact once duplicate rates and connect rates come in, and that spread stays invisible unless dialler outcomes flow back to whoever signs the data purchase orders.
Where volume allows, push one step further to cost per qualified outcome by source. Be honest about sample size, a fortnight of one vendor's file is a hint rather than a verdict, but even the hint beats renewing on the vendor's own numbers. The feedback loop is the point: sources that produce cheap leads and expensive conversations get killed or renegotiated, and the budget migrates to whatever your own outcomes say is working.
Fresh and aged data both have a price at which they work. Fresh enquiries cost more because their value decays in minutes, which makes them a workflow problem before a data problem, the case set out in speed to lead. Aged files are cheap because much of the contactable value has already been extracted by someone else, and they can still pay when worked patiently in low-cost capacity windows under a strict cap. What kills programmes is paying fresh prices for aged behaviour. Data brokers price on optimism; buy on your own numbers instead.
Where each control lives, and the three settings that do most of the work.
Everything above is platform-agnostic. Here is where each control actually lives in Five9, because a strategy that never reaches the admin console is a memo, not a strategy.
| Discipline | Where it lives in Five9 |
|---|---|
| Segmentation | Separate lists per segment on the campaign. List priority sets the order: lower numbers dial first, equal priorities merge lists into one virtual sortable list, and List Dialing Ratios balance volume between lists sharing a priority. |
| Recycling rules | Disposition types. Redial Number carries a redial timer and an attempt limit, Do Not Dial Number for Campaign parks a number with an optional reactivation timer, a final disposition retires the whole record, and Add Number to DNC List writes straight to the domain suppression list. |
| Retry timing | Minimum Duration Before Redialing on the campaign, default 10 minutes, sets the floor. A disposition-level redial timer overrides it until that disposition's attempt limit is reached. |
| Attempt caps and recycling filters | Campaign profile filters: up to 20 conditions combined with AND, OR and NOT across contact fields, last disposition, last attempt time and dial-attempt counts, plus sorting on up to three fields. |
| Timezone integrity | Time Zone Settings per campaign: area code and prefix by default, with postal code or state options where the record carries them, and state dialling rules layered on top. |
| Penetration measurement | Standard reports: List Penetration, List Statistics by Campaign, Unreachable List Record Reasons, and record-level List Details for audits. |
Three details in that table do more work than the rest. Setting a list's priority to 0 parks it without detaching it, so its history and reporting stay intact while it rests, and Workflow Rules can flip priorities on a condition such as time of day, which is how a floor runs fresh-lead bursts in the morning and aged passes after lunch with nobody touching the console (Five9, Defining List Priority). Campaign profile filters are the recycling engine most admins never open: a filter on dial attempts and last disposition is "work no-answers under the cap, leave callbacks alone" expressed as configuration (Five9, Filtering and Sorting Lists). And one warning from the same documentation matches our experience exactly: Five9 advises against sorting lists larger than 50,000 records, because large sorts slow campaign startup and list recycling enough to cost agent productivity.
On measurement, List Penetration gives dialled records and percentages by campaign and list, List Statistics by Campaign gives average and maximum attempts to first contact plus time to first contact, and Unreachable List Record Reasons breaks out exactly why the remainder cannot be dialled: DNC, do-not-dial dispositions, attempt caps reached, or the calling window closed (Five9, List Reports). If those are not on a weekly review, penetration is a feeling rather than a number. The setups we audit fail here more than anywhere else: the controls exist, configured once, reviewed never.
Use two measures together. Attempts per record is total dials divided by records loaded, and unique reach is the share of records attempted at least once, plus the share that reached a live conversation. Quoting attempts per record alone flatters campaigns that hammer responsive segments while never touching a third of the file. In Five9, the standard List Penetration report shows dialled records and percentages by campaign and list, with contacted and finalised counts alongside.
There is no universal number, and we distrust anyone selling one. In our experience the early attempts carry most of a record's contact probability, so vary the daypart before adding attempts: three calls at the same hour are barely more than one. Set a per-number cap and a per-record cap, review them monthly against contact depth by segment, and remember that high-frequency repeat dialling is a pattern carrier analytics associate with spam.
Recycling rules decide when and whether a record re-enters the dialling pool based on its last outcome. No-answers cycle back after a delay and a daypart change, busy signals retry on a short interval, voicemails move to a different time of day, and agreed callbacks are honoured at the promised moment. Hard outcomes such as not interested, wrong number or a do-not-call request exit permanently. In Five9 these rules are built with disposition types, redial timers and attempt limits.
A list is done when the records still active have been worked to your attempt cap across at least two dayparts and more than one day, and unique contact depth has stopped moving. It is lazily abandoned when answer rates dipped on day two and a fresh file got loaded instead. Before retiring a list, check why the remainder is unreachable: in Five9 the Unreachable List Record Reasons report separates DNC, do-not-dial dispositions, max attempts and closed calling windows.
No. They are different products with different economics. Fresh enquiries decay in minutes, so they belong in their own high-priority list with speed as the design goal. Aged records are cheap because much of their contactable value has been extracted, so they suit patient passes in low-cost capacity windows with strict caps. In Five9, keep them as separate lists on the campaign and use list priorities and ratios to control the blend rather than mixing files.
The FTC's Telemarketing Sales Rule requires covered sellers and telemarketers to synchronise calling lists with an updated version of the National Do Not Call Registry at least every 31 days, and its safe harbour depends on using a registry version obtained no more than 31 days before any call. Entity-specific requests, where a person tells one company to stop, apply regardless of registry status. Rules change, so verify against the current text and take qualified advice where it matters.
Regulation 21 of PECR prohibits unsolicited live marketing calls to numbers listed on the Telephone Preference Service unless the person has notified the caller that they want the calls, and a number gains protection once registered for 28 days. ICO guidance says to screen lists against the TPS, screen business calling against the CTPS too, and keep your own do-not-call list. Since February 2026, maximum PECR penalties align with UK GDPR levels: £17.5 million or 4 percent of global turnover.
List priority on the campaign's Lists tab. Lists with lower priority numbers dial before higher ones, lists set to the same priority are treated as one combined virtual list, and List Dialing Ratios balance volume between lists sharing a priority. Setting a priority of 0 stops a list dialling entirely while keeping it attached for reporting, and Workflow Rules can change priorities automatically on conditions such as time of day. Campaign-level priority and ratio then govern between campaigns.
AL Performance is a Five9 Certified Implementation Partner working with UK and US outbound teams. You get a dedicated pod that designs, builds and tunes the whole loop, lists, recycling, reporting and buying feedback, as day-job work rather than a ticket queue.
Talk to us