Compare Two Sales Call Lists Without Mixing Attempts and People

A list with a higher answer percentage isn't necessarily the better list. It may contain warmer contacts, more repeat attempts or fewer people who are hard to reach. Before replacing a list source, compare the same kind of work and keep the people count beside the dial count.

This worksheet is for a sales manager reviewing two existing call lists. It helps separate reachability, contact coverage and downstream progress. It isn't a method for proving that one data provider causes better results, and it doesn't set a universal target for your team.

Write the decision before opening the export

“Which list performed better?” leaves too much room to choose the flattering number. A more useful question is: “Which list should receive the next calling block for this audience, given the relevant conversations and time spent?” That question allows a list to be harder to reach but still useful. It also allows a high answer rate to be commercially disappointing.

Name the audience and the period. Keep genuinely cold contacts separate from people who requested a callback. Include the prospect's local calling window when it is available. A list called on Monday morning and another called over several different evenings are observations of different working conditions.

Write down what you cannot match. Perhaps one source has more direct numbers, or one group includes contacts previously reached by email. Those differences may be part of the purchasing decision, but they prevent a clean claim about the source alone. An honest comparison can still guide the next block without pretending to be a controlled experiment.

Build two views, not one headline percentage

The attempt view asks what happened each time a call was placed. The person view asks how many different intended contacts the team reached. They answer different questions.

For the attempt view, use actual outbound attempts and a documented answer category. State whether it includes any person or only the intended prospect. Don't switch definitions between lists. The existing connect-rate guide explains those definitions and their limits.

For the person view, use a stable contact reference where available. Count each person once within the stated period, even when they answer more than once. Two phone numbers for the same person aren't two people. Conversely, a shared company phone number doesn't make two different people the same contact. Keep this reached-person measure separate from list penetration, which asks how much of the assigned list received an attempt.

Don't force uncertain identities into the count. If the export cannot distinguish people reliably, report the attempt view and mark person coverage unavailable. Guessing a deduplication key from names can merge different people or split the same person. Resolve the data question before using coverage to rank lists.

A fictional comparison that changes with the question

Suppose List A receives 200 attempts to 100 people. Twenty attempts reach the intended person, but those answers come from 12 different people. List B receives 100 attempts to 100 people. Fifteen attempts reach 15 different intended people. These are illustrative numbers, not customer results or recommended targets.

• List A: 20 answered attempts out of 200, or 10%; 12 people reached out of 100, or 12%.
• List B: 15 answered attempts out of 100, or 15%; 15 people reached out of 100, or 15%.

B looks stronger on both percentages. A still produced more answered attempts, because it received twice the attempt volume. Those extra answers could include useful follow-up conversations; they aren't automatically wasted. But they aren't eight additional people reached either.

You cannot conclude that A needs exactly twice as much work to reach the same audience. The lists haven't received the same attempt pattern. Some people in B may answer on a later attempt; some may never answer. Keep the observed counts, then decide whether a comparable next block would answer the practical question.

Now add conversation quality. Suppose A produces six relevant conversations and B produces four. Record the rule for “relevant” before counting. If A's six are repeated conversations with three people, say so. If B's four are all new contacts, that matters too. A raw conversation total doesn't establish broader account coverage, meetings held or revenue.

Check the mix before declaring a winner

A blended percentage can hide a change in audience. Consider another fictional example, with “answer” defined consistently as reaching the intended person:

• List A's new contacts: 8 answers from 100 attempts, or 8%. Its requested callbacks: 30 from 100, or 30%. Combined: 38 from 200, or 19%.
• List B's new contacts: 18 answers from 200 attempts, or 9%. Its requested callbacks: 16 from 50, or 32%. Combined: 34 from 250, or 13.6%.

A has the higher combined rate. B has the higher rate within each of the two groups. The combined result differs because A contains a much larger share of requested callbacks. Replacing B solely because 13.6% is below 19% would ignore that difference.

This example doesn't prove B is the better source. The groups may still differ by role, region, age of data or caller. It shows why you should inspect comparable groups before relying on the total. Don't erase the total either: it describes the work the team actually did.

Start with the differences most likely to affect your decision: cold versus requested follow-up, first versus later attempts, and major audience segments. Avoid slicing a small export into dozens of tiny groups. A percentage based on one or two answers can look dramatic without supporting a stable conclusion.

Keep missing outcomes visible

An attempt without a recorded outcome is unknown, not an unanswered call. Show the number of unknown outcomes beside the total. If you calculate a rate only among attempts with known outcomes, label that denominator and explain the excluded count. Don't present it as the rate for every attempt.

Check duplicate activities before comparing. One call logged twice shouldn't become two attempts. Preserve the original export and record how duplicates were identified. The dialer-to-CRM reconciliation worksheet covers that separate task.

Also keep invalid numbers, wrong people and unanswered calls distinguishable when the source records allow it. A wrong-number problem calls for data repair; it isn't evidence that prospects rejected the message. Never remove inconvenient outcomes after seeing which list wins.

Copy a comparison note your team can reuse

• Decision: Which next block, source or audience choice will this review inform?
• Period: Start, end, timezone and whether outcomes have had time to appear.
• Included work: List sources, audience, reps, calling windows and attempt rules.
• Definitions: What counts as an attempt, intended-person answer, relevant conversation and agreed next step?
• Counts: Attempts, answered attempts, unique people attempted and reached, unknown outcomes and duplicates removed.
• Comparable groups: Cold versus follow-up, attempt number and important audience differences.
• Effort and next steps: Observed calling time where reliably available; agreed actions, booked meetings and held meetings kept separate.
• Decision and limits: What will change, what remains uncertain and when the team will review another comparable block.

For example: “We'll use B for the next new-contact block because its cold-contact answers were slightly higher under the recorded conditions. A's overall lead came mainly from requested callbacks. We'll keep A's callbacks with their existing owners and compare another equivalent new-contact block before changing providers.”

That is a working decision, not a claim of proven superiority. It protects existing conversations while giving the team a specific next observation. Don't add calls solely to manufacture a larger sample, disregard opt-outs or move a promised callback just to balance the worksheet.

When the comparison should wait

Pause the ranking when outcome definitions differ, identities cannot be resolved, most calls are still missing outcomes or the supposed comparison is really two different audiences. Fix the reporting problem or narrow the question. “We can't compare these yet” is more useful than a precise percentage with an unreliable denominator.

Finish with the decision the numbers can support. Often that's one comparable block, a data-quality repair or a clearer reporting rule—not a new vendor, a rep leaderboard or a revenue forecast.