THE 98,518 DIAL STUDY: DATA DICTIONARY AND SOURCE NOTE Dataset release: 1.0 Web page last updated: 31 August 2026 Frozen analysis window: 18 May 2026 through 11 August 2026, inclusive Canonical source: https://albatalent.io/research/98518-dial-study Aggregate CSV: https://albatalent.io/research/98518-dial-study-aggregates.csv Licence: Creative Commons Attribution 4.0 International (CC BY 4.0) SCOPE This release contains published aggregate results only. It contains no phone numbers, names, companies, caller-level rows, call recordings or other contact data. It describes one firm's US B2B outbound operation: 98,518 analysed call attempts to 60,054 distinct numbers by nine callers. OUTCOME DEFINITION "Reached a human" means that the call ended in an outcome recording a person on the line: a conversation, gatekeeper, wrong contact or refusal. It measures whether a person answered, not whether the call produced a meeting, opportunity or sale. RECONCILIATION The dialler table held 132,570 records at the freeze. Of these, 22,316 had no usable timestamp and 1,205 timestamped records fell outside the frozen window. That left 109,049 in-window records. Excluding 9,911 cancelled records and 620 records with no outcome produced the final 98,518 analysed dials. The 9,911 cancelled records averaged 0.8 seconds and were treated as abandoned before an attempt rather than completed attempts. TIME OF DAY Prospect local hour was derived from North American area code using a fixed UTC offset. The study window was entirely within daylight saving time. Of the 98,518 analysed dials, 91,477 (92.9%) mapped through the 284-area-code table. Ported and VoIP numbers can make area-code location wrong. Hours with fewer than 300 dials were not displayed; this removed hours before 08:00 and after 17:00. ATTEMPT NUMBER Attempt number was calculated inside the frozen study window. A number dialled before 18 May can therefore appear as a first attempt in this release. The all-attempt series is subject to survivorship because numbers that answer leave the redial pool. The fixed-cohort series uses the 1,578 numbers that received at least six within-window dials, so the same numbers appear at attempts one to six. CSV DATA DICTIONARY - analysis: the published analysis or robustness check. - segment: the displayed hour, attempt, weekday or reconciliation category. - metric: what the row measures. - value: the published value, preserved at its published precision. - unit: percent, count, seconds or decimal hour. - sample_dials: published denominator where the article states one; blank when no cell denominator was published. - human_reaches: published numerator where the article states one; blank when no numerator was published. - notes: qualifications required to interpret the row. Blank cells are deliberate. They mean the source article did not publish that cell's denominator or numerator. Do not reverse-engineer counts from rounded percentages. LIMITATIONS - This is one firm, nine callers and twelve weeks, not a cross-industry survey. - The firm's ICP, lists, scripts and operating choices are present in the data. - Area-code timezones are noisy for ported and VoIP numbers. - The fixed cohort is 1,578 numbers; the sixth-attempt recovery is 53 human reaches and is the least certain reported finding. - The 08:00 cell is the smallest displayed hourly cell at 2,899 dials. - Granular outcomes include caller-entered dispositions. - The study is observational. It does not establish that changing call time causes a particular connect rate. REUSE The article, aggregate CSV and chart are licensed under CC BY 4.0. Attribute: Alba Talent, "The 98,518 Dial Study" (2026), analysis of 98,518 US B2B cold calls placed 18 May to 11 August 2026, https://albatalent.io/research/98518-dial-study When reproducing a chart or statistic, retain the prospect-local-time basis, the one-firm scope and the observational limitation. Do not relabel connect rate as meetings, pipeline or revenue, and do not present these aggregates as a universal industry benchmark.