Summary

  • ARIN opened its 2026 Customer Satisfaction Survey on 28 May and said that many questions repeated earlier surveys so that results could be benchmarked; the announced closing date was 19 June.
  • ARIN’s published 2023 research gives a usable starting frame: 317 online respondents split across three stated relationships to ARIN, with a reported overall margin of error of plus or minus six percentage points at 95% confidence.
  • The available 2026 invitation does not publish a response count, response rate, weighting plan, instrument, results or action plan. It therefore does not yet support a conclusion that customer views rose, fell or stayed constant.
  • Before ARIN presents a 2026-versus-2023 result, it should publish a composition bridge: fielding and invitation rules, relationship classes, completed and excluded counts, weighting, question-version changes, uncertainty and the rule that connects a finding to a follow-on decision.

A benchmark starts with a question about the sample

ARIN’s 28 May announcement invited the community to a 2026 survey on services and operations. The accompanying ARIN blog post says the exercise occurs every three years, should take about 15 minutes, and deliberately repeats many questions from earlier work to permit benchmarking. It also identifies Radius Insights, formerly Rockbridge, as the research firm that has conducted ARIN’s prior surveys.

That is a credible reason to preserve continuity. A repeated question can reveal movement that an entirely new questionnaire cannot. Yet it can reveal a trend only after readers know whether the people who answered, their relationship to the registry, and the conditions of participation remained comparable enough for the stated comparison.

The 2023 Customer Satisfaction Tracking Research makes that point unusually concrete. It reports an online field period from 19 April to 15 May 2023 and 317 respondents. Of those, 278 had direct ARIN IP allocations and membership; 30 had no direct ARIN number resources but used ARIN services; and nine had neither direct resources nor service use but belonged to the ARIN community. The report also records that the direct-assignment relationship question was removed in 2023, gives a median completion time of 15 minutes, and cautions that the total sample’s stated margin of error was plus or minus six points at 95% confidence, with greater uncertainty for subgroups.

Those details do not make the 2023 answers representative of every ARIN stakeholder. They do, however, describe the comparison baseline much better than an undifferentiated total would.

The missing bridge is specific, not ceremonial

The 2026 invitation provides dates and intentions; it does not provide the data needed to establish comparability. It does not state how many invitations were sent, which eligibility groups received them, how many starts or completions occurred, whether entries were excluded, whether weights were used, or whether the published question wording and response scales matched 2023. Nor does it publish the 2026 respondent mix or any result. None of those absences proves a flaw. They mark information that is not yet public.

A composition bridge should make the later benchmark auditable without exposing individual respondents. At minimum it should show, for both years: field dates; invitation and recruitment channels; eligibility and relationship definitions; completions, exclusions and weighted counts for each relationship class; question wording, response-scale and coding changes; the research vendor’s role; and the comparison rule. If an item was removed, merged, recoded or asked of a different population, the bridge should label it non-comparable or qualify the comparison.

The bridge should separately state uncertainty. The 2023 report’s overall margin does not automatically apply to a future sample, to a subgroup, or to a year-on-year difference. A simple table can distinguish an observed result from an inference that the available design can actually sustain.

An action claim needs its own join

ARIN’s announcement frames the survey as an opportunity to understand customer experience. That is valuable, but a result becomes operationally meaningful only when readers can see what decision surface it informs. A transparent release can name the service area, the relevant result, the owner who will assess it, the possible response, and the date for reporting back. It need not promise a change for every score.

This is not a demand to turn a satisfaction survey into a referendum on ARIN’s legitimacy. Nor is it an accusation about the survey firm or the people who respond. It is a modest governance rule: when a registry asks the public to read a new number against an old one, it should show the factual bridge that makes the comparison proportionate.

Sources