Summary

  • LACNIC reported meetings with 214 member organizations in 2024, more than 330 organizations in 2025 and close to 90 organizations in Venezuela in 2026.
  • Its 2025 paragraph says that, among those who provided feedback, 98% reported the highest levels of satisfaction. It does not say how many responses were received.
  • Attendance, eligible respondents, completed responses and top-level ratings are different units. The missing base does not prove a weak survey; it prevents the reader from knowing how the result relates to the more-than-330 organizations.
  • A short methods note could preserve privacy while publishing the respondent unit, response count and rate, question, scale, top-box rule, duplicate treatment, dates, rounding and revision history.

The percentage has a missing noun

LACNIC's 26 August account of its member outreach opens with a persuasive institutional story. Since 2015, the registry says, it has gone into the region to meet the organizations it serves. Between 2024 and 2026, it shifted from door-to-door visits toward working breakfasts and a larger presence at industry events. That change allowed it to reach more organizations at once.

The post supplies useful scale. In 2024, LACNIC says it met 214 member organizations in the Dominican Republic, Costa Rica, El Salvador and Guatemala. Its booths at Andina Link in Cartagena, Internet Day in Buenos Aires and CANTO also served as consultation spaces. In 2025, the programme expanded to seven cities in five countries and brought together more than 330 organizations—nearly 10% of LACNIC's direct membership. In 2026, the registry says it visited close to 90 organizations in Venezuela, participated in Expo ISP Venezuela and again attended CANTO. Paraguay and Bolivia were next on the itinerary.

Then comes the result most likely to travel beyond the post: “Among those who provided feedback, 98% reported the highest levels of satisfaction.” The Spanish original is even clearer about the unit: 98% of the responses received were in the highest satisfaction levels.

That wording is careful. LACNIC does not say that 98% of all attendees were highly satisfied. It does not say that 98% of direct members approved the programme. It narrows the claim to feedback received.

The missing fact is therefore simple: received from how many?

A percentage without a base can describe ten responses or hundreds. It can be a complete census of the people offered a form or a small fraction of them. Nothing on the page shows which. That does not make 98% false. It makes the institutional meaning of 98% unavailable to the reader.

Reach is not a response rate

The more-than-330 figure and the 98% figure sit next to one another, but they do not share a disclosed denominator. The first counts organizations present at gatherings. The second counts feedback responses, or possibly the people or organizations behind those responses. The post does not define the respondent unit.

One organization may have sent several people. One person may have attended more than one meeting. An organization may have attended but not been asked to complete a form. A form may have been offered to every person, one representative per organization or only some participants. Several people from one organization may have answered. The programme may have removed duplicates across cities, or it may have counted each event appearance. Each design can be legitimate. Each produces a different interpretation.

Suppose 98% is calculated from nearly every organization represented. The result would be strong evidence about the programme's reception among the organizations reached. Suppose it comes from a much smaller self-selecting group. It would still be an accurate report of those responses, but it would answer a narrower question. The post does not allow either scenario to be selected.

The same restraint applies to the remaining 2%. Highest levels is plural, but no scale is shown. It might mean the top two categories on a five-point scale, two categories on another scale, or a local classification used for event feedback. The page does not say whether the rest were neutral, dissatisfied, incomplete or missing. Subtracting 98 from 100 does not create a published dissatisfaction rate.

Nor can non-response bias be declared. Bias depends on who had an opportunity to respond, who did respond and whether non-respondents differed in a relevant way. Without those facts, the correct conclusion is not that bias occurred. It is that the page provides no way to measure the risk.

LACNIC already distinguishes other measurement series

The outreach post says face-to-face meetings complement information gathered through surveys. That sentence matters because LACNIC has several measurement channels, not one pool of interchangeable percentages.

Its official 2024 Customer Satisfaction Survey summary reports that 96% of respondents were satisfied or very satisfied. The survey index lists a biennial series going back through 2022, 2020, 2018, 2016, 2014 and 2012. That is a named member-survey programme.

The 26 August post describes feedback attached to 2025 gatherings. It does not say that this feedback used the 2024 survey instrument, sampled the same population or applied the same weighting. The similar percentages cannot be averaged, treated as a trend line or used to validate each other without a methodological bridge.

A separate February 2026 services post demonstrates the same need to preserve units. It reports more than 5,000 support tickets, 97.9% of first responses within two business days, 96.6% of more complex cases resolved within five business days and a separate perception result in which all surveyed users rated a process easy or very easy. Ticket timeliness and survey perception are both useful; they measure different events.

The lesson is not that LACNIC lacks measurements. It is that its public evidence is strongest when every number carries its population, unit and clock.

The programme produced more than a score

The outreach account is valuable even without the percentage. LACNIC says members discussed IPv4 and address transfers, IPv6 deployment, MiLACNIC, account security and membership benefits. Meetings were not only an opinion poll. They were also places to explain tools, diagnose local problems and collect service ideas.

The post names digital signatures for contracts, self-service request-status checks, co-administrators for resources, 2FA updates, a fee calculator and improvements to a security report. It says many of these changes came from ideas that originated in the meetings or were validated there.

That is deliberately broader than a causal claim. An idea can originate in one conversation and be developed through several teams. A feature can already be planned and then be validated by members. A satisfaction response can evaluate an event without endorsing any particular change. The 98% result should not be asked to prove the service lineage, and the service list should not be used to infer the survey base.

LACNIC's February 2025 services account described CRM, support tickets, local-currency payment in Argentina, training and member engagement. It said the prior year's outreach met more than 200 members in four countries. The new post sharpens that picture to 214 member organizations in four named countries. That continuity makes the programme newsworthy. It also makes a stable measurement note more useful: later reports could be compared without guessing whether members, organizations, visits and responses were counted the same way.

A small methods box would be enough

The remedy is not publication of respondent names or raw comments. LACNIC could attach a compact note to the 98% sentence.

It would identify the events and dates included. It would say how many organizations participated, whether repeat attendance was deduplicated and how many people were eligible to answer. It would define whether the respondent unit is a person, organization, completed form or answer to one question. It would give the number of forms distributed, usable responses and response rate.

The note would reproduce the question and scale and explain what highest levels includes. It would say whether several replies from one organization are weighted separately, how partial forms and missing answers are handled, and whether the result is weighted across cities, languages, organization sizes or membership tenure. A rounding rule would explain why the displayed value is 98% rather than a decimal. A version and correction date would protect the record if a duplicate or late form is later found.

Small groups require care. City-level or category-level counts can be suppressed or combined where they risk identifying an organization. Aggregate method does not require public comments, emails or attendance lists.

That note would let three claims coexist cleanly. Reach: more than 330 organizations participated. Response: a disclosed number completed the relevant question. Satisfaction: a disclosed share of those responses fell into defined top categories. None has to pretend to be the other.

Good outreach deserves evidence that can travel

A regional registry operates through repeated relationships. A member may need help with a transfer, an IPv6 plan, a compromised account or an unfamiliar MiLACNIC workflow. Meeting people in their own markets can reveal friction that a central ticket queue misses. Working breakfasts can place operators with similar problems in the same room. Industry events can reach organizations that would not request a formal visit.

Those are credible reasons for the programme. The 98% figure is not needed to prove that conversation has value. Its job is narrower: describe the reception of a defined set of encounters.

Once a percentage is published, it becomes portable. It may appear in an annual report, a Board presentation, a budget case or a future claim that the programme improved. Readers will remember the numerator and forget the qualifying phrase among those who provided feedback. A methods note preserves the qualification when the number moves.

LACNIC has already done the important editorial work by limiting the statement to feedback received. Publishing the response count would complete that discipline. It would not turn outreach into a referendum. It would make one institutional claim legible on its own terms.

Sources