The Question Behind the Question

Suppose you ask 500 voters whether they support a ballot measure and 52 percent say yes. The lazy interpretation is that the measure is ahead. The useful interpretation begins with a fresh stack of questions. Which voters are certain? Which have merely recognized a friendly phrase? What happens when the price is explained? Whom do they trust to explain it? Where do they get their news? Peak Insights lives in that second stack.

The privately held research firm is headquartered in Missoula, Montana, and traces its history to 1981. For most of that history it was called Moore Information, after founder Bob Moore. Today its staff conducts live-phone, online, text-to-online and automated surveys; moderates focus groups and long-form interviews; tests advertisements and messages; builds turnout models; and helps campaigns contact voters. The clients range from Republican Senate campaigns to Intel, Univision, hospital systems, universities, trade groups and nonprofits.

That list can make the company sound like a menu of methods. It is better understood as a machine for converting uncertainty into a sequence. Establish the baseline. Divide the audience into meaningful groups. Test the frame. Test the messenger. Watch what moves. Then tell the client where to spend.

The first result was bad news

In 2018, Washington voters considered Initiative 1631, a proposed carbon fee. The coalition supporting it had more than $15 million, sympathetic headlines and ballot language that made the policy sound comfortably public-spirited. Peak Insights was hired as lead pollster by the opposition campaign. Its first statewide survey found majority support for the measure.

This is the moment when a poll becomes valuable. The campaign could not simply shout louder. It had to decide what argument the undecided middle would hear as relevant, and who could deliver it credibly. Peak ran in-person focus groups, online advertising tests, repeated message-refinement surveys and nightly tracking. The research isolated unaffiliated voters and people who had cut the cable cord - people less likely to encounter traditional political advertising. Those voters responded to a broad pocketbook argument: whatever its label, the proposal could raise energy and fuel costs without guaranteeing a meaningful emissions reduction.

The poll was not a verdict. It was a map of the people who had not reached one.What Peak's I-1631 case study reveals

The measure lost. Peak's telling of the case is, naturally, the vendor's account, not a controlled experiment proving which consultant caused which vote. But the sequence is instructive. What failed first was the opposition's starting position. What changed the strategy was not a clever slogan dreamed up in isolation, but evidence that the contest turned on cost, credibility and a hard-to-reach independent audience.

A pollster with two desks

One Peak desk speaks campaign. President and managing partner Erik Iverson has managed congressional races, chaired the Montana Republican Party and served as lead pollster to Tim Scott's presidential campaign. Other senior employees have run statewide campaigns, worked at national Republican committees and directed voter-contact operations. That experience distinguishes Peak from a survey shop that delivers a deck and disappears. Its political clients describe the firm as part of the strategic team.

The other desk speaks institution. Since 2004, the firm says it has tracked perceptions around Intel manufacturing sites in Arizona, California, New Mexico and Oregon. For a Washington hospital system, it established perception baselines and followed the public response through mergers and market expansion. For Port Blakely, a family-owned forestry company, Peak researched employees, customers, opinion leaders and industry figures over a relationship lasting more than 20 years. A cold-weather outerwear study crossed North America, Asia and Europe.

The two desks share a premise: every institution eventually meets a public it does not control. A candidate meets voters. A factory meets its neighboring towns. A hospital merger meets patients who have never seen the integration plan. A reputation study cannot make the decision for them, but it can reveal where the decision will snag.

What the work costs

Peak publishes no price list. Political filings, however, offer peepholes into the economics. A Texas local campaign reported a $19,500 polling expense in 2024. America PAC reported two same-day survey-research payments of $29,500 and $30,500. A Senate campaign reported a $76,400 polling-services payment in April 2026. Another public-data aggregation counted $300,000 across 11 transactions and six committees in its dataset, with an average payment of roughly $27,000.

A payment is not a standardized project price, a profit figure or the firm's total revenue. Survey mode, sample size, number of waves, modeling, analysis and advisory work can all change the scope.

The business model is therefore conventional consulting with unusually visible fragments: custom projects and ongoing relationships, priced around research design, fieldwork, analysis and advice. The client is not buying certainty. The client is buying a more disciplined way to risk the next dollar - on an advertisement, a campaign, a merger announcement or a long-term reputation program.

The name finally caught up

In spring 2024, Moore Information Group became Peak Insights. The firm added Jeff Glassburner, who had run Tim Scott's Iowa operation, and Sam Lancaster to lead voter contact, identification and modeling. It also hired commercial researchers Larry Jacobs and Tamara Wisinski Wynn from Ipsos. The old name described a founder and an era. The new one described the deliverable.

The expansion matters because survey research has become harder to conduct as people abandon landlines, ignore unknown callers and scatter across media channels. Peak's answer is methodological variety: live interviews when nuance matters, text and online collection when reach matters, focus groups when language matters, and models that combine surveys with voter files, registration patterns, consumer behavior and geography when action matters.

That mix still has limits. A focus group cannot stand in for an electorate. A partisan internal poll should be read with its sponsor in view. A model inherits the blind spots of its data. Message testing can identify persuasion; it cannot make an unpopular product good or a distrusted messenger credible. The method is strongest when the target population is identifiable, the client can act on the result and there is time to test before committing the larger budget.

The parts worth stealing

You do not need a Senate budget to copy the logic. The most portable Peak habit is separating the question you want answered from the decision you actually face. “Do people like this?” is a weak brief. “Which audience is uncertain, which objection blocks it, and which proof makes that objection smaller?” is a research program.

Peak's work with Hispanic voters offers the cleanest warning against lazy segmentation. Its public case material distinguishes Cuban and Puerto Rican audiences in Florida rather than treating “Hispanic voters” as a single block. Its Hometown Lenders case makes the same point in miniature: an advertisement may be memorable, clear, effective and positively received - or memorable for exactly the wrong reason. Those are separate questions, so ask them separately.

There is no romance in this conclusion. Public opinion remains fickle. Sampling remains imperfect. Campaign clients still publish the flattering numbers and bury the painful ones. Yet Peak Insights has endured because institutions keep making expensive decisions in public. Somewhere between the first percentage and the final decision sits a smaller, more useful question: what would have to be true for this person to change their mind?

Go deeper

• Peak Insights website
• Research services
• Peak Insights on LinkedIn
• Erik Iverson on Voices of Montana