The Best Business Openings Start With Attention Others Overlook

Jeff Harmon’s rule for starting a business is to find a “glitch”: a pocket of attention or demand that established players are overlooking. The opening may not look large enough to interest a conventional business. It only has to be large and motivated enough to give a new venture a foothold—and a way to learn.

That logic first surfaced when Harmon was asked to help sell Orabrush, a tongue cleaner invented by dentist Dr. Bob Wagstaff. Wagstaff had boxes of the product in his garage. In a university class, students told him that only about 8% of people wanted to buy it online and advised him to seek a partnership with a large consumer-goods company.

Jeff Harmon saw the 8% differently. Eight percent of the US population, he said, was about 30 million people. If even a fraction of that group wanted the product, it could be enough to sell the stock in Wagstaff’s garage. The point was not that 8% was a large share. It was that the share represented a potentially viable group of customers.

Harmon’s first attempt to reach them was a Facebook page called Kisses, with 1.3 million followers. He persuaded Wagstaff to pay $5,000 for it, then promoted Orabrush to people drawn to a page about thinking of someone they wanted to kiss. The connection was bad breath: the product was not the message so much as a means of avoiding embarrassment. Harmon later renamed the page Orabrush.

The large following did not make the campaign work. Harmon says the team sold very little through the page, and Google ads were also uneconomic: they spent three to five dollars for every dollar of sales. The lesson was not simply to find an audience. The team needed a way to reach people that connected the product to a problem they wanted solved, and a way to do so at a cost the business could sustain.

Harmon found a more promising approach by combining several things he had noticed. A Howcast video about testing for bad breath—with a spoon scraped along the back of the tongue—had about 400,000 views. He had also been watching early YouTuber iJustine, whose jump cuts removed pauses and uninteresting stretches. His roommate Joel Ackerman, a screenwriter and comedian, could help shape a script. Harmon brought in sales experience from door-to-door potato sales and missionary work, where he had learned to handle rejection.

The resulting Orabrush video explained the spoon test, connected bad breath to bacteria on the tongue, and demonstrated how to use the cleaner. Harmon says it cost about $500 to make. His brother Neal built a website, and the team offered a free brush if customers paid shipping, setting the shipping charge to cover both the product and delivery.

The economics depended on timing as much as the ad itself. YouTube had recently launched its ad platform, and Harmon says views cost less than a penny each. Including ads, shipping and the brush, he says the team could spend a dollar and bring in two or three dollars. Cheap traffic gave them enough data to test and improve the offer. As ad prices rose, he says, the team became more efficient because it had already been able to learn from that traffic. Harmon says Orabrush eventually sold for about $10 million.

The specific opening—very cheap YouTube views—belonged to that moment. The broader method is to look for a motivated group that others discount, then find a route to reach it that lets the business learn before the opening disappears. Harmon calls the first foothold a “beachhead”: a limited position the business can hold and expand from.

Harmon’s product-side test is whether a product solves a problem its founder understands as a user. Wagstaff had encountered bad breath among missionaries he supervised in the Philippines and wanted to address it. Drawing on his professional knowledge of surgical scrub brushes, he combined a toothbrush handle, brush and tongue scraper into an early prototype. He was solving a problem he had seen up close, even though the eventual market extended beyond the people he knew.

Harmon calls this the “user method,” drawing on Jeff Schwarting’s book The User Method. He uses Airbnb as another example. In his account, the founders needed rent money during a San Francisco convention. One was a graphic designer; they bought air mattresses at Walmart, inflated them and rented out floor space. The workaround became AirBed & Breakfast. The initial idea came from a problem the founders themselves faced, rather than from trying to invent a market at a distance.

For Harmon, that product insight is separate from the marketing question: where is attention going, and where are existing businesses not looking? A product may answer a real need, but the business still has to find a channel or mechanism that puts it in front of people who might act.

He offers shopping through AI agents as a possible example of an emerging marketing opening. In the scenario he describes, someone asks an agent to buy groceries without searching through websites or clicking through conventional results. A business that lets an agent connect directly to its system, rather than requiring it to scrape information, could serve people choosing to shop that way. Harmon presents this as a potential glitch, not as a settled or universal change in consumer behavior.

Shaan Puri sharpens the competitive logic: an established company may have built its people, skills and processes around a channel that has worked for years. It may give a new channel only a fraction of its attention. A smaller business can instead design the whole experience for that channel and concentrate on it.

Harmon makes a related point about new tools in Hollywood. He says his team is moving faster partly because it uses tools such as Claude, while many in the industry are wary of them or actively fighting their use. His claim is about an uneven willingness to adopt tools, not simply about the tools themselves: an incumbent’s reluctance can create room for a smaller team willing to change its process.

The two kinds of opening reinforce each other but should not be conflated. The user method helps identify a problem worth solving; the attention glitch concerns how to reach people who might care. In Orabrush, the product addressed a problem Wagstaff understood, while cheap early YouTube views helped the team reach potential customers and learn. In both cases, the advantage came from taking seriously a signal that others could dismiss.

Harmon and Puri describe VidAngel as an attempt to solve a problem they wanted addressed themselves: watching films or television with selected sexual content, nudity or strong language removed. Harmon says they saw demand in the edited versions of Game of Thrones episodes that appeared on torrent sites within a day of new episodes. Those versions removed scenes including sex, nudity and rape, and, he says, were downloaded by millions of people. The opportunity, as he saw it, was to offer a legal way to watch content with those scenes removed.

The audience was not necessarily large in proportion to all viewers. Puri observes that in an ordinary group discussing Game of Thrones, few might say they had watched a cleaned-up torrent. Harmon agrees the group was small. The question was whether the people who did want the option were numerous and motivated enough to support a business.

That question shaped an unusual test for The Chosen. The team had a 19-minute film and wanted to learn whether people would support turning it into a series. Rather than cut the film into a conventional short ad, they uploaded the full version as a skippable YouTube ad. At the end, viewers were asked to help fund the series.

Harmon says someone asked why they would upload a 19-minute ad. His answer was that even if only 3% of viewers made it to the end, those might be the people who wanted the series to exist—and might invest in it. The point of the long ad was not to make everyone watch. It was to let viewers leave and see whether a smaller group chose to stay.

Harmon connects that test to the skip-ad button, which he says his team helped develop with YouTube. The platform wanted more advertisers to make work like Orabrush, and its people spent several days with the team in Provo, Utah. Harmon argued that viewers should be able to skip after five seconds, but that advertisers should not have to stop after a fixed 15- or 30-second limit. If the ad held a viewer’s interest, he said, it should be able to continue.

The arrangement made a long ad more acceptable to the platform, in Harmon’s account. He says YouTube would not have wanted an advertiser to demand five uninterrupted minutes in front of its content. A five-second skip gave viewers a choice; the advertiser could then earn more time by keeping them interested. For The Chosen, the length of the film became part of the test: staying through it was a stronger signal of interest than simply being served an impression.

Harmon says 17,000 people invested $11 million in The Chosen. He puts the advertising spend at $800,000, a little less than 8% of the total raise. Director Dallas Jenkins was initially upset, expecting viral marketing to raise the money for free. Harmon’s response was that the agency had spent money building other brands as well. He compared the advertising cost with the 8% to 12% he said an investment bank might charge to raise money.

The ad was not the only fundraising mechanism. Harmon says his friend Derral Eves helped the team run livestreams near the end of the campaign. During the last 24 or 48 hours, Jenkins streamed for an extended period, while notifications appeared when people invested—for example, someone investing $100 from Idaho. The visible updates made it possible for viewers to see that other people were contributing while the campaign continued.

In Harmon’s telling, the campaign’s mechanics were complementary: advertising brought people to the project, the long ad gave them a chance to decide whether they cared enough to stay, and the livestream made contributions visible as they arrived. The team raised enough to make the series. Harmon says the subsequent task was for Jenkins to make a good show, and credits him with doing so. Harmon describes The Chosen as reaching about 200 million viewers worldwide.

Harmon’s work on Orabrush led into the Harmon Brothers ad agency, which worked on campaigns for Poo-Pourri, Squatty Potty, Purple mattress and Lume deodorant. He says Poo-Pourri was doing about $4 million a year when his team began working with it. Purple was close to bankruptcy; one of its founders called Harmon two weeks after the campaign launched and said it had saved the business. Harmon recalls thinking that the agency might not have been paid if the company had failed.

The Poo-Pourri premise made the product’s function part of the joke. The spray is applied before using the toilet; Harmon describes its essential oils as trapping the smell beneath the water. The ad put a polished British woman, dressed neatly and seated on a toilet, in the position of speaking frankly about going to the bathroom. The contrast between the composed presentation and the blunt language made an otherwise awkward product easy to remember.

Harmon says his team spent three or four days developing the ad with Joel Ackerman. Their idea was to create a series of juxtapositions: a well-dressed British woman on a toilet, discussing something people normally avoid talking about. In the ad shown during the conversation, she describes what she has just done and the problem of leaving a shared toilet smelling bad. The humor depends on the product’s use case rather than being detached from it.

Harmon does not describe the creative process as waiting for inspiration. Weeks before a retreat, the agency sent the product to three or four trained comedian-writers. They used it and arrived with rough scripts. On the first day, each writer read a version. By that afternoon or evening, the group chose the strongest basic structure. Over the next three days, they took useful elements from the other scripts and incorporated them into the lead version.

The retreat included the creatives, business leaders, attorneys and other people who needed to approve the work. Harmon says having those stakeholders in the room allowed writers to pitch directly and revise with decision-makers present. Months of back-and-forth by email can neuter a script, he argues; a few days of face-to-face work can produce something the necessary parties are excited to approve.

His division of labor is specific: it is easier to teach a comedian to think like a marketer than to teach a marketer to be funny. The marketing structure comes first—the problem, the solution, the product’s key elements, the credibility behind the claims and the call to action. Then comedians contribute the premise and humor.

But the jokes have a constraint. Harmon told writers to cut anything that did not add to the message. In the campaigns he cites, a joke should bring viewers closer to understanding what the product does and why they might use it. Comedy is not a decorative layer on top of the sales pitch; it is one way the ad communicates the pitch.

For movies, Harmon says, the campaign may need to do more than explain a product. A theatrical release is a time-limited launch, and the team has to build attention for a particular film within that window. His examples include scripted comedy, character-driven social posts and stunts that prompt people to ask what they are seeing.

For Solo Mio, starring Kevin James, the team built a social-media character around Matt Taylor, an art teacher who is stood up at his wedding in Italy. James proposed making the character feel real outside the film: he would post videos of himself painting and speaking in a calm, Bob Ross-like manner while staying in character. Harmon says the team sent people to Long Island to work with James and produce the posts.

The team also tried to “warm up” James’s face in social platforms’ recommendation systems before launching the new account. Harmon’s explanation is that platforms can recognize faces in videos and associate a face in new material with faces viewers have seen elsewhere, without needing the person’s name in the caption. James reposted older material and made other videos to build activity around his face. Harmon says that when the Matt Taylor account launched, the platforms recognized James and some of the new videos began reaching people who had watched his other content.

Harmon says the team produced thousands of clips in search of a few that would take off. He compares the approach to creators who pay people to cut livestreams into short videos: they are taking many swings, looking for a viral clip and trying to ride waves of attention that already exist. He also points to movie reviewer Critical Drinker, whose videos use scenes from well-known films. In Harmon’s account, showing famous actors and scenes can connect those videos to viewers already watching material featuring those people.

The Matt Taylor campaign then extended beyond the account. On Super Bowl Sunday, James appeared in a tuxedo, holding flowers and seated among empty seats at the stadium. His outfit matched the character’s wardrobe in the movie. The team bought seats around him for people with Instagram and TikTok accounts, who posted questions about whether the man was really Kevin James. The image shown during the conversation puts James alone among empty red seats, with crowds visible behind him.

The stunt did not explain itself or make an explicit commercial pitch. Harmon says James appeared on the jumbotron and the campaign drew hundreds of millions of views. He estimates the broader campaign generated more than a billion views. The image gave people a question to pass along—why was Kevin James sitting alone with flowers?—before it told them what the campaign was about.

Harmon says the method takes substantial creative effort. The aim is not simply to post repeatedly or to rely on a recognizable face. For the Solo Mio campaign, the character, James’s existing recognition, the volume of short-form material and the unexplained stadium appearance worked together to draw attention. The campaign could then connect that attention to the film.

Harmon describes Angel Studios as an alternative to entertainment he sees as nihilistic: stories in which no one is good and life appears to have no purpose. Angel’s offering is values-based, he says, rather than exclusively faith-based. The studio promotes faith content around Easter and Christmas, while its broader slate includes westerns and action films. He points to Runner, an action film with Alan Ritchson and Owen Wilson, as an example of a movie with action and explosions but no sex or strong language.

Angel’s Guild gives its audience a role beyond watching. The Guild page shown during the conversation describes members watching a “torch,” voting and leaving feedback for the filmmaker. Harmon says members can vote on films Angel considers, give feedback on rough cuts and veto a film. If the Guild vetoes it, he says, Angel cannot take that film to theaters. Members can also attend Angel releases with the tickets included in the membership.

Harmon describes the Guild as having about 3 million members. At one point he describes the benefit as two movie tickets for every movie Angel releases and says a Guild membership costs $20 a month. Later, he gives a monthly range of $12 to $20. These are Harmon’s descriptions of the membership, not a single consistent price point. His account presents the tickets as a benefit available to members, alongside voting, feedback and the chance to support filmmakers.

Shaan Puri compares the voting process to a prediction market because paying members have more at stake than respondents in a casual online poll. Harmon calls it “the user method at scale.” He says a study nearly a year old at the time of the interview found Angel’s average box office and theater reach higher than those of other independent studios. He also says Angel had been among the top ten studios in the US for the preceding three years. These are Harmon’s descriptions of the study and the company’s position.

He characterizes Angel as a base-hits business that gets home runs occasionally. Sound of Freedom, he says, cost $14 million to make and earned $250 million worldwide at the box office. The production budget did not include marketing: Harmon says Angel spent another $40 million marketing the film.

Those figures describe different parts of a release, not the amount Angel kept. Puri notes that theaters take a substantial share of box-office revenue. Harmon says a good distributor may take half, while a less favorable arrangement may leave the distributor with more. He says Angel takes profits rather than a portion of the other half of box-office receipts, then splits those profits with the filmmaker. According to Harmon, other distributors typically take 12.5% to 30% of that other share.

The Guild also provides a way for members to support filmmakers collectively. Harmon says Guild members have paid $300 million to filmmakers. He says filmmakers receive 50% of the Guild’s profits, while Angel uses the other half to build up the Guild. The figure is presented as money paid out to filmmakers, not as the Guild’s profit or as a box-office total.

Angel’s releases create a different kind of marketing challenge from selling a repeatable consumer product. Harmon describes a theatrical launch as a 90-day period in which the company may spend $5 million to $20 million on advertising for a film, with new releases every four to six weeks. Some films, he says, can represent $10 million to $50 million businesses, and occasionally $100 million businesses. He compares each release to launching a startup with a large advertising budget and a short window to find its audience.

When Puri asks why Angel reported a loss despite member revenue and not funding every film, Harmon says marketing is a major expense and acquiring new members is not cheap. The member system does not remove the costs of reaching audiences or launching films. It changes how Angel involves its audience in judging and supporting projects.

Harmon calls Hollywood a knife fight and says he asks himself every week whether he wants to stay in the business. He connects the difficulty to what Angel is trying to build: an alternative system for making excellent entertainment that reflects values he wants to pass on to his children. He sees the Guild’s members as the company’s strength and as a way to change how films are selected, improved and brought to audiences.

Harmon also describes using AI tools to build prototypes, but says his executive responsibilities repeatedly interrupted that work. He would build a prototype and then get pulled into meetings, fires and other demands. His response was to find someone who could take an experiment forward while he was occupied. If it reached a minimum viable product and showed signs of product-market fit, the product team could take over and build it to scale.

One example is a system for testing movie trailers and scenes. The team can send links to friends and family, assign viewers randomly to trailer A, B or C, and gather data about which version works better. Harmon also says AI lets the team model attrition and customer lifetime value faster, as long as the data is clean.

The tools do not operate without company-specific rules. Angel has built internal “harnesses,” including a brand-guide harness, that can be connected to different models and check outputs against the company’s requirements. Harmon says he often prototypes until something breaks; then the engineering team builds a new harness or other support to address the failure. He also sets agents to work through plans overnight, though he describes them sometimes getting stuck when they need approval to take an action.

Harmon argues that improved productivity does not necessarily mean a company should employ fewer people. If AI makes a team much more effective, he says, it may be able to take on more work and need more people to build and serve what becomes possible. He sees AI-driven layoffs as short-term cost-cutting rather than a way to build long-term value, and says he does not expect AI to produce mass unemployment.

The constraint he describes is more immediate: access to a model is not enough if the executive building with it has no time to carry the experiment forward. A dedicated person who can keep a prototype moving, and systems that check work against company rules, help bridge that gap.

Harmon’s advice to his younger self starts with family. He says marriage and children have mattered more to him than professional success, and urges people not to wait until they believe the timing is perfect to have a family.

His business advice is to avoid treating any one idea or dispute as the only chance to win. A mentor, Steve Oskoui, told him not to spend years fighting over a claim worth less than $5 million. Even if the case were likely to succeed, Oskoui argued, 18 months to three years of litigation could drain the time and energy an innovative person might use to create more value.

Harmon had to apply that advice when a client owed his agency a significant amount of money and threatened legal action. The amount mattered because the agency did not have much money. After talking it through with his brother Neal, Harmon decided not to sue. He says Neal had opened the Bible to a verse about giving a coat as well as a cloak to someone suing for the cloak; the combination of that passage and his mentor’s advice helped him let go of the claim.

Harmon says the decision changed the direction of his life. Rather than spend years on a dispute, he returned to building ideas and the ad agency. His image for the choice is the golden goose: if you are the goose, you can lay more eggs. The money he gave up mattered, but he did not consider it the only opportunity he would have.