# The psychology of B2B ads: how persuasion moves a buyer from noticed to acted on

> A business-to-business (B2B) ad must be noticed, understood, believed and acted on, and each step has its own persuasion levers. This page covers eight in depth, each with a public root: Loewenstein on curiosity, Ein-Gar on small admissions, Kahneman and Tversky on reference points and losses, Bosson on shared dislikes, Cialdini and Goldstein on peers, Hopkins on specific facts and Bain on anxiety. We found no public test that shows any of them lifts the results of B2B ads on LinkedIn or Meta. So find the step where your buyer stalls and use the sheet to pick one bet per brief, not a promise.

Source: https://nimblet.io/guides/psychology-of-b2b-ads. Updated 2026-10-23.

Before a creative review, a team needs words for two questions. Why should this ad move a buyer? At which step could the buyer drop off? This page follows the buyer through four steps: the ad is noticed, understood, believed and acted on. For each step it gives two levers with their public roots, real ads and an evidence label. At the end you get a sheet to paste into your strategy doc, a reading card and two prompts.

The real ads come from the LinkedIn Ad Library and the Meta Ad Library. We filtered to Germany and read them on 22 October 2026. The libraries show no spend and no results, such as clicks or leads ([LinkedIn's help page](https://www.linkedin.com/help/linkedin/answer/a1517918) lists what its library holds). So we never say an ad worked, and run length is at most a weak hint. German ads appear in German with our English translation. A number inside an ad is the advertiser's own claim unless the ad names a source. We do not name the people who post thought leader ads, because we read ads and not people.

Fernmark is an invented product: it lists every software licence and contract, shows each renewal and cancel date and the last use of each seat, and warns the owner before an auto-renewal. Every Fernmark line is constructed. A bracket marks a name or number you must take from your own data.

Three labels mark the evidence. Measured elsewhere means a published study we opened. It did not test B2B ads on LinkedIn or Meta. Practitioner or vendor claim means a recipe or a number from someone who sells ads, tools or training. Our reading means our judgement, not tested. We opened the external pages linked here in October 2026.

## Why does a B2B buyer respond to an ad at all?

Most people who see a B2B ad are not buying yet, so the ad has two jobs. It must be remembered when a need appears. It must also give the buyer a reason to repeat to colleagues. A lever is a way for the ad to do one of these jobs.

The first job is about memory. The B2B Institute at LinkedIn quotes Professor John Dawes on his 95-5 rule: advertising "mainly works by building and refreshing memory links to the brand. These memory links activate when buyers do come into the market." Jenni Romaniuk of the Ehrenberg-Bass Institute adds that remembered brands are the starting point of a purchase. Buyers who look further, she writes, "still show a bias for the brands they already know" ([Romaniuk, June 2022](https://business.linkedin.com/content/dam/lem/business/en/advertise/resources/b2b-institute/b2bi-cepinb2b-final.pdf)). Both texts are frameworks and neither tests an ad. We found Dawes's rule only as the B2B Institute quotes it, so we print no share of buyers.

The second job is about the person. Romaniuk writes that a buyer has "their own desires and ambitions too, such as making a splash to help get promoted, demonstrating skills, or even just not getting fired". Bain reports from three decades of its client studies that, in some purchases, whether a product "can enhance the buyer's reputation or reduce anxiety" plays a large role ([Bain, 2018](https://www.bain.com/insights/the-b2b-elements-of-value-hbr/)). The same article holds the limit. In its survey of more than 2,300 decision makers in two industries, product quality, expertise and responsiveness were the strongest predictors of supplier loyalty. Insurance buyers named reduced anxiety among their top wants, yet other elements predicted loyalty better ([reprint of the article](https://ellisonchair.tamu.edu/files/2020/06/The-B2B-Elements-of-Value.pdf)). The person also sits in a buying group of five to 16 people, as [Gartner](https://www.gartner.com/en/newsroom/press-releases/2025-05-07-gartner-sales-survey-finds-74-percent-of-b2b-buyer-teams-demonstrate-unhealthy-conflict-during-the-decision-process) puts it. The guide on [B2B ad angles](/guides/b2b-ad-angles) covers who carries which risk.

Levers matter because most B2B ads draw a weak emotional response in System1's tests. The B2B Institute and System1 rated 1,700 B2B ads, and 77 percent scored one star. System1 expects a one-star ad to bring no market share growth, even with a significant budget ([compendium](https://business.linkedin.com/content/dam/me/business/en-us/marketing-solutions/resources/pdfs/b2b-creative-v2.pdf)). The compendium's tech sample is 587 US ads of 5 to 120 seconds, rated in a forced exposure setting, so the figure says little about a feed image. Kantar says B2B ads draw a more rational response, with stronger persuasion and relevance but weaker attention, engagement and brand connection ([Kantar, April 2024](https://www.kantar.com/north-america/inspiration/advertising-media/beyond-rationality-how-emotion-drives-b2b-advertising-success)). Its page gives no sample size.

Four terms sit close together, so keep them apart.

| Term | The question it answers | Where to read more |
| --- | --- | --- |
| Angle | Why should this buying role say yes? | [B2B ad angles](/guides/b2b-ad-angles) |
| Hook | What does the buyer meet first? | [B2B ad hooks](/guides/b2b-ad-hooks) |
| Proof | What makes the claim believable? | [Proof points in B2B ads](/guides/proof-points-in-b2b-ads) |
| Lever | Which mechanism moves the buyer? | This page |

## How does persuasion move a B2B buyer from noticed to acted on?

An ad persuades in steps. First the buyer notices it. Then the buyer understands the point, believes it and, last, acts. The ad can lose the buyer at any of these steps, and each step asks for a different kind of help. The four steps are our reading. They follow the old advertising sequence of attention, interest, desire and action, which textbooks credit to E. St. Elmo Lewis (textbook claim, not checked here). We put belief where desire was, because a B2B buyer must defend a choice to colleagues before acting on it.

A lever is one way to help the buyer through one step. Many lists of levers exist. Cialdini's [page](https://www.influenceatwork.com/7-principles-of-persuasion/) lists seven principles: reciprocity, scarcity, authority, consistency, liking, social proof and unity. Unity is the later addition, and the page gives it no description.

This page covers eight levers in depth, two per step. We chose them because each one has a public root that we could open or name, and because we found a real B2B ad for each one. The choice is ours. Other levers work too.

| Step | The buyer's question | Levers on this page | Public root |
| --- | --- | --- | --- |
| Noticed | Is this for me, and why look now? | An information gap. A small admission. | Loewenstein, 1994. Ein-Gar and colleagues, 2012. |
| Understood | What is the point, in words I can repeat? | Framing. A common opponent. | Kahneman and Tversky, 1979. Bosson and colleagues, 2006. |
| Believed | Why should I trust this claim? | Peers like the buyer. Specific facts. | Cialdini; Goldstein, Cialdini and Griskevicius, 2008. Hopkins, 1923. |
| Acted on | Is the next step safe for me now? | Loss framing. A smaller first step. | Kahneman and Tversky, 1979. Bain, 2018, and Hopkins, 1923. |

Reciprocity, consistency and liking are not on this page. A free report in an ad uses reciprocity in practice. We treat it as an offer choice, which the [brief template](/guides/b2b-creative-brief-template) covers. Scarcity sits close to loss framing.

## How does the awareness stage change the first line?

The step tells you which help the buyer needs. The awareness stage tells you where the first line must start. The five stages come from Eugene Schwartz's book Breakthrough Advertising, first published in 1966 (textbook claim, not checked here; we read summaries such as [this one](https://optimizesmart.com/blog/schwartz-five-stages-of-awareness-in-marketing/)). The right column is our reading of the summaries.

| Stage | The reader | Where the first line starts |
| --- | --- | --- |
| Unaware | Has no problem in mind | A situation from the reader's week, with no product yet |
| Problem-aware | Feels the pain and sees no fix | The pain, in the words the reader uses for it |
| Solution-aware | Knows fixes exist and compares kinds | Why your kind of fix is different |
| Product-aware | Knows you and doubts the fit | The doubt, then the proof that answers it |
| Most-aware | Is ready and wants a reason now | The offer and the next step |

A cold B2B audience is mostly unaware or problem-aware. This follows from Dawes's point that most buyers are not in the market, and it is our reading. A stage belongs to the reader, not to the product. One reader can sit at different stages for different concerns. The [angles guide](/guides/b2b-ad-angles) shows how to map this for your own market.

Here is a first line set to its stage.

| Ad | The line | Our reading |
| --- | --- | --- |
| [GitLab, LinkedIn message ad](/guides/ads/1600409724.jpg) | "Most leadership teams I talk to have moved past 'should we use AI?' Now they're asking how to scale it across the software lifecycle without creating new risk." | The reader is solution-aware: the team already uses AI. The first sentence says where the reader's thinking is today. The second gives the next question, with the risk in it. It adds "No sales pitch, just industry and peer conversations." The library preview shows the merge field %FIRSTNAME% unresolved. |

## How does an ad get noticed?

A buyer scrolls past most ads in a feed. To be noticed, an ad must give this reader a reason to stop. Kantar's point above applies here: B2B ads tend to score weaker on attention than on persuasion.

### An information gap

The root. George Loewenstein defines curiosity as "a form of cognitively induced deprivation that arises from the perception of a gap in knowledge or understanding" ([Loewenstein, 1994](https://www.cmu.edu/dietrich/sds/docs/loewenstein/PsychofCuriosity.pdf)). In his account, the pull is strongest when the reader knows most of the answer. A person who knows 47 of 50 state capitals frames it as not knowing three. A person who knows three frames it as knowing three. He also notes that satisfied curiosity often disappoints.

In a B2B ad. Ask a question about the reader's own work that the reader knows enough to feel. A question about your product is a weaker pull, because the reader knows little about it. Write the answer before you write the question, and check that the answer is worth the click. For readers new to the field, start from a situation they recognise instead.

Fernmark line (constructed): "Can you name the software contract that renews first this quarter, and who owns it?"

| Ad | The line | Our reading |
| --- | --- | --- |
| [Pipedrive, LinkedIn video ad](/guides/ads/915417333.jpg) | "Which should come first: the discovery call or the product demo?" Later: "Watch this short clip to find out what actually works best." | Sales teams know this debate, so the reader holds most of the answer and sees the missing part. The ad withholds it. We did not watch the clip, so we cannot say whether it earns the click. |

### A small admission

The root. Ein-Gar and colleagues ran four studies in labs and in the field. Adding a small negative detail to an otherwise positive description raised appeal when people processed it with little effort, and when the negative came after the positive ([abstract](https://en-coller.tau.ac.il/sites/coller-english.tau.ac.il/files/RP_222_Ein-Gar.pdf)). The literature calls this the blemishing effect.

In a B2B ad. Put one true limit after the main promise, and keep it small. A reader who expects a vendor to praise itself meets something else, and looks again. Humour can break the same expectation. A joke helps a B2B ad when it carries the argument and comes from something the profession laughs at in its own week. Both steps from the studies to feed ads are our reading.

Fernmark line (constructed): "Fernmark cannot cancel a contract you already renewed. It can warn you before the next one."

| Ad | The line (our translation) | Our reading |
| --- | --- | --- |
| [Snowflake, LinkedIn document ad](/guides/ads/1633878686.jpg) | "Snowflake's leadership share the exact framework behind their GTM AI transformation, including what didn't work. Free ebook." (GTM means go-to-market. AI means artificial intelligence.) | "Including what didn't work" is a small admission placed after a large promise. That is the order that raised appeal in the lab studies. |
| [HubSpot-promoted post, LinkedIn](/guides/ads/1577956593.jpg) | "Die KI-Investitionen deutscher Unternehmen sind inzwischen ungefähr so hoch wie der Berliner Fernsehturm. Die messbaren Ergebnisse davon passen locker in den Fahrstuhl." (German companies' AI spending is now about as tall as the Berlin TV tower. The measurable results fit easily in the lift.) | A joke that carries the argument: high spend against small results. The reader gets the point in the punchline, so no second line is needed to explain it. |

## How does an ad get understood?

A noticed ad still fails if the buyer cannot say its point back. In a buying group, the point must also survive when one person retells it to another.

### Framing

The root. Kahneman and Tversky's prospect theory says people judge an outcome as a gain or a loss from a reference point, not as a final state ([Econometrica, 1979](https://www.econometricsociety.org/publications/econometrica/1979/03/01/prospect-theory-analysis-decision-under-risk)). Their later work on framing showed that two descriptions of the same facts can lead people to different choices (Tversky and Kahneman, Science, 1981; textbook claim, not checked here).

In a B2B ad. Framing is your choice of what the buyer compares against. The ads below show three forms. One sets today's tool next to the new one. One moves the cause of a pain from people to a plan. One borrows a familiar situation, so an abstract problem takes a shape the buyer already knows. For pictures, Harry Dry's copy test asks "Can you visualize it?" ([Demand Curve, 8 August 2024](https://www.demandcurve.com/newsletters/growth-newsletter-202)). A recipe in [Exit Five](https://exitfive.com/newsletter/b2b-paid-ads-that-dont-suck-proven-creative-recipes-that-sell-anything-even-boring-products/) takes the customer's own phrase literally, such as "email ping pong". That is a good source for a picture, because buyers already use the phrase.

Fernmark lines (constructed): "Today: a renewal notice in an old inbox. With Fernmark: every cancel date on one list." Or: "Every auto-renewal is a gym membership for people who stopped coming."

| Ad | The line (our translation) | Our reading |
| --- | --- | --- |
| [monday.com, LinkedIn single image ad](/guides/ads/1578255844.jpg) | "Board statt Tabelle" (Board instead of spreadsheet). "Aufgaben, Verantwortliche, Deadlines und Status in einem Board. Deine Projekte brauchen mehr als eine Tabelle." (Tasks, owners, deadlines and status in one board. Your projects need more than a spreadsheet.) | The reference point is the spreadsheet most project owners use today. The board is described by four things it holds, so a reader can state the difference in one sentence. |
| [monday.com, LinkedIn document ad](/guides/ads/1588880964.jpg) | "Dein Team ist nicht zu langsam. Es ist überplant." (Your team is not too slow. It is overplanned.) Then: "So planst du Ressourcen auf drei Ebenen und erkennst Engpässe, bevor sie bremsen." (This is how you plan resources on three levels and spot bottlenecks before they slow you down.) | The blame for slow delivery moves from the people to the plan. A manager hears that the problem is not them. The ad cannot check whether the reader's team is overplanned. |
| [Post promoted by Celonis, LinkedIn](/guides/ads/1605611964.jpg) | "Picture this: you bring in a temp through a staffing agency, and over two years they learn exactly how your business runs." Then: "the agency owns everything that temp learned about how your company operates." | The parts map one to one. The temp is the AI agent, the agency is the AI vendor, and what the temp learned is company context. The post decodes the picture two paragraphs later: "a lot of companies are signing it for their AI agents". It also frames lock-in as a loss. The post is marked #ad. |

### A common opponent

The root. Across two surveys and an experiment, people who shared a negative attitude about a third party felt closer than people who shared a positive one. In the experiment this held only for weak attitudes ([Bosson and colleagues, 2006](https://doi.org/10.1111/j.1475-6811.2006.00109.x)). The third party was a person, not a process or a habit.

In a B2B ad. Name a chore, a habit or a routine the reader is tired of, and stand next to the reader against it. The study found the bond for mild dislikes, so we would pick a small, shared annoyance and leave people out of it, the reader most of all. The [Exit Five issue](https://exitfive.com/newsletter/b2b-paid-ads-that-dont-suck-proven-creative-recipes-that-sell-anything-even-boring-products/) above has a recipe of this kind. The step from people to processes is our reading.

Fernmark line (constructed): "Renewal notices go to the person who signed up years ago. The bill goes to someone else."

| Ad | The line (our translation) | Our reading |
| --- | --- | --- |
| [Miro, LinkedIn single image ad](/guides/ads/1644816336.jpg) | "Proximie's hardest problem was never engineering the product. It was deciding what to build next without defaulting to the loudest voice in the room." | The opponent is a meeting habit, not a person or a rival. In many teams the loudest voice is the reader's boss, or the reader, so the habit is safer to name than the person. The post also names a customer, which is proof (see the [proof guide](/guides/proof-points-in-b2b-ads)). |
| [Workday, LinkedIn single image ad](/guides/ads/1640400306.jpg) | "Niemand wird Recruiter, um ständig Kalendern und Postfächern hinterherzujagen." (Nobody becomes a recruiter to keep chasing calendars and inboxes.) Then: "Die Entscheidung, wer eingestellt wird, liegt weiterhin bei Ihnen." (The decision on who gets hired stays with you.) Last line: "Weniger Admin, mehr Zeit für Ihre Recruiter." (Less admin, more time for your recruiters.) | The opponent is the admin of calendars and inboxes. The first line names the craft recruiters chose and puts the admin outside it. The second line keeps the hiring decision with the reader, which makes the next step safer. The last line shows that the reader leads recruiters. It speaks to one role, and a buying group has other people. |

## How does an ad get believed?

A B2B buyer expects a vendor to praise its own product, so the buyer gives vendor claims little weight. Belief comes from people like the buyer, or from facts the buyer can check.

### Peers like the buyer

The root. Cialdini's page states the principle: "Especially when they are uncertain, people will look to the actions and behaviors of others to determine their own" ([Cialdini](https://www.influenceatwork.com/7-principles-of-persuasion/)). In two field experiments in hotel rooms, a sign saying most guests reuse towels beat a standard environmental sign. A sign about guests "in this room" did better than the other norm signs ([Goldstein, Cialdini and Griskevicius, 2008](https://carlsonschool-d8.prd.umn.edu/sites/carlsonschool.umn.edu/files/2018-10/118359.pdf)). The figures are in the evidence section.

In a B2B ad. The closest B2B match to "in this room" is a peer of the reader's size, field and role. That step is our reading. A peer's quote or result needs a source, a date and permission. The [proof guide](/guides/proof-points-in-b2b-ads) ranks the kinds of proof.

Fernmark line (constructed): "[Role] at [company type, size]: [real quote about a renewal]. [Source, date, permission]"

| Ad | The line (our translation) | Our reading |
| --- | --- | --- |
| [monday.com, LinkedIn single image ad](/guides/ads/1578413474.jpg) | "Über 245.000 Teams. Eine Plattform." (More than 245,000 teams. One platform.) | A raw total. It says many teams use the product, and it says nothing about teams like the reader's. |
| [Personio, LinkedIn video ad](/guides/ads/1585022103.jpg) | "Ein HR-System für die Fertigung" (One human resources system for manufacturing). Body: "'Echte Digitalisierung.' So beschreibt D+H den Wechsel auf Personio." (Real digitalisation. That is how D+H describes the move to Personio.) | It names a customer and a segment, manufacturing. By the hotel logic, a manufacturer is the closer match for a manufacturing reader. That step is our reading, and we have no results for either ad. |

### Specific facts

The root. Claude Hopkins wrote in 1923: "Platitudes and generalities roll off the human understanding like water from a duck." He added: "Specific facts, when stated, have their full weight and effect" ([Scientific Advertising](https://archive.org/details/scientific-advertising)). Cialdini's page adds the source: "people follow the lead of credible, knowledgeable experts". A fact that only an expert could know does both jobs.

In a B2B ad. Give one exact figure and say what it counts and where it comes from. If you hold data that nobody else holds, show it and name the sample. A cause and effect sentence works in the same way. It gives the buyer a reason, and a reason is easier to check than a bare promise. Harry Dry's three tests in the same Demand Curve issue help here: can the reader picture it, can it be proven wrong, and can nobody else say it?

Fernmark lines (constructed): "[N] seats, last used [N] days ago, cancel by [date]." With data you hold: "We read [N] software contracts from [type of company]. [One finding about auto-renewal clauses.]"

| Ad | The line (our translation) | Our reading |
| --- | --- | --- |
| [Lexware, thought leader ad](/guides/ads/1584431403.jpg) | "100 Euro gespendet, rund 65 Euro bezahlt. Den Rest trägt das Finanzamt, und kaum jemand rechnet das je nach." (100 euros donated, about 65 euros paid. The tax office carries the rest, and hardly anyone ever checks the sum.) Lower in the post: "Die 65 Euro gelten zum Beispiel als Single mit 50.000 Euro zu versteuerndem Einkommen." (The 65 euros apply, for example, to a single person with 50,000 euros of taxable income.) | The number is in plain euros and the reader can check it. The post states its base further down. That is the habit to copy: an exact figure that says what it counts. |
| [Freshworks, LinkedIn document ad](/guides/ads/1597821666.jpg) | "43.6% now use AI across service workflows, up from 32%. 48.3% plan to spend more on it. Swipe for what the shift looks like." | The figures come from data the advertiser holds. The text we read gives no sample size or source. As a sign of expertise, an own survey works: it shows the advertiser watches the field. As proof of a product result it does not (see the [proof guide](/guides/proof-points-in-b2b-ads)). A named expert's post shows expertise by its form: see [thought leader ads](/guides/linkedin-thought-leader-ads). |
| [Snowflake, LinkedIn message ad](/guides/ads/1635840686.jpg) | "The gap usually comes down to data quality and context. Without a semantic layer that defines what your data means, AI models hallucinate and fail to earn trust." | A cause and effect sentence. The message credits three companies with the same fix ("SAP, Snowflake and AWS are solving this together"), so the reason is a shared idea and not one vendor's. |

## How does an ad get acted on?

Most B2B ads ask for a step: a click, a download, a demo, a trial. The buyer weighs that step against what it could cost them at work.

### Loss framing

The root. Prospect theory gives a value function that is steeper for losses than for gains ([Kahneman and Tversky, 1979](https://www.econometricsociety.org/publications/econometrica/1979/03/01/prospect-theory-analysis-decision-under-risk)). How much steeper is debated, and the evidence section has the numbers. Write "weighs more", not "weighs twice", in a brief.

In a B2B ad. Name what the current way costs, in a figure the buyer can check against their own records. Put the next step in the same ad. A meta-analysis of fear appeals, in the section on backfire below, found the effect grew when the message said what people can do.

Fernmark line (constructed): "[N] seats sat unused last year at [amount] a year. The next renewal date is [date]."

| Ad | The line (our translation) | Our reading |
| --- | --- | --- |
| [Adobe for Business, LinkedIn carousel ad](/guides/ads/1605684644.jpg) | "Repurposing assets consumes 41% of creative teams' time", then "leaving less room for new content creation, personalisation, and experimentation." | The cost is time. The 41 percent has no source in the text. The next step appears elsewhere in the card text ("Teams need AI-driven systems."). |
| [Lexware, Meta video ad](/guides/ads/1078108978065275.jpg) | "Wenn die Buchhaltung kippt, kippt alles", then "verpasste Fristen, Mahnungen, unnötiger Stress" (When bookkeeping tips over, everything tips over; missed deadlines, payment reminders, needless stress). | The losses are events in the reader's week. The ad then adds "Über 400.000 Kunden vertrauen darauf" (More than 400,000 customers trust it). Both numbers in these two ads are the advertisers' own claims. |

### A smaller first step

The root. Bain reports that, in some purchases, whether a product can "reduce anxiety" plays a large role ([Bain, 2018](https://www.bain.com/insights/the-b2b-elements-of-value-hbr/)). Romaniuk lists "not getting fired" among a buyer's own wants. Hopkins wrote in 1923 that guarantees "have ceased to be impressive", yet one firm "made a fortune" by offering a dealer's signed promise to pay back ([Scientific Advertising](https://archive.org/details/scientific-advertising)).

In a B2B ad. Make the first step small, cheap and easy to undo: a trial, a check of the buyer's own data, a short pilot. Say how long the step takes. The buyer then risks little by trying, and has a step to suggest to colleagues.

Fernmark line (constructed): "Upload one export. See your next renewals in [time]. Delete the data if it does not help."

| Ad | The line (our translation) | Our reading |
| --- | --- | --- |
| [sevDesk, Meta video ad](/guides/ads/28011665928529647.jpg) | "Tabellen für Rechnungen und Ausgaben werden schnell unübersichtlich. sevdesk ersetzt die manuelle Arbeit und bringt alles an einen Ort. Jetzt kostenlos testen." (Spreadsheets for invoices and expenses quickly become hard to follow. sevdesk replaces the manual work and brings everything to one place. Try it free now.) Card: "Keine Kreditkarte nötig." (No credit card needed.) | The step is small and reversible: a free trial, and no card to cancel later. That removes two risks at once, a cost and a commitment. |

## How does persuasion backfire with B2B buyers?

In our reading, persuasion backfires with B2B buyers in three ways. Each one comes from how they buy.

The buyer carries a personal risk. A message that makes the buyer feel exposed can raise that risk instead of lowering it. A 2015 meta-analysis covered 127 articles, 248 samples and 27,372 participants. It found that fear appeals changed attitudes, intentions and behaviour on average (effect size d = 0.29). The effect grew when the message included efficacy statements, which say what people can do about the threat ([Tannenbaum and colleagues](https://pubmed.ncbi.nlm.nih.gov/26501228/)). We read the abstract only, and it names no B2B ads. A promise that the product cannot keep moves the risk onto the buyer who repeated it to colleagues.

The buyer knows the field. A 2020 paper tested loss aversion in five samples with 17,720 participants. People at every knowledge level showed it, and people with more domain knowledge showed less ([Mrkva and colleagues](https://researchportal.bath.ac.uk/en/publications/moderating-loss-aversion-loss-aversion-has-moderators-but-reports/)). The samples were consumers. Our reading: an expert tests each figure and each device against what they know, and a trick they see through costs more trust than a plain ad. Humour has the same edge. A Marketing Week column cites Kantar for a fall in ads with humour from 53 percent in 2000 to 34 percent in 2024. It also cites System1 data that put top-quartile laugh ads at about five times the extra share of voice (ESOV) efficiency of bottom-quartile ones ([Marketing Week, 8 May 2026](https://www.marketingweek.com/humour-b2b-medicine-pizza/)). The author co-founded an agency, and the column is opinion.

The buyer does not decide alone. Gartner found that content built on individual-level relevance had a 59 percent negative impact on buying group consensus, while content tailored to the group had a 20 percent positive impact ([Gartner](https://www.gartner.com/en/newsroom/press-releases/2025-05-07-gartner-sales-survey-finds-74-percent-of-b2b-buyer-teams-demonstrate-unhealthy-conflict-during-the-decision-process)). Gartner surveyed 632 buyers about content from sellers, not about ads. Our reading: a line that flatters one role can split the group, so give the reader something to say to colleagues. On Meta the reader is a stranger in leisure time, and a line about status reads as pressure ([LinkedIn ads vs Facebook ads for B2B](/guides/linkedin-ads-vs-facebook-ads-b2b)).

## One ad, read with the card

Use one reading card per ad. Fill it by hand, or let the first prompt below fill most of it. The filled card after it is for the Workday ad. It is our reading. A model did not produce it.

```template
LEVER CARD
Ad: [advertiser, platform, format, library entry or ID, date read]
First line, word for word: "[quote]"
Lead lever: [a lever from the sheet on this page, or another you can name and source]
Words that carry it: "[quote]"
Support lever: [a lever, or none]
Words that carry it: "[quote, or none]"
Step it works on: [noticed / understood / believed / acted on]
Risk the ad speaks to: [one sentence, or "not clear from the ad"]
Awareness stage it assumes: [stage], because "[quote]"
Condition from the lever's root, and whether the ad meets it: [one or two sentences]
Evidence label for the lead lever: [measured elsewhere / practitioner or vendor claim / our reading]
Not known from the ad: [whether it worked, because the library shows no spend or results]
```

```text
EXAMPLE ONLY. Filled by hand from a real ad. Our reading.

LEVER CARD
Ad: Workday, LinkedIn single image ad, read 22 October 2026
First line, word for word: "Niemand wird Recruiter, um ständig Kalendern und Postfächern hinterherzujagen."
Lead lever: a common opponent
Words that carry it: "Kalendern und Postfächern hinterherzujagen" (chasing calendars and inboxes)
Support lever: a smaller first step
Words that carry it: "Die Entscheidung, wer eingestellt wird, liegt weiterhin bei Ihnen." (the decision stays with you)
Step it works on: understood, because the ad tells the reader what the problem is (admin) before it names a fix
Risk the ad speaks to: a recruiting lead fears that a scheduling tool takes the hiring decision from the team (our reading)
Awareness stage it assumes: problem-aware, because the chore is named and the fix comes second
Condition from the lever's root, and whether the ad meets it: the bond in the study held for mild dislikes. Calendar admin is a mild dislike that recruiters share, and the ad blames the chore, not the recruiter. The ad speaks to the recruiting lead and names no other role in the buying group.
Evidence label: measured elsewhere (about people, not chores), our reading for B2B
Not known from the ad: whether it worked. The library shows no spend or results.
```

## How strong is the evidence for each lever?

Weak, and uneven. For no lever did we find a public test on B2B ads on LinkedIn or Meta. The roots are lab and field studies on consumers. The B2B evidence is correlation from award cases, panel ratings and commissioned surveys. The table gives each lever a label and a limit.

| Lever | Label | What we opened | Limit |
| --- | --- | --- | --- |
| An information gap | Measured elsewhere | [Loewenstein, 1994](https://www.cmu.edu/dietrich/sds/docs/loewenstein/PsychofCuriosity.pdf) and [Ghibellini and Meier, 2025](https://www.nature.com/articles/s41599-025-05000-w). | Theory and lab tasks. The meta-analysis found no memory advantage for unfinished tasks. |
| A small admission | Measured elsewhere, plus vendor claims | [Ein-Gar and colleagues, 2012](https://en-coller.tau.ac.il/sites/coller-english.tau.ac.il/files/RP_222_Ein-Gar.pdf). Kantar and the B2B Institute compendium on attention. | The lab studies test product descriptions, not feed ads. Kantar gives no sample size. The compendium rates video ads. |
| Framing | Measured elsewhere for the reference point. Practitioner claim for pictures. | [Kahneman and Tversky, 1979](https://www.econometricsociety.org/publications/econometrica/1979/03/01/prospect-theory-analysis-decision-under-risk). Harry Dry's test ([Demand Curve](https://www.demandcurve.com/newsletters/growth-newsletter-202)) and the [Exit Five](https://exitfive.com/newsletter/b2b-paid-ads-that-dont-suck-proven-creative-recipes-that-sell-anything-even-boring-products/) recipe. | Choices about money in lab samples. The 1981 framing paper and the research on pictures and memory, dual coding, are textbook claims, not checked here. The Exit Five results come from the presenter, and the page shows no account data, so we do not repeat them. |
| A common opponent | Measured elsewhere, plus a practitioner recipe | [Bosson and colleagues, 2006](https://doi.org/10.1111/j.1475-6811.2006.00109.x). The recipe is in the same Exit Five issue. | The targets in the study were people, not processes. The recipe's results come from the presenter. |
| Peers like the buyer | Measured elsewhere | [Goldstein, Cialdini and Griskevicius, 2008](https://carlsonschool-d8.prd.umn.edu/sites/carlsonschool.umn.edu/files/2018-10/118359.pdf). | Hotel towels. It shows that similar others help. It does not test B2B. |
| Specific facts | Practitioner claim | [Hopkins, 1923](https://archive.org/details/scientific-advertising), Harry Dry's three tests, and Cialdini's [page](https://www.influenceatwork.com/7-principles-of-persuasion/) on authority. For B2B, see the survey in the [thought leader guide](/guides/linkedin-thought-leader-ads). | Examples, not data. The Cialdini page is a summary by his firm, and we did not open the studies behind authority. |
| Loss framing | Measured elsewhere, debated | Four papers, below. | Money and product choices in consumer and lab samples. No test on ads. |
| A smaller first step | Practitioner claim | [Hopkins, 1923](https://archive.org/details/scientific-advertising), and the Bain survey. | No controlled test. Bain asks about suppliers, not ads. |

### What do the academic roots say?

Loss aversion. Kahneman and Tversky's prospect theory gives a value function that is steeper for losses than for gains ([Econometrica, 1979](https://www.econometricsociety.org/publications/econometrica/1979/03/01/prospect-theory-analysis-decision-under-risk)). The shorthand "losses weigh twice as much" is a rounded textbook figure. A 2024 meta-analysis of 607 estimates from 150 articles puts the mean at 1.955, with a 95 percent interval from 1.820 to 2.102 ([Brown, Imai, Vieider and Camerer](https://www.aeaweb.org/doi/10.1257/jel.20221698)). Gal and Rucker reviewed the evidence in 2018 and wrote that it "does not support that losses, on balance, tend to be any more impactful than gains" ([abstract](https://doi.org/10.1002/jcpy.1047)). The 2020 paper by Mrkva and colleagues, in the section on backfire, found loss aversion at every knowledge level and less of it in experts. So the effect is real in many studies and debated as a general law.

Curiosity and open loops. Loewenstein reports that "mystery" ads, which reveal the product only at the end, produced greater later brand recognition than matched ads that named the product first. The study is by others (Fazio, Herr and Powell, 1992), and we did not open it. The idea that unfinished items are remembered better, the Zeigarnik effect, failed in a 2025 meta-analysis that "found no memory advantage for unfinished tasks but found a general tendency to resume tasks" ([Ghibellini and Meier](https://www.nature.com/articles/s41599-025-05000-w)). For an ad, the pull of an open loop is the urge to finish it, which is the click.

Social proof. In two field experiments in hotel rooms, a sign saying most guests reuse towels beat a standard environmental sign, 44.1 percent against 35.1 percent in the first experiment. In the second, a sign about guests "in this room" reached 49.3 percent, against 42.8 percent for three other norm signs combined ([Goldstein, Cialdini and Griskevicius](https://carlsonschool-d8.prd.umn.edu/sites/carlsonschool.umn.edu/files/2018-10/118359.pdf)).

Small admissions. In the abstract's account of the Ein-Gar studies, putting the negative first, or making people work to process it, meant the effect did not appear ([abstract](https://en-coller.tau.ac.il/sites/coller-english.tau.ac.il/files/RP_222_Ein-Gar.pdf)). A short admission after a promise fits that order, as in the Snowflake ad above. The step from product descriptions to ads is our reading.

Shared dislikes. Bosson and colleagues found the bonding effect of a shared negative attitude in two surveys and an experiment, and in the experiment only for weak attitudes ([Bosson and colleagues](https://doi.org/10.1111/j.1475-6811.2006.00109.x)). The step to a common opponent that is a process or a chore is our reading.

Specific facts and guarantees. Hopkins gave this example of a specific claim: "Say it gives three and one-third times the light and people realize that you have made tests and comparisons." His lines on guarantees are quoted in the section on the smaller first step ([Scientific Advertising](https://archive.org/details/scientific-advertising)). These are a practitioner's claims, with examples and no data.

### What do B2B studies say, and what are their limits?

| Source | What it says | Limit |
| --- | --- | --- |
| [B2B Institute and System1 compendium](https://business.linkedin.com/content/dam/me/business/en-us/marketing-solutions/resources/pdfs/b2b-creative-v2.pdf) | In 587 US tech ads, 71 percent scored one star. An everyday setting (+32.8 percent), fast cuts (+19.7) and a story arc (+14.3) went with higher positive emotion. The business setting used in 71 percent of ads showed +2.9 percent. | Viewers rated ads of 5 to 120 seconds in forced exposure. Emotion is a score, not sales. LinkedIn's own research unit published it. |
| [B2B Effectiveness Code](https://business.linkedin.com/content/dam/lem/business/en/advertise/resources/b2b-institute/the-b2b-effectiveness-code-27-october.pdf) (James Hurman, October 2021) | 435 B2B effectiveness award papers from the WARC database, 2010 to 2021. B2B work skews to short-term, rational, tightly targeted campaigns. It cites a 46 and 54 percent split of spend between long-term brand building and activation. | Award papers are written by the entrants. The split is Binet and Field's, quoted by the Code. |
| [LinkedIn blog page](https://www.linkedin.com/business/marketing/blog/content-marketing/why-b2b-marketing-needs-emotion) on Binet and Field (10 November 2019) | Prints "7x more very large business effects" for emotional campaigns in the long term. Says rational campaigns gave roughly twice the short-term activation effects. | The page names no sample, period or method for the 7x. It cites data from the Institute of Practitioners in Advertising (IPA) awards. We do not repeat the figure as a fact. |
| [Kantar, Allyson Wuensch, 23 April 2024](https://www.kantar.com/north-america/inspiration/advertising-media/beyond-rationality-how-emotion-drives-b2b-advertising-success) | B2B ads draw a more rational response. Persuasion and relevance run stronger, attention and engagement weaker. Ads that avoid typical office settings are more likely to have impact. | Draws on Kantar's LINK database. The page gives no sample size. |
| [Brand Finance and System1, 14 May 2025](https://brandfinance.com/insights/feelings-vs-function-the-power-of-emotional-advertising-in-b2b-brand-building) | One Adobe ad scored 5.3 stars, a level only 1.1 percent of ads in its category reach. Another Adobe ad scored 1.0 star, a level 65 percent reach. | Two ads, no sample size, correlation only. Brand Finance describes itself as a brand valuation consultancy. |
| [LinkedIn and Magna, as MediaPost reports it](https://www.mediapost.com/publications/article/398528) (14 August 2024) | In controlled tests of 67 ads from 26 brands, shown to 1,773 users in 3,024 exposures, creative ads got 53 percent more viewing time and an average of 61 percent more likes. When decision makers saw an ad as creative, they were 40 percent more likely to consider the brand. A human in the ad helped, and a known name did not help or hurt. | LinkedIn co-ran the study with Magna. The page does not say how an ad was classed as creative or what the baseline was. The ads span many verticals, such as auto, retail, healthcare and tech. The outcomes are brand measures, not sales. |
| [Marketing Week column, 8 May 2026](https://www.marketingweek.com/humour-b2b-medicine-pizza/) | Humour fell from 53 percent of ads in 2000 to 34 percent in 2024. System1 data put top-quartile laugh ads at about five times the ESOV efficiency of bottom-quartile ones. | Opinion by an agency co-founder. The case in it reports more than ten times its commercial target and a client-calculated return on investment (ROI) above 1,200 percent. We did not open the System1 source. |
| [SaaS Hero, published 15 February 2026](https://www.saashero.net/strategy/human-centric-b2b-ad-frameworks/) | Names buyer tensions such as "Fear of looking wrong internally" and "Preference for peer social proof over vendor claims". | Shows no real ad and no test result of its own. |

## How do you choose one lever for a brief?

Find the step where this buyer is most likely to stall, then pick one lever for that step whose condition you can meet today. The method is our reading, not a tested rule. One lead lever and at most one support lever is our reading too. Real ads often pull two, as the Celonis-promoted post does. In our reading, a third lever blurs the point.

1. Take the angle sentence from the [angle card](/guides/b2b-ad-angles): role, risk, reason and proof.
2. Name the awareness stage the angle assumes, and set the start of the first line from the stage table.
3. Find the stall. Ask the four step questions in order and stop at the first "no". Will this reader stop for the ad? Can the reader say the point back? Will the reader trust it? Is the next step safe for the reader? The first "no" is the step to work on.
4. Pick one lever for that step from the sheet below. Read its condition and what you must hold. If you cannot meet the condition today, take the other lever for that step, or close the gap first. For example, source a number before you use loss framing.
5. Add one support lever at most, from any step.
6. Write the lever line in the brief, using the template below.
7. If you test, change only the lever. Keep angle, proof and offer the same. Read the result on the goal the ad was bought for, and see the [angles guide](/guides/b2b-ad-angles) for how to test on a small budget.

The sheet below is the core of this page. A pipe character separates the columns. Paste it into a spreadsheet and split the text at the pipe to get a table, or paste it into your strategy doc as it is. The last two columns are for your team to fill in.

```template
Lever | Step | Public root | What the root found or claims | Condition to meet | What you must hold | Evidence | Our line for [product] | Review verdict
An information gap | Noticed | Loewenstein, 1994 | Curiosity is a felt gap in knowledge, and it pulls hardest when the reader knows most of the answer | The reader knows the field, and the answer is worth the click | The answer, written before the question | Measured elsewhere (theory and lab tasks). The memory claim failed in a 2025 meta-analysis. | [ ] | [ ]
A small admission | Noticed | Ein-Gar and colleagues, 2012 | A small negative detail after a positive one raised appeal when people processed it with little effort | The admission is small, true and comes after the promise | One real limit of the product | Measured elsewhere (product descriptions, not ads) | [ ] | [ ]
Framing | Understood | Kahneman and Tversky, 1979 | People judge outcomes as gains or losses from a reference point | The reference point is one the buyer uses today | A plain account of how the buyer works today, or a picture that maps part by part to the problem | Measured elsewhere (reference point). Practitioner claim (pictures). | [ ] | [ ]
A common opponent | Understood | Bosson and colleagues, 2006 | A shared dislike of a third party raised closeness, in the experiment only for mild dislikes | The opponent is a chore or habit the reader does not own | A true account of how the chore happens | Measured elsewhere (about people), plus a practitioner recipe | [ ] | [ ]
Peers like the buyer | Believed | Cialdini; Goldstein, Cialdini and Griskevicius, 2008 | A norm about guests in the same room beat general norms | The peer matches the reader in size, field or role | A peer quote or result with source, date and permission | Measured elsewhere (hotel field experiments) | [ ] | [ ]
Specific facts | Believed | Hopkins, 1923; Cialdini on authority | Specific facts carry their full weight, and generalities roll off | The figure says what it counts and where it comes from | One sourced number, or data only you hold, with its sample | Practitioner claim | [ ] | [ ]
Loss framing | Acted on | Kahneman and Tversky, 1979; Brown and colleagues, 2024 | Losses weigh more than equal gains, a mean of about 1.96 in a 2024 meta-analysis, and debated | The buyer can check the loss, and the next step is in the same ad | A loss figure from the buyer's own kind of records | Measured elsewhere, debated | [ ] | [ ]
A smaller first step | Acted on | Bain, 2018; Hopkins, 1923 | Reduced anxiety plays a large role in some purchases | The step is small, cheap to undo and true on the first day | A trial, pilot or data check the product can deliver now | Practitioner claim, plus survey context | [ ] | [ ]

Lever sheet. Source: nimblet.io/guides/psychology-of-b2b-ads
Evidence labels: measured elsewhere = a published study that did not test B2B ads on LinkedIn or Meta. Practitioner or vendor claim = a recipe or number from someone who sells ads, tools or training. Our reading = judgement, not tested.
Step = where the lever helps the buyer: noticed, understood, believed or acted on. The steps, and which lever sits at which step, are our reading.
Stages for the first line = the five awareness stages from Eugene Schwartz: unaware, problem-aware, solution-aware, product-aware, most-aware.
No lever is shown to lift B2B ad results on LinkedIn or Meta.
```

Write the lever into the brief with this line. The [brief template](/guides/b2b-creative-brief-template) has a place for it under the angle.

```template
LEVER LINE (add under the angle in the brief)
Lead lever: [a lever from the sheet]
Support lever: [a lever, or none]
Step where the buyer stalls: [noticed / understood / believed / acted on]
Why this lever for this buyer: [one sentence that names the role's risk]
Awareness stage: [unaware / problem-aware / solution-aware / product-aware / most-aware]
Condition from the sheet: [the condition, in this ad's words]
Do we meet it today: [yes, with what / no, and what is missing]
Evidence label: [measured elsewhere / practitioner or vendor claim / our reading]
Proof the lever needs: [source, date and permission, or "none yet"]
Test design: [change only the lever, or "no test"]
Goal metric we read after launch: [the metric this ad is bought for]
```

Here is the line filled in for Fernmark. All of it is constructed. The role is an IT asset manager who fears being blamed for a renewal nobody chose, and the stage is problem-aware. The manager knows renewals hurt, so the ad gets noticed. The stall is at understood: colleagues blame the manager, and nobody has words for the real cause. The team has no proof yet, which rules out the two levers for belief. Loss framing needs a number the buyer can check, and the team has none. A common opponent needs only a true account of how renewal notices travel.

```text
EXAMPLE ONLY. Fernmark is invented, and every line below is constructed.

LEVER LINE
Lead lever: a common opponent
Support lever: framing
Step where the buyer stalls: understood
Why this lever for this buyer: the IT asset manager is blamed for a renewal nobody chose, and the opponent is the notice that reaches the wrong inbox, so the blame moves off the person
Awareness stage: problem-aware
Condition from the sheet: the opponent is a chore the reader does not own. The reader owns the licence list and not the notices, so the ad blames how notices are routed
Do we meet it today: yes, with a screen that shows how the product routes notices
Evidence label: measured elsewhere (about people, not processes), plus a practitioner recipe. Our reading for B2B.
Proof the lever needs: none beyond that screen
Test design: same angle, proof and offer, with a loss framing version once the team has a number it can source
Goal metric we read after launch: [the metric this ad is bought for]
```

The next prompt reads the levers in one ad. Use it on your own ads, on a competitor's ads from the ad library, or on a draft. Check every quote it returns against the ad yourself. Models over-read, and rule 1 is there to hold them to the words.

```prompt
You are a senior creative strategist who reads business-to-business (B2B) ads for LinkedIn and Meta. Read the ad below and name the persuasion levers it uses. Describe what the ad does. You have no spend and no results, so never say or imply that the ad worked or did not work.

STEPS
An ad must be noticed, understood, believed and acted on. Each lever below helps at one step.

LEVERS (use these names; if the ad uses another lever, name it and say which step it serves)
- an information gap (noticed): a question about the reader's work stays open until the click. Condition: the reader knows enough to feel the gap.
- a small admission (noticed): one true, small limit placed after the main promise, or a joke that carries the argument.
- framing (understood): the ad sets what the buyer compares against, such as today's tool, a new cause for a pain, or a familiar picture.
- a common opponent (understood): a chore, habit or routine the reader dislikes stands against the ad and the reader. Condition: the reader does not own it.
- peers like the buyer (believed): a customer or group like the reader carries the claim. Condition: the peer matches the reader in size, field or role.
- specific facts (believed): one exact figure, a fact only an expert could know, or a cause and effect sentence. Condition: the ad says what the figure counts.
- loss framing (acted on): the ad names what the current way costs. Condition: the next step is in the same ad.
- a smaller first step (acted on): the step the ad asks for is small, cheap and easy to undo.

THE AD
Platform: [LinkedIn or Meta]
Format: [single image, video, carousel, document, message, thought leader post, other]
Advertiser: [name]
Text, word for word, with headline and button: [paste]
What the picture or video shows: [describe it, or write "not available"]

RULES
1. Quote the words that carry each lever, in the original language, with an English translation in brackets. If no words or visible device carry a lever, the lever is not present. Do not guess a lever from the product category.
2. Name one lead lever and at most one support lever.
3. Do not invent facts about the advertiser, the audience or results. A number in the ad is the advertiser's own claim unless the ad names a source. Say so.
4. Name the risk the ad speaks to, as something that happens to a person. If the ad names none, write "not clear from the ad".
5. Name the awareness stage the ad assumes: unaware, problem-aware, solution-aware, product-aware or most-aware. Quote the words that show it.
6. Write short, plain sentences. Use no dashes and no praise words.

ANSWER IN THIS FORMAT
Lead lever: [name]
Words that carry it: "[quote]"
Support lever: [name, or none]
Words that carry it: "[quote, or none]"
Step it works on: [noticed / understood / believed / acted on]
Risk the ad speaks to: [one sentence]
Awareness stage it assumes: [stage], because "[quote]"
Condition of the lead lever, and whether the ad meets it: [two sentences at most]
Not known from the ad: [one sentence]
```

The second prompt helps before the ad exists. It finds the step where the buyer stalls, ranks three levers for one angle and flags what each one needs. The [writing guide](/guides/write-b2b-ads-with-claude-or-chatgpt) covers the full draft workflow.

```prompt
You help a business-to-business (B2B) creative team choose a persuasion lever for one angle on LinkedIn or Meta. Do not write the finished ad.

THE ANGLE
Product, in one sentence: [what it does]
Buying role: [one job, not "decision makers"]
Risk that falls on that role if the choice goes wrong: [one sentence]
Reason to say yes: [one sentence]
Proof we can source, with source and date: [list, or "none yet"]
Awareness stage to reach: [unaware / problem-aware / solution-aware / product-aware / most-aware]
Step where we think the buyer stalls: [noticed / understood / believed / acted on / not sure]
Platform: [LinkedIn or Meta]
Words the buyer uses, from reviews, calls or tickets: [3 to 6 phrases, or "none"]

THE LEVERS (step, then what the team must hold)
- an information gap (noticed): an answer worth the click, for a reader who knows the field.
- a small admission (noticed): one real, small limit of the product.
- framing (understood): a plain account of how the buyer works today, or a picture that maps part by part to the problem.
- a common opponent (understood): a true account of a chore or habit the reader dislikes and does not own.
- peers like the buyer (believed): a peer of the reader's size, field or role, with source, date and permission.
- specific facts (believed): one number with its source and what it counts, or data only we hold, with its sample.
- loss framing (acted on): a loss figure the buyer can check, and a next step in the same ad.
- a smaller first step (acted on): a trial, pilot or data check the product can deliver now.

TASK
1. Name the step where this buyer most likely stalls. Ask in order: will the reader stop, can the reader say the point back, will the reader trust it, is the next step safe? Stop at the first no. If my input names a step, test it and say if you disagree.
2. Pick three levers. Take the first from the stall step. Prefer levers whose needs my input already meets. Rank them.
3. For each lever give:
   a. Why it fits this role and this risk, in one sentence.
   b. One example first line of at most 20 words, in the buyer's words where I gave them. Where a line needs a number or name I did not give, write [number] or [name]. Never invent one.
   c. What the lever needs, from the list. If my input does not give it, write "needs" and the item, for example "needs a checkable loss figure".
4. Say which lever to test first, and why, in two sentences.
5. List what you could not decide because my input was missing.

RULES
- Do not claim that any lever will raise results. No lever is shown to lift B2B ad results.
- Do not invent customers, quotes, numbers or sources.
- Use plain words and short sentences. Use no dashes.
```

## Nimblet runs this check

Nimblet's intelligence report groups your LinkedIn and Meta ads into patterns, judges them on the goal each ad was bought for and says what each pattern tells you about your buyer. Each brief it writes names the psychology principle behind an angle when one applies. Nimblet only reads your ad accounts and never changes a campaign.

## Frequently asked questions

### What is a lever in a B2B ad?

A lever is the mechanism an ad uses to move a buyer, such as a peer, an exact number or a loss. It differs from the angle (the reason to say yes), the hook (the first line) and the proof (the evidence). This page covers eight levers, two for each step the buyer takes: noticed, understood, believed and acted on. Other levers exist, such as Cialdini's reciprocity and consistency. One ad has one lead lever and at most one support lever.

### How many levers should one ad use?

One lead lever and at most one support lever. This is our reading. No test backs it. Each extra lever takes space from the point. Real ads often pull two, such as the Celonis-promoted post above, which pairs a familiar picture with the fear of lock-in.

### Is a loss twice as strong as a gain?

Not as a fixed figure. A 2024 meta-analysis of 607 estimates puts the mean at 1.955, with a 95 percent interval from 1.820 to 2.102 ([Brown and colleagues](https://www.aeaweb.org/doi/10.1257/jel.20221698)). Gal and Rucker argue that the evidence does not show losses weigh more than gains in general ([abstract](https://doi.org/10.1002/jcpy.1047)). No study we found tests it in B2B ads.

### Do open loops work because of the Zeigarnik effect?

The memory claim did not hold. A 2025 meta-analysis found no memory advantage for unfinished tasks and a general tendency to resume them ([Ghibellini and Meier](https://www.nature.com/articles/s41599-025-05000-w)). An open loop in an ad leans on the urge to finish, which is the click or the swipe. Even that comes from lab tasks, not ads.

### Are emotional B2B ads seven times more effective?

We do not repeat that figure. A LinkedIn blog page of 10 November 2019 prints "7x more very large business effects" for emotional campaigns in the long term, and names no sample, period or method ([LinkedIn](https://www.linkedin.com/business/marketing/blog/content-marketing/why-b2b-marketing-needs-emotion)). The same page says rational campaigns gave roughly twice the short-term activation effects. We also found the often-quoted 90 percent buyer-memory figure only second hand, in a Marketing Week column that credits Harvard Business Review.

### Should B2B ads use fear?

Fear can work, if the ad gives a way out. A meta-analysis of fear appeals found a positive average effect (d = 0.29) that grew when the message said what to do about the threat ([abstract](https://pubmed.ncbi.nlm.nih.gov/26501228/)). It did not test B2B ads. Our reading: a loss with a number the buyer can check and a next step beats vague doom.

### Do status plays work on Meta?

We found no test. Our reading is that they land worse. On LinkedIn the reader is a named professional who cares how a reaction looks at work. On Meta the reader is a stranger in leisure time, and a line about status reads as pressure. See [LinkedIn ads vs Facebook ads for B2B](/guides/linkedin-ads-vs-facebook-ads-b2b).

### How do you read a competitor's levers?

Pull 10 to 20 of their ads from the library, as the [Ad Library guide](/guides/linkedin-ad-library) shows. Code each with the first prompt, one row per ad: lead lever, step and stage. Count the rows. A step nobody works on may be open space, but only if you can meet the lever's condition. The library shows what a competitor says and never what works.

### Can Claude or ChatGPT code the levers?

Yes, with the first prompt. Rule 1 asks for the words that carry each lever, so you can check them. Models over-read, so compare every quote with the ad. For the full workflow of briefs, drafts and review, see [writing B2B ads with Claude or ChatGPT](/guides/write-b2b-ads-with-claude-or-chatgpt).

### Does Nimblet find the levers in my account?

Nimblet records each ad's hook, offer and proof from the creative itself. Its intelligence report groups your ads into patterns and says what each pattern tells you about your buyer. Nimblet does not tag a lever on every live ad. The reading card and the prompt on this page do that step by hand or in Claude or ChatGPT. When Nimblet writes a brief, it names the mechanism behind each angle ([AI ad generator](/features/ai-ad-generator)). Nimblet only reads your ad accounts and never changes a campaign. See [creative intelligence](/features/creative-intelligence).