A Research Review of ROI, Consumer Response, and Environmental Impact

Generative artificial intelligence has dramatically lowered the cost and time required to produce marketing graphics. A business owner can now describe an event, promotion, product, or service and generate a complete visual concept in seconds.

For small businesses, this raises an obvious question:

Are AI-generated flyers actually more effective than traditionally designed flyers?

The answer is complicated.

The emerging research suggests that AI-generated advertising can perform extremely well in certain circumstances. AI imagery can match or outperform human-created imagery on measures such as visual quality and click-through rate, while dramatically reducing the cost and time associated with producing creative assets. At the same time, consumers can respond negatively when they know an advertisement was generated by AI, particularly when the advertisement depends on authenticity, emotional connection, or trust.

There is also an environmental question. AI does not make marketing “free” from an environmental perspective. Generating images requires computational resources, while printed flyers require paper, ink, manufacturing, transportation, and disposal.

The most useful conclusion is therefore not that AI-generated flyers are either good or bad.

The better question is:

When does AI-generated creative produce enough additional value to justify its financial, environmental, and brand costs?

1. The Economics of AI-Generated Advertising

One of the strongest arguments for AI-generated creative is cost.

Traditional marketing graphics can require a designer, photographer, illustrator, copywriter, production specialist, and multiple rounds of revisions. AI can reduce some of those production requirements dramatically.

Research published in the International Journal of Research in Marketing provides some of the strongest evidence so far.

Researchers generated 10,320 synthetic marketing images using seven different text-to-image models and compared them with 2,400 real-world human-created images. More than 254,000 human evaluations were collected.

The researchers found that AI-generated marketing imagery could outperform human-made imagery in areas including quality, realism, and aesthetics. In a field study involving more than 173,000 advertising impressions, some AI-generated banner ads achieved click-through rates as much as 50% higher than comparable human-made imagery.

That does not mean that an AI-generated flyer will automatically generate 50% more customers.

The experiment involved digital banner advertising rather than printed flyers, and click-through rate is an upper-funnel metric rather than a completed sale.

Nevertheless, the research demonstrates something important:

AI-generated imagery is capable of producing commercially competitive advertising creative.

A separate 2026 working paper from Columbia Business School provides even larger-scale evidence. Researchers analyzed more than two million ad-day observations from more than 7,000 advertisers across nearly 50 product categories, representing more than 16 billion impressions and 116 million clicks.

They found that ads using AI-generated images outperformed ads using human-generated images on click-through rate—but only when the AI-generated imagery did not visibly “look like AI.”

This distinction is critical.

The advantage wasn’t simply “AI.”

The advantage was effective creative produced with AI.

2. AI Can Improve the Economics Without Improving the Marketing

This is an important distinction for businesses.

Suppose a traditional flyer costs $300 to develop.

An AI-assisted workflow might reduce the production cost to $50.

At first glance, that appears to be an 83% cost reduction.

But if the traditional flyer generates 30 customers and the AI flyer generates only 10, the cheaper flyer isn’t necessarily the better marketing investment.

The correct measure is not:

Cost per flyer.

It is:

Cost per customer acquired.

And ultimately:

Profit generated per marketing dollar.

This is why ROI needs to be evaluated at the campaign level rather than simply comparing design-production costs.

A useful formula is:

Marketing ROI = (Incremental profit generated − campaign cost) ÷ campaign cost

For example:

A $500 campaign generates $2,000 in incremental profit.

ROI:

($2,000 − $500) ÷ $500 = 300%

If AI reduces creative production from $300 to $50, the savings are meaningful—but only if the resulting creative performs sufficiently well to maintain or improve the campaign’s actual business results.

3. Production Speed May Be One of AI’s Biggest Advantages

The biggest advantage of AI may not actually be the image itself.

It may be the ability to produce and test many alternatives.

A traditional workflow might produce one or two flyer concepts.

An AI-assisted workflow can produce dozens of variations.

That makes experimentation much cheaper.

A business could test:

Different headlines
Different images
Different calls to action
Different layouts
Different offers
Different emotional appeals
Different audiences
Different color treatments
Different product presentations

Instead of asking:

“Is this flyer good?”

the business can ask:

“Which version produces the best response?”

That is a much more useful marketing question.

Research on generative AI-powered advertising supports this broader productivity argument. A 2025 study examining more than 10,000 consumer comments on AI-powered and comparable human-created advertising found that a majority of the AI-powered ads in its sample were classified as productive, with consumers generally responding positively.

Other field research has found that generative AI can increase advertising engagement. One Marketing Science field experiment found that personalized GenAI video advertisements increased engagement by approximately six to nine percentage points compared with the study’s baseline conditions.

The implication for small businesses is significant:

AI can make experimentation economically practical for organizations that previously could not afford extensive creative testing.

4. But Attention Isn’t the Same as Conversion

This may be the most important finding for marketers.

A flyer can get attention without generating a customer.

A person can:

Notice the flyer.
Read the headline.
Scan the QR code.
Visit the website.
And still never buy anything.

Recent research illustrates this distinction.

A 2026 randomized field experiment comparing more than 150 AI-generated and human-created video advertisements found that AI-generated ads performed better on upper-funnel measures such as click-through rate and viewer retention.

However, those gains did not carry through to conversions. AI-generated creatives were less likely than human-created creatives to generate downstream conversions in that experiment.

This suggests that AI can be particularly good at capturing attention, while human-created elements may remain important for trust, credibility, and persuasion.

For a flyer, this distinction is especially relevant.

A visually spectacular AI-generated image might make someone stop and look.

But if the flyer is promoting:

A local therapist
A farm
An artist
A craft workshop
A nonprofit
A restaurant
A retreat
A professional service
A community event

then authenticity may matter as much as visual impact.

5. How Consumers Respond When They Know AI Was Used

One of the most interesting findings in current advertising research is the effect of disclosure.

Consumers don’t necessarily react negatively to AI-generated content simply because it was created by AI.

The reaction depends on context.

Research published in 2025 found that when AI involvement was disclosed, consumers could penalize AI-generated imagery compared with human-made alternatives. However, those negative effects varied depending on the reason given for using AI, including explanations involving cost savings, privacy protection, or visual quality.

More recent research has found similar effects.

A 2026 study examining AI disclosure in advertising found that labeling an advertisement as AI-generated reduced advertising attitudes in several experimental settings. Perceived eeriness helped explain the effect, although stronger brands appeared better able to buffer it.

Another 2026 study found that transparent explanation of AI’s role can improve perceptions of the creation process under some circumstances, suggesting that the way AI involvement is communicated can matter.

The research therefore doesn’t support a simple rule such as:

“Never tell customers that AI was used.”

Instead, it suggests that disclosure, context, brand strength, and the role AI played in creating the advertising all influence consumer response.

6. Authenticity May Matter More for Local Businesses

This is where the flyer question becomes particularly interesting.

Large companies can potentially use synthetic imagery without consumers expecting the imagery to represent an actual person or location.

Small local businesses are different.

Imagine a local farm promoting a Saturday market.

An AI-generated photograph might show a beautiful basket of vegetables, a perfect barn, and an idyllic farm landscape.

But if none of those things actually exist, the image creates a potential authenticity problem.

The same applies to:

Restaurants
Artists
Contractors
Farms
Retreat centers
Medical practices
Salons
Local events
Churches
Nonprofits
Community organizations

For these businesses, an actual photograph of the business, owner, product, property, staff, or event may carry information that an AI image cannot reproduce.

This suggests a useful hybrid strategy:

Use AI for visual enhancement and concept development while using authentic photography for things customers actually need to trust.

7. The Environmental Cost of AI-Generated Images

AI-generated advertising is often described as environmentally efficient because it eliminates photography, travel, physical prototypes, or other traditional production requirements.

That can sometimes be true.

But AI generation itself requires energy.

The environmental footprint varies considerably between models and generation methods.

A 2025 study that measured the energy consumption of 17 image-generation models found differences of as much as 46× between models. The study also found that image resolution could substantially affect energy consumption, with energy increases ranging from approximately 1.3× to 4.7× when resolution was doubled, depending on the model.

Earlier research published in Scientific Reports estimated that generating an image using certain AI systems could produce emissions on the order of grams of CO₂ equivalent per image, although the authors noted substantial uncertainty and differences between systems.

The important point is not the exact number.

There is no single environmental cost for “an AI image.”

The footprint depends on:

The model
Hardware
Resolution
Number of generations
Number of revisions
Data-center efficiency
Electricity source
Geographic location
Whether the image is generated once or thousands of times

This makes the common practice of generating dozens or hundreds of disposable AI images worth considering environmentally.

8. The Hidden Environmental Cost of “Try Again”

This is an underappreciated issue.

Imagine a designer creating one final image.

Now imagine someone prompting an AI system:

“Make it brighter.”

“Try again.”

“More realistic.”

“Different background.”

“Make the person younger.”

“Make it more dramatic.”

“Try another.”

“Add flowers.”

“Remove flowers.”

“Make it photorealistic.”

Ten, twenty, fifty, or hundreds of generated images may never be used.

Each generation requires computational resources.

The 2025 image-generation energy study found substantial variation in energy consumption between models, reinforcing the importance of how much generation is performed rather than simply whether AI is involved.

This creates an interesting environmental principle:

AI can reduce waste in physical production while simultaneously creating a new form of digital resource consumption.

The environmentally responsible approach is therefore not necessarily “never use AI.”

It is:

Generate intentionally. Select carefully. Avoid unnecessary iterations. Reuse useful assets.

9. The Flyer Still Has an Environmental Footprint

Even if the artwork is generated digitally, a printed flyer still requires physical materials.

The paper and printing industries have measurable environmental impacts.

The U.S. Environmental Protection Agency reports that the U.S. pulp and paper sector reported 31.7 million metric tons of direct greenhouse-gas emissions in 2023. The sector includes facilities involved in pulp and paper manufacturing as well as other paper production and related activities.

Printing itself can also generate air emissions. According to the EPA, printing operations can release volatile organic compounds and hazardous air pollutants associated particularly with solvents in inks, although the environmental profile varies significantly by printing technology and ink system.

Paper disposal is another factor.

EPA data show that paper and paperboard represented approximately 23.1% of U.S. municipal solid waste generation in 2018. About 68.2% of paper and paperboard was recycled that year, but rates varied substantially by product category; nondurable paper goods excluding newspapers had a considerably lower recovery rate.

Therefore, the environmental impact of a flyer is not simply:

AI versus human design.

It is more accurately:

AI production + design workflow + paper + ink + printing + transportation + distribution + disposal

versus whatever alternative is being considered.

10. Digital Advertising Isn’t Automatically Environmentally Free Either

It is tempting to assume that a digital flyer is environmentally superior because nothing gets printed.

But digital media also requires infrastructure.

Servers, networks, devices, data storage, and display systems consume electricity.

A recent life-cycle assessment of digital poster systems found that distribution infrastructure and display operations accounted for 89% of the assessed lifecycle impacts in its particular case study.

That study should not be interpreted as a universal carbon number for every digital advertisement.

It does, however, demonstrate an important principle:

Digital communication has an environmental footprint too.

The environmentally preferable option depends on the entire system and how the communication is used.

11. AI May Actually Reduce Environmental Impact in Some Marketing Workflows

Consider a traditional advertising shoot for a fictional product.

It could involve:

A photographer
Models
Travel
A studio
Lighting equipment
Props
Physical products
Makeup and wardrobe
Multiple locations
Shipping
Retouching
Multiple production rounds

An AI-generated image could potentially eliminate some or all of those activities.

In that situation, the computational energy used to generate the image may be substantially less than the environmental cost of producing the equivalent physical advertising asset.

But this comparison has to be made on a case-by-case basis.

The environmental question isn’t:

“Is AI green?”

It’s:

“What physical and digital production process is AI replacing?”

That is a much more useful question.

12. What About the ROI of an AI Flyer?

For businesses, this is ultimately the question that matters.

The best way to measure an AI-generated flyer is through controlled testing.

For example:

Version A

Human-designed flyer

Version B

AI-generated flyer

Version C

Human-designed flyer using AI-assisted imagery

Then measure:

Impressions
Response rate
QR scans
Website visits
Phone calls
Coupon redemptions
Event registrations
Leads
Purchases
Revenue
Gross profit
Cost per acquisition

The most important number is not the number of people who saw the flyer.

It is the number of incremental customers and dollars generated by the flyer.

13. A Simple Small-Business ROI Experiment

A local business could conduct a relatively inexpensive experiment.

Suppose a business distributes 3,000 flyers.

1,000 receive Version A.

1,000 receive Version B.

1,000 receive Version C.

Each version receives a unique QR code, URL, coupon code, or tracking number.

After the campaign, the business might discover:

Version	Cost	Responses	Customers	Revenue
Human-designed	$700	70	35	$3,500
AI-generated	$350	80	24	$2,400
AI-assisted human design	$500	85	42	$4,200

In this hypothetical example, the AI-only flyer generates more responses than the traditional flyer but fewer customers.

The AI-assisted version generates the most customers.

That would lead to a very different conclusion than simply saying:

“The AI flyer got more responses.”

This is why marketing should be measured throughout the funnel.

14. The Emerging Evidence Favors Hybrid Workflows

Taken together, the research points toward an interesting model.

AI appears particularly strong at:

Generating visual concepts
Creating variations
Increasing production speed
Reducing certain creative costs
Supporting personalization
Producing attention-grabbing imagery
Rapid experimentation

Humans remain particularly important for:

Brand strategy
Authenticity
Local knowledge
Emotional storytelling
Art direction
Quality control
Fact checking
Legal and ethical judgment
Understanding customers
Final creative decisions

This is consistent with field research examining advertising agencies in Rio de Janeiro and Catalonia, which found that agencies were using generative AI primarily as a complement to creative processes and that human verification remained necessary.

The strongest model may therefore be neither:

Human versus AI

nor:

AI replaces the designer.

Instead, it is:

Human strategy + AI-assisted production + human judgment.

15. What This Means for Small Businesses

For small businesses, AI-generated flyers can be particularly attractive because they reduce the financial barrier to experimentation.

A business that previously could afford one professionally designed campaign might now be able to test five or ten creative directions.

That can be valuable.

But businesses should avoid confusing inexpensive production with effective marketing.

A $5 flyer that produces no customers is not cheaper than a $500 flyer that produces $5,000 in profit.

Likewise, an AI image that receives thousands of impressions but damages consumer trust may not be a successful marketing asset.

The objective should always be business performance, not simply inexpensive content production.

16. A Better Model for AI-Assisted Flyer Design

For businesses considering AI-generated flyers, a strong workflow looks something like this:

1. Define the marketing objective.

What are you trying to make the customer do?

Visit?

Call?

Buy?

Attend?

Donate?

2. Identify the audience.

Who specifically needs to respond?

3. Develop the offer.

What is the actual reason someone should take action?

4. Develop the message.

The headline and offer should come before the decorative imagery.

5. Use AI strategically.

Generate concepts, imagery, variations, and visual directions where appropriate.

6. Add authentic elements.

Use real photographs, real locations, actual products, real customer experiences, and accurate information whenever authenticity matters.

7. Have a human review the result.

Check typography, facts, brand consistency, imagery, accessibility, legal considerations, and overall communication.

8. Test multiple versions.

Don’t assume the first AI-generated design is the best one.

9. Track actual business outcomes.

Measure leads, customers, revenue, and profit—not simply impressions or likes.

10. Calculate the complete ROI.

Include creative production, printing, distribution, advertising spend, labor, and other campaign costs.

Conclusion: AI Can Make Flyers Cheaper. That Doesn’t Automatically Make Them Better.

The research emerging in 2025 and 2026 suggests that generative AI has genuine potential to transform advertising production.

AI-generated imagery can compete with or outperform human-created imagery in controlled evaluations and digital advertising environments. Some field studies have reported substantially higher click-through rates for AI-generated advertising imagery.

At the same time, other research shows that stronger attention does not necessarily translate into stronger conversion. Consumer trust, perceived authenticity, brand context, and awareness of AI involvement can significantly affect outcomes.

The environmental picture is similarly complicated.

AI image generation consumes electricity and its energy requirements vary enormously between models and workflows.

But traditional printed advertising also has an environmental footprint involving paper production, printing, transportation, and disposal.

Consequently, there is no universal answer to the question:

“Are AI-generated flyers better?”

The evidence points toward a more useful conclusion:

AI-generated flyers can be cheaper, faster, and highly effective—but their success depends on how they are created, what they communicate, how consumers perceive them, and whether they ultimately produce measurable business results.

For many businesses, the most promising approach may not be replacing the designer with AI.

It may be using AI to make the designer—and the marketing process—more productive.

The winning formula is likely to be:

Human strategy + AI-assisted creativity + authentic content + measurable testing + responsible production.

That combination gives businesses something much more valuable than a cheap flyer.

It gives them a marketing system that can learn what actually works.

Research Notes

The evidence base for AI-generated flyers specifically is still developing. Most current studies examine AI-generated advertising imagery, social-media advertisements, banner ads, video advertisements, or digital marketing rather than printed flyers. Therefore, conclusions about printed flyers should be treated as an application of broader advertising research rather than direct experimental proof.

The strongest directly relevant evidence currently includes large-scale field research on AI-generated advertising imagery, controlled experiments on consumer reactions to AI disclosure, research on AI advertising productivity, and environmental studies measuring the energy requirements of image-generation systems.

This distinction matters because a digital banner, social-media advertisement, and printed local flyer operate under different conditions. They have different production costs, distribution mechanisms, exposure times, measurement capabilities, and environmental footprints.

AI-Generated Flyers: Do They Actually Work?