Overview

nCino, Inc. is a leader in intelligent banking solutions and a pioneer in the financial technology space. As their business continues to grow and their digital strategy evolves, they needed to ensure their content management capabilities could keep pace with their expanding marketing needs.

Our collaboration with nCino focuses on optimizing their use of the Contentful platform to enhance their internal marketing and website capabilities. By leveraging Contentful, we’re helping nCino streamline and improve their content management processes, allowing their teams to focus more on innovation and growth within their core business.

The approach

Why this project matters

Overview

Aerospike specializes in high-performance NoSQL databases known for their speed, scalability, and reliability in handling large volumes of data in real-time applications. Their database technology is designed to meet the demands of modern data-intensive applications.

The company’s platform offers features such as strong consistency, high availability, and automatic failover to ensure continuous operations even in the event of hardware failures or network issues. Aerospike also provides tools and integrations to support analytics, monitoring, and management of the database environment, empowering developers and operations teams to optimize performance and scalability.

Think of a NoSQL database as a big, organized digital filing cabinet for storing different kinds of information. Traditional databases are like organized spreadsheets where everything is neatly arranged in rows and columns. However, NoSQL databases are designed to handle different types of data—like text, numbers, pictures, and more—and can store huge amounts of it very quickly. They’re a super-fast and flexible digital storage system.

The challenge

The approach

Headless CMS functions similarly to NoSQL databases, allowing content managers to view and manage content of all forms easily and from one place. Under the guidance of our team, Aerospike decided that a migration to a headless CMS with Contentful was the right option.

This decision, along with brand redesign and consistent messaging, were all instrumental parts of Aerospike’s digital revolution. We were able to work side by side with Aerospike to give them a refresh that tells a story.

The Results

Overview

For over twenty years, RepTrak has been the go-to provider for reputation data and insights, helping organizations understand and improve their corporate reputation. With their flagship Global RepTrak 100 report, RepTrak offers an annual definitive ranking of corporate reputation for the world’s leading companies, providing valuable benchmarks that influence strategic decisions and stakeholder relationships.

The RepTrak Platform draws on the world’s largest reputation database with over 20 years of data. Their reputation scores serve as a leading indicator, allowing teams to interpret constantly updating streams of reputation, brand, ESG, and media data.

RepTrak’s Home and Global RepTrak 100 Landing Page are their most important lead generators, making it imperative to get these digital experiences right.

Key results

The Challenge

The Global RepTrak 100 report is more than just data — it’s a definitive ranking system recognized industry-wide that reinforces RepTrak’s leadership in the reputation industry. Their homepage demands similar attention as the first or second touchpoint for leads.

The challenge was to design landing pages that not only met the aesthetic and functional needs of their users but also reinforced RepTrak’s brand as a trusted and authoritative source. With the report being their top lead generator for the year, the landing page needed to be engaging, fast-loading, and seamlessly integrated into their Contentful site.

Beyond aesthetics for outside visitors, their internal team required Contentful modeling conducive to empowering Content Managers, guidance on technical integrations, and a new design system.

The Approach

Redefining Technical Support

With any project, proper guidance is an often overlooked prerequisite. It’s fairly common to “know what you want” and have no idea how to get there. It’s even more common to “know what you want” and for that journey to achieve the “want” be ill-advised. Without the outside perspective of a technical solutions partner, internal biases and inefficiencies multiply.

We approached this project interdisciplinary and agile. Assuming the role of impartial confidant, we were able to give the RepTrak team objective recommendations, allowing us to focus on speed with a collaborative touch.

Collaborative and Strategic Design

The Home and Global RepTrak 100 landing page received a complete overhaul, designed to elevate user experience, increase engagement, and drive conversions. Not to mention, make content editing and management easier for all parties internally.

The redesigned landing page is a testament to our collaborative efforts with RepTrak, merging aesthetics with functionality. By focusing on user experience and leveraging Contentful’s robust capabilities, we created a page that not only highlights the significance of the Global RepTrak 100 report but also aligns with RepTrak’s brand values and business goals.

The design features intuitive navigation, clear calls to action, and visually appealing elements that draw attention to key insights from the report. We also incorporated responsive design principles to ensure the page performs well across devices, catering to a global audience.

The Results

The redesign delivered measurable impact on RepTrak’s most important lead generation channels. Report downloads increased by 6% and conversions saw an impressive 40% year-over-year boost.

The new landing page is not just a one-time update — it’s a strategic investment in RepTrak’s digital presence. By ensuring a seamless and engaging experience, we’ve laid the groundwork for future enhancements that will extend to other areas of their Contentful site.

[Oomph] truly understood how important this report was to the company and helped us build something that can be translated across our website — so every piece we release can be just as powerful.

— Bianca Martucci-FiNk, Director of Global Content Marketing, The RepTrak Company

Overview

Stats/Key Outcomes

The Challenge

The Solution

We partnered with Contentful to deliver a unified content architecture, beginning with a full migration of MyCase.com from WordPress to Contentful.

Key elements of the MyCase solution included a full migration of MyCase.com encompassing 166+ pages, 400+ blog posts, and 4,700+ media files. We built a structured content model for blogs, guides, webinars, press releases, videos, and landing pages. Design system standardization streamlined and standardized design components across the site. We ensured SEO continuity and GEO visibility by preserving 143+ SEO-critical redirects and improving metadata governance. For MarOps enablement, we integrated Marketo forms directly into Contentful’s structured model. Tech enablement connected Frontify DAM to manage brand assets consistently across all brands.

With the help of an Atomic Design System, this solution became the blueprint for ongoing migrations and updates across the broader 8am portfolio.

Shared DAM: One asset library powers all 8am brands — no duplication, no outdated files.

The Results

Overview

Ibotta is a leading performance marketing platform that helps brands reach over 200 million consumers through results-driven digital promotions. Backed by Walmart, Ibotta’s Performance Network (IPN) empowers marketers to influence purchasing decisions, shopping locations, and shopping frequency.

Millions of consumers use Ibotta to get cash back every time they shop, earning an average of $261 each year.

We partnered with Ibotta through a collaboration focused on Contentful. This partnership brought together Ibotta’s innovative consumer engagement model and our digital strategy expertise. Our shared goal: build smarter, results-driven experiences that move people and move product.

The approach

The Results

Why this project matters

We’re proud to support partners like Ibotta who are reshaping the future of performance marketing.

This collaboration reflects what we do best: delivering flexible, expert support that helps teams move faster, scale smarter, and stay focused on what drives results. We look forward to continuing to build alongside innovative brands making an impact.

With Ibotta, you can get cash back every time you shop, on the app or in-store. Join the millions of Savers who earn $261 each year on average.

The Results

Key Outcomes

The Challenge

The Approach

The Result

Workhuman launched a headless content system that unified previously fragmented operations. Their editorial and development teams now work from a single source of truth, publishing consistent brand experiences across digital channels while tracking the performance indicators that drive continuous improvement.

The system doesn’t just manage content—it enables the operational clarity and measurement infrastructure Workhuman needs to optimize performance over time.

Why This Matters

THE BRIEF

THE CHALLENGE

MULTIPLE COUNTRIES AND A GLOBAL OPTION

SIX LANGUAGES

MANY 3RD-PARTY INTEGRATIONS

THE APPROACH

LEARN VS. SHOP

Supporting Visitor Intention with Two Different Modes

While the technical challenges were being worked out, the user experience and design had to reflect a cutting-edge company. With the launch of their revolutionary product, the BLK 360, in 2018, Leica positioned itself as the Apple of the geospatial measurement community — sleek, cool, cutting-edge and easy to use. While many companies want to look as good as Apple, few of them actually have the content and product to back it up.

The navigation for the site went through many rounds of feedback and testing before deciding on something radically simple — Learn or Shop. A customer on the website is either in an exploratory state of mind — browsing, comparing, reviewing pricing and specifications — or they are ready to buy. We made it very clear which part of the website was for which.

This allowed us to talk directly to the customer in two very different ways. On the Learn side, the pages educate and convince. They give the customer information about the product, reviews, articles, sample data files, and the like. The content is big, sleek, and leverages video and other embedded content, like VR, to educate.

On the Shop side the pages are unapologetically transactional. Give the visitor the right information to support a purchase, clearly deliver specs and options like software and warranties, without any marketing. We could assume the customer was here to purchase, not to be convinced, so the page content could concentrate on order completion. The entire checkout process was simplified as much as possible to reduce friction. Buying habits and patterns of their user base over the past few site iterations were studied to inform our choices about where to simplify and where to offer options.

THE RESULTS

Generative Engine Optimization (GEO) is making organizations scramble — our clients have been asking “Are we ready for the new ways LLMs crawl, index, and return content to users? Does our site support evolving GEO best practices? What can we do to boost results and citations?”  

Large language models (LLMs) and the services that power AI summaries don’t “think” like humans but they do perform similar actions. They seek content, split it into memorable chunks, and rank the chunks for trust and accuracy. If pages use semantic HTML, include facts and cite sources, and include structured metadata, AI crawlers and retrieval systems will find, store, and reproduce content accurately. That improves your chance of being cited correctly in AI overviews.

While GEO has disrupted the way people use search engines, the fundamentals of SEO and digital accessibility continue to be strong indicators of content performance in LLM search results. Making content understandable, usable, and memorable for humans also has benefits for LLMs and GEO.

How LLM systems (and AI-driven overviews) get their facts

Understanding how LLMs crawl, process, and retrieve web content helps us understand why semantic structure and accessibility best practices have a positive effect. When an AI system generates an answer that cites the web, several distinct back-end steps usually happen: 

1. Crawling — Bots visit URLs and download page content. Some crawlers execute javascript like a browser (Googlebot) while others prefer raw HTML and limit their rendering.
2. Chunking — Large documents are split into small, logical “chunks” of paragraphs, sections, or other units. These chunks are the pieces that are later retrieved for an answer. How a page’s content is structured with headings, paragraphs, and lists determines the likely chunk boundaries for storage.
3. Vectorization — Each chunk is then converted into a numeric vector that captures its semantic meaning. These embeddings live in a vector database and enable systems to find chunks quickly. The quality of the vector depends on the clarity of the chunk’s text.
4. Indexing — Systems will store additional metadata (URL, title, headings, metadata) to filter and rank results. Structured data like schema metadata is especially valuable. 
5. Retrieval — A user asks a question or performs a search and the system retrieves the most semantically similar chunks via a vector search. It re-ranks those chunks using metadata and other signals and then composes its answer while citing sources (sometimes). 

The Case for Human-Accessible Content

There are many more reasons why digital accessibility is simply the right thing to do. It turns out that in addition to boosting SEO, accessibility best practices help LLMs crawl, chunk, store, and retrieve content more accurately.

During retrieval, small errors like missing text, ambiguous links, or poor heading order can fail to expose the best chunks. Let’s dive into how this can happen and what common accessibility pitfalls contribute to the confusion.

For Content Teams — Authors, Writers, Editors

Lack of descriptive “alt” text

While some LLMs can employ machine-vision techniques to “see” images as a human would, descriptive alt text verifies what they are seeing and the context in which the image is relevant. The same best practices for describing images for people will help LLMs accurately understand the content. 

Out-of-order heading structures

Similar to semantic HTML, headings provide a clear outline of a page. Machines (and screen readers!) use heading structure to understand hierarchy and context. When a heading level skips from an <h2> to an <h4>, an LLM may fail to determine the proper relationship between content chunks. During retrieval, the model’s understanding is dictated by the flawed structure, not the content’s intrinsic importance. (Source: research thesis PDF, “Investigating Large Language Models ability to evaluate heading-related accessibility barriers”) 

Descriptive and unique links

All of the accessibility barriers surrounding poor link practices affect how LLMs evaluate their importance. Link text is a short textual signal that is vectorized to make proper retrieval possible. Vague link text like “Click here” or “Learn More” does not provide valuable signals. In fact, the same “Learn More” text multiple times on a page can dilute the signals for the URLs they point to.

Using the same link text for more than one destination URLs creates a knowledge conflict. Like people, an LLM is subject to “anchoring bias,” which means it is likely to overweight the first link it processes and underweight or ignore the second, since they both have the same text signal. 

Example of the duplicate link problem: <a href=“[URL-A]”>Duplicate Link Text</a>, and then later in the same article, <a href=“[URL-B]”>Duplicate Link Text</a>. Conversely, when the same URL is used more than once on a page, the same link text should be repeated exactly.

Logical order and readable content

Simple, direct sentences (one fact per sentence) produce cleaner embeddings for LLM retrieval. Human accessibility best practices of plain language and clear structure are the same practices that improve chunking and indexing for LLMs

For Technical Teams — IT, Developers, Engineers

Poorly structured semantic HTML

Semantic elements (<article>, <nav>, <main>, <h1>, etc.) add context and suggest relative ranking weight. They make content boundaries explicit, which helps retrieval systems isolate your content from less important elements like ad slots or lists of related articles. 

Lack of schema

This is technical and under the hood of your human-readable content. Machines love additional context and structured schema data is how facts are declared in code — product names, prices, event dates, authors, etc. Search engines have used schema for rich results and LLMs are no different. Right now, server-rendered schema data will guarantee the widest visibility, as not all crawlers execute client-side Javascript completely. 

How to make accessibility even more actionable

The work of digital accessibility is often pushed to the bottom of the priority list. But once again, there are additional ways to frame this work as high value. While this work is beneficial for SEO, our recent research uncovers that it continues to be impactful in the new and evolving world of GEO.

If you need to frame an argument to those that control the investments of time and money, some talking points are: 

• Accurate brand representation — Poor accessibility hides facts from LLMs. When customers ask an AI  assistant for “best X for Y,” your content may not be shown — or worse, misrepresented. Fixing accessibility reduces brand risk and increases content authority.
• Engagement boost — Improvements that increase accurate citations and AI visibility can increase referral traffic, feature mentions, and lead quality. In a landscape where AI Answers are reducing click-through rates, keeping the traffic you have on your site for longer and building brand trust becomes vital.
• Increased exposure — Digital inclusion makes your content widely accessible to machines and the machines that assist humans. Think about a search engine as another human-assistive device, just like a keyboard or screen reader. 
• Multi-pronged benefits — Accessibility improvement improves traditional SEO, can benefit mobile performance, and reduces the risks associated with accessibility compliance policies. 

Staying steady in the storm

Let’s be clear — this summer was a “generative AI search freak out.” Content teams have scrambled to get smart about LLM-powered search quickly while search providers rolled out new tools and updates weekly. It’s been a tough ride in a rough sea of constant change.

If you continue to have questions within this rapidly evolving landscape, talk to us about your questions around SEO, GEO, content strategy, and accessibility conformance. Ask about our training and documentation available for content teams.

Additional Reading

• AHREFs.com: Is SEO Dead? Real Data vs. Internet Hysteria
• SearchEngineJournal.com: How LLMs Interpret Content: How To Structure Information For AI Search
• InclusionHub.com: SEO and Web Accessibility: What You Need to Know (from 2020, but still relevant)

As a digital services firm partnering with destination marketing organizations (DMOs) across the U.S., we’re helping teams navigate what’s already proving to be a volatile 2025—especially on the inbound side. Analysis from the World Travel & Tourism Council (WTTC) projects a stark reality: the U.S. economy will miss out on $12.5 billion in international visitor spending this year, with inbound spend expected to dip to just under $169B, down from $181B in 2024. Even more concerning, the U.S. is the only country among 184 economies in WTTC’s study forecast to see an inbound-spend decline this year.

While external market forces remain largely beyond control, we’ve identified three strategic areas where DMOs can focus their digital platforms to weather this storm and continue demonstrating measurable demand to their partners.

1. Transform Content Into Action-Driving Experiences

Why this strategic shift matters now

With inbound spend shrinking by $12.5B and key feeder markets weakening, undecided travelers need clarity and confidence to choose your destination. Content that reduces uncertainty and highlights immediate value converts better than generic inspiration.

Strategic implementation approach

Activate “Go Now” signals. Combine always-on inspiration with time-sensitive reasons to visit—shoulder-season value, midweek deals, cooling weather breaks—strategically mapped to the soft periods your analytics reveal. 

Elevate discovery through intelligent architecture. Curate SEO-optimized content hubs organized by Themes (outdoors, arts, culinary) and Moments (fall colors, winter lights). Implement structured data (FAQ, Event, Attraction) with strategic internal linking architecture so travelers find relevant options fast.

Deploy micro-itineraries for immediate conversion. Design 24–48-hour “micro-itins” featuring embedded maps, transit and parking guidance, and seamless handoffs to bookable partners. Partnering with platforms like MindTrip reduces content team effort while accelerating output—a strategy that’s proven particularly effective for our DMO clients facing resource constraints.

Authority-driven event content optimization. Event pages generate the highest intent traffic. Enhance them with rich media, last-minute planning resources, and strategic “if sold-out, try this” alternatives.

Transparent value communication. Feature free experiences prominently, implement intuitive budget filters, and deploy “Best Time to Visit” calendars comparing crowds and pricing by week and month. Transparency builds trust, and trust drives conversion.

2. Build Your Competitive Moat Through Data-Driven Audience Cultivation

Your first-party data represents your most defensible competitive advantage. As platform targeting becomes increasingly constrained and inbound spending softens, DMOs that build and activate their own audience will capture attention far more efficiently than those relying solely on paid channels.

Strategic audience development

Implement high-intent capture everywhere. Deploy contextual email and SMS prompts across high-intent templates—events, itineraries, trip planners, partner directories. Offer valuable micro-perks like exclusive maps and early event alerts. 

Master progressive profiling. Collect visitor preferences—season, interests, party type, origin market—over multiple touchpoints rather than overwhelming users with lengthy initial forms. 

Create actionable audience segments. Develop cohorts around 2025’s market realities: last-minute planners, shoulder-season seekers, road-trippers, value hunters, family weekenders, and meetings planners. 

Future-proof attribution systems. Combine GA4 with server-side tagging and standardized UTM schemas for every partner handoff. Track outbound clicks, partner session quality, itinerary saves and usage, offer redemptions, and newsletter-driven sessions. This comprehensive approach ensures you maintain visibility into conversion paths as third-party cookies disappear.

Deploy trend-driven editorial strategy. Develop weekly dashboards blending organic query trends, on-site search terms, partner click-through rates, and feeder-market signals. When interest dips in one market, pivot homepage modules and paid social toward value and itinerary content targeting more resilient markets.

3. Transform Partner Relationships Through Measurable Value Delivery

In a softening inbound environment where domestic spending carries approximately 90% of the economic load, your partners need two critical elements: qualified attention and proof of conversion. Your website should function as the region’s premier meta-directory and conversion engine.

Experience optimization strategies

Enable one-click handoffs with context preservation. Pass user filters—dates, neighborhoods, price ranges—directly into partner sites and booking engines while preserving state if travelers return. 

Deploy persistent trip planning tools. Allow users to save places and generate shareable itineraries with intelligent handoffs: “Book these two hotels,” “Reserve rentals,” “Get festival passes.” 

Create compelling partner storefronts. Develop rich partner profiles featuring availability widgets, authentic reviews, social proof, and clear calls-to-action. 

Implement strategic co-op modules. Design paid placements that provide value rather than feeling like advertisements: “Local Favorites” carousels, sponsor highlights, seasonal deal tiles—rotated by audience cohort and season. This generates additional revenue while maintaining user experience quality.

Establish closed-loop reporting systems. Standardize UTM tracking, monitor outbound events, and where permitted, implement partner pixels and offer codes to report assisted conversions by category and campaign. Partners need proof of ROI, and data-driven reporting builds stronger, more profitable relationships.

How Oomph Can Accelerate Your Success

If you’re experiencing softer international interest, shorter booking windows, or declining partner satisfaction, you’re facing the same challenges as DMOs nationwide. The organizations pulling ahead aren’t waiting for market recovery—they’re strengthening their digital platforms through strategic content optimization, systematic audience cultivation, and demonstrable partner value creation.

Our proven methodology transforms these challenges into competitive advantages.

We’ll conduct a comprehensive audit of your digital platform against these three strategic pillars, quantify immediate optimization opportunities, and provide your partners with what they need most: qualified, measurable demand. The market headwinds are real, but the right strategic approach can help you maintain resilience and emerge stronger when conditions improve. Let’s navigate these challenges together.

The Brief

The APPROACH

The Results

THE BRIEF

THE APPROACH

Understanding the Educational Purpose

THE RESULTS

In our design, the whole community drives the learning experience, not just the teachers and coursework. It’s easy for students to connect with others who are taking the same courses, while members-only forums provide a place for productive networking, questions, stories, and support. Some forums are attached to specific lessons, so that the dialogue isn’t just between teachers and students; all members, including alumni, can participate and share their learnings on a given topic.

Another component, the accountability partner system, was crucial for achieving Seraphic’s goal of driving lasting change. Research shows that publicly sharing a goal gives people a 65% chance of success, while reporting to a specific accountability partner boosts that chance to 95%.

Finally, our learning tools were designed to enhance both content comprehension and retention. Course videos were a key feature, designed not just for the course, but for reference over time. Students have the ability to bookmark videos and attach notes to specific sections, letting them revisit important info whenever they need it.

---

THE IMPACT

While online learning has been around for a long time, recent advancements in design and functionality make it possible for learning platforms to have a transformative impact on individuals and across society.

In the case of Atlus∗U, it’s not just the coursework that drives users’ learning; an entire community is mobilized to help you succeed. With a focus on collaborative, lifelong learning, our design brings together farmers from around the world to improve their business, grow healthier food, and protect our world.

Need help building an effective online learning platform? Let’s talk about your goals and how to achieve them.

Oomph has been quiet about our excitement for artificial intelligence (A.I.). While the tech world has exploded with new A.I. products, offerings, and add-ons to existing product suites, we have been formulating an approach to recommend A.I.-related services to our clients. 

One of the biggest reasons why we have been quiet is the complexity and the fast-pace of change in the landscape. Giant companies have been trying A.I. with some loud public failures. The investment and venture capitalist community is hyped on A.I. but has recently become cautious as productivity and profit have not been boosted. It is a familiar boom-then-bust of attention that we have seen before — most recently with AR/VR after the Apple Vision Pro five months ago and previously with the Metaverse, Blockchain/NFTs, and Bitcoin. 

There are many reasons to be optimistic about applications for A.I. in business. And there continue to be many reasons to be cautious as well. Just like any digital tool, A.I. has pros and cons and Oomph has carefully evaluated each. We are sharing our internal thoughts in the hopes that your business can use the same criteria when considering a potential investment in A.I. 

Using A.I.: Not If, but How

Most digital tools now have some kind of A.I. or machine-learning built into them. A.I. has become ubiquitous and embedded in many systems we use every day. Given investor hype for companies that are leveraging A.I., more and more tools are likely to incorporate A.I.

This is not a new phenomenon. Grammarly has been around since 2015 and by many measures, it is an A.I. tool — it is trained on human written language to provide contextual corrections and suggestions for improvements.

Recently, though, embedded A.I. has exploded across markets. Many of the tools Oomph team members use every day have A.I. embedded in them, across sales, design, engineering, and project management — from Google Suite and Zoom to Github and Figma.

The market has already decided that business customers want access to time-saving A.I. tools. Some welcome these options, and others will use them reluctantly.

The Risks that A.I. Pose

Every technological breakthrough comes with risks. Some pundits (both for and against A.I. advancements) have likened its emergence to the Industrial Revolution of the early 20th century. And a high-level of positive significance is possible, while the cultural, societal, and environmental repercussions could also follow a similar trajectory.

A.I. has its downsides. When evaluating A.I. tools as a solution to our client’s problems, we keep this list of drawbacks and negative effects handy, so that we may review it and think about how to mitigate their negative effects:

• A.I. is built upon biased and flawed data
• Bias & flawed data leads to the perpetuation of stereotypes
• Flawed data leads to Hallucinations & harms Brands
• Poor A.I. answers erode Consumer Trust
• A.I.’s appetite for electricity is unsustainable

We have also found that our company values are a lens through which we can evaluate new technology and any proposed solutions. Oomph has three cultural values that form the center of our approach and our mission, and we add our stated 1% For the Planet commitment to that list as well: 

• Smart
• Driven
• Personal
• Environmentally Committed

For each of A.I.’s drawbacks, we use the lens of our cultural values to guide our approach to evaluating and mitigating those potential ill effects. 

A.I. is built upon biased and flawed data

At its core, A.I. is built upon terabytes of data and billions, if not trillions, of individual pieces of content. Training data for Large Language Models (LLMs) like Chat GPT, Llama, and Claude encompass mostly public content as well as special subscriptions through relationships with data providers like the New York Times and Reddit. Image generation tools like Midjourney and Adobe Firefly require billions of images to train them and have skirted similar copyright issues while gobbling up as much free public data as they can find. 

Because LLMs require such a massive amount of data, it is impossible to curate those data sets to only what we may deem as “true” facts or the “perfect” images. Even if we were able to curate these training sets, who makes the determination of what to include or exclude?

The training data would need to be free of bias and free of sarcasm (a very human trait) for it to be reliable and useful. We’ve seen this play out with sometimes hilarious results. Google “A.I. Overviews” have told people to put glue on pizza to prevent the cheese from sliding off or to eat one rock a day for vitamins & minerals. Researchers and journalists traced these suggestions back to the training data from Reddit and The Onion.

Information architects have a saying: “All Data is Dirty.” It means no one creates “perfect” data, where every entry is reviewed, cross-checked for accuracy, and evaluated by a shared set of objective standards. Human bias and accidents always enter the data. Even the simple act of deciding what data to include (and therefore, which data is excluded) is bias. All data is dirty.

Bias & flawed data leads to the perpetuation of stereotypes

Many of the drawbacks of A.I. are interrelated — All data is dirty is related to D.E.I. Gender and racial biases surface in the answers A.I. provides. A.I. will perpetuate the harms that these biases produce as they become easier and easier to use and more and more prevalent. These harms are ones which society is only recently grappling with in a deep and meaningful way, and A.I. could roll back much of our progress.

We’ve seen this start to happen. Early reports from image creation tools discuss a European white male bias inherent in these tools — ask it to generate an image of someone in a specific occupation, and receive many white males in the results, unless that occupation is stereotypically “women’s work.” When AI is used to perform HR tasks, the software often advances those it perceives as males more quickly, and penalizes applications that contain female names and pronouns.

The bias is in the data and very, very difficult to remove. The entirety of digital written language over-indexes privileged white Europeans who can afford the tools to become authors. This comparably small pool of participants is also dominantly male, and the content they have created emphasizes white male perspectives. To curate bias out of the training data and create an equally representative pool is nearly impossible, especially when you consider the exponentially larger and larger sets of data new LLM models require for training.

Further, D.E.I. overflows into environmental impact. Last fall, the Fifth National Climate Assessment outlined the country’s climate status. Not only is the U.S. warming faster than the rest of the world, but they directly linked reductions in greenhouse gas emissions with reducing racial disparities. Climate impacts are felt most heavily in communities of color and low incomes, therefore, climate justice and racial justice are directly related.

Flawed data leads to “Hallucinations” & harms Brands

“Brand Safety” and How A.I. can harm Brands

Brand safety is the practice of protecting a company’s brand and reputation by monitoring online content related to the brand. This includes content the brand is directly responsible for creating about itself as well as the content created by authorized agents (most typically customer service reps, but now AI systems as well).

The data that comes out of A.I. agents will reflect on the brand employing the agent. A real life example is Air Canada. The A.I. chatbot gave a customer an answer that contradicted the information in the URL it provided. The customer chose to believe the A.I. answer, while the company tried to say that it could not be responsible if the customer didn’t follow the URL to the more authoritative information. In court, the customer won and Air Canada lost, resulting in bad publicity for the company.

Brand safety can also be compromised when a 3rd party feeds A.I. tools proprietary client data. Some terms and condition statements for A.I. tools are murky while others are direct. Midjourney’s terms state,

“By using the Services, You grant to Midjourney […] a perpetual, worldwide, non-exclusive, sublicensable no-charge, royalty-free, irrevocable copyright license to reproduce, prepare derivative works of, publicly display, publicly perform, sublicense, and distribute text and image prompts You input into the Services” 

Midjourney’s Terms of Service Statement

That makes it pretty clear that by using Midjourney, you implicitly agree that your data will become part of their system.

The implication that our client’s data might become available to everyone is a huge professional risk that Oomph avoids. Even using ChatGPT to provide content summaries on NDA data can open hidden risks.

What are “Hallucinations” and why do they happen?

It’s important to remember how current A.I. chatbots work. Like a smartphone’s predictive text tool, LLMs form statements by stitching together words, characters, and numbers based on the probability of each unit succeeding the previously generated units. The predictions can be very complex, adhering to grammatical structure and situational context as well as the initial prompt. Given this, they do not truly understand language or context. 

At best, A.I. chatbots are a mirror that reflects how humans sound without a deep understanding of what any of the words mean. 

A.I. systems are trying its best to provide an accurate and truthful answer without a complete understanding of the words it is using. A “hallucination” can occur for a variety of reasons and it is not always possible to trace their origins or reverse-engineer them out of a system. 

As many recent news stories state, hallucinations are a huge problem with A.I. Companies like IBM and McDonald’s can’t get hallucinations under control and have pulled A.I. from their stores because of the headaches they cause. If they can’t make their investments in A.I. pay off, it makes us wonder about the usefulness of A.I. for consumer applications in general. And all of these gaffes hurt consumer’s perception of the brands and the services they provide.

Poor A.I. answers erode Consumer Trust

The aforementioned problems with A.I. are well-known in the tech industry. In the consumer sphere, A.I. has only just started to break into the public consciousness. Consumers are outcome-driven. If A.I. is a tool that can reliably save them time and reduce work, they don’t care how it works, but they do care about its accuracy. 

Consumers are also misinformed or have a very surface level understanding of how A.I. works. In one study, only 30% of people correctly identified six different applications of A.I. People don’t have a complete picture of how pervasive A.I.-powered services already are.

The news media loves a good fail story, and A.I. has been providing plenty of those. With most of the media coverage of A.I. being either fear-mongering (“A.I. will take your job!”) or about hilarious hallucinations (“A.I. suggests you eat rocks!”), consumers will be conditioned to mistrust products and tools labeled “A.I.” 

And for those who have had a first-hand experience with an A.I. tool, a poor A.I. experience makes all A.I. seem poor. 

A.I.’s appetite for electricity is unsustainable

The environmental impact of our digital lives is invisible. Cloud services that store our lifetime of photographs sound like featherly, lightweight repositories that are actually giant, electricity-guzzling warehouses full of heat-producing servers. Cooling these data factories and providing the electricity to run them are a major infrastructure issue cities around the country face. And then A.I. came along.

While difficult to quantify, there are some scientists and journalists studying this issue, and they have found some alarming statistics: 

• Training GPT-3 required more than 1,200 MWh which led to 500 metric tons of greenhouse gas emissions — equivalent to the amount of energy used for 1 million homes in one hour and the emissions of driving 1 million miles. GPT-4 has even greater needs.
• Research suggests a single generative A.I. query consumes energy at four or five times the magnitude of a typical search engine request.
• Northern Virginia needs the equivalent of several large nuclear power plants to serve all the new data centers planned and under construction.
• In order to support less consumer demand on fossil fuels (think electric cars, more electric heat and cooking), power plant executives are lobbying to keep coal-powered plants around for longer to meet increased demands. Already, soaring power consumption is delaying coal plant closures in Kansas, Nebraska, Wisconsin, and South Carolina.
• Google emissions grew 48% in the past five years in large part because of its wide deployment of A.I. 

While the consumption needs are troubling, quickly creating more infrastructure to support these needs is not possible. New energy grids take multiple years and millions if not billions of dollars of investment. Parts of the country are already straining under the weight of our current energy needs and will continue to do so — peak summer demand is projected to grow by 38,000 megawatts nationwide in the next five years. 

While a data center can be built in about a year, it can take five years or longer to connect renewable energy projects to the grid. While most new power projects built in 2024 are clean energy (solar, wind, hydro), they are not being built fast enough. And utilities note that data centers need power 24 hours a day, something most clean sources can’t provide. It should be heartbreaking that carbon-producing fuels like coal and gas are being kept online to support our data needs.

Oomph’s commitment to 1% for the Planet means that we want to design specific uses for A.I. instead of very broad ones. The environmental impact of A.I.’s energy demands is a major factor we consider when deciding how and when to use A.I.

Using our Values to Guide the Evaluation of A.I.

As we previously stated, our company values provide a lens through which we can evaluate A.I. and look to mitigate its negative effects. Many of the solutions cross over and mitigate more than one effect and represent a shared commitment to extracting the best results from any tool in our set

Smart

• Limit direct consumer access to the outputs of any A.I. tools, and put a well-trained human in the middle as curator. Despite the pitfalls of human bias, it’s better to be aware of them rather than allow A.I. to run unchecked
• Employ 3rd-party solutions with a proven track-record of hallucination reduction

Driven

• When possible, introduce a second proprietary dataset that can counterbalance training data or provide additional context for generated answers that are specific to the client’s use case and audience
• Restrict A.I. answers when qualifying, quantifying, or categorizing other humans, directly or indirectly

Personal

• Always provide training to authors using A.I. tools and be clear with help text and microcopy instructions about the limitations and biases of such datasets

1% for the Planet

• Limit the amount of A.I. an interface pushes at people without first allowing them to opt in — A.I. should not be the default
• Leverage “green” data centers if possible, or encourage the client using A.I. to purchase carbon offset credits 

In Summary

While this article feels like we are strongly anti-A.I., we still have optimism and excitement about how A.I. systems can be used to augment and support human effort. Tools created with A.I. can make tasks and interactions more efficient, can help non-creatives jumpstart their creativity, and can eventually become agents that assist with complex tasks that are draining and unfulfilling for humans to perform. 

For consumers or our clients to trust A.I., however, we need to provide ethical evaluation criteria. We can not use A.I. as a solve-all tool when it has clearly displayed limitations. We aim to continue to learn from others, experiment ourselves, and evaluate appropriate uses for A.I. with a clear set of criteria that align with our company culture. 

To have a conversation about how your company might want to leverage A.I. responsibly, please contact us anytime.

Additional Reading List

• “The Politics of Classification” (YouTube). Dan Klyn, guest lecture at UM School of Information Architecture. 09 April 2024. A review of IA problems vs. AI problems, how classification is problematic, and how mathematical smoothness is unattainable.
• “Models All the Way Down.” Christo Buschek and Jer Thorp, Knowing Machines. A fascinating visual deep dive into training sets and the problematic ways in which these sets were curated by AI or humans, both with their own pitfalls.
• “AI spam is already starting to ruin the internet.” Katie Notopoulos, Business Insider, 29 January 2024. When garbage results flood Google, it’s bad for users — and Google.
• Racial Discrimination in Face Recognition Technology, Harvard, 24 October 2020. The title of this article explains itself well. 
• Women are more likely to be replaced by AI, according to LinkedIn, Fast Company, 04 April 2024. Many workers are worried that their jobs will be replaced by artificial intelligence, and a growing body of research suggests that women have the most cause for concern. 
• Brand Safety and AI, Writer.com. An overview of what brand safety means and how it is usually governed.
• AI and designers: the ethical and legal implications, UX Design, 25 February 2024. Not only can using training data potentially introduce legal troubles, but submitting your data to be processed by A.I. does as well.
• Can Generative AI’s Hallucination Problem be Overcome? Louis Poirier, C3.ai. 31 August 2023. A company claims to have a solution for A.I. hallucinations but doesn’t completely describe how in their marketing.
• Why AI-generated hands are the stuff of nightmares, explained by a scientist, Science Focus, 04 February 2023. Whether it’s hands with seven fingers or extra long palms, AI just can’t seem to get it right.
• Sycophancy in Generative-AI Chatbots, NNg. 12 January 2024. Human summary: Beyond hallucinations, LLMs have other problems that can erode trust: “Large language models like ChatGPT can lie to elicit approval from users. This phenomenon, called sycophancy, can be detected in state-of-the-art models.”
• Consumer attitudes towards AI and ML’s brand usage U.S. 2023. Valentina Dencheva, Statistica. 09 February 2023
• What the data says about Americans’ views of artificial intelligence. Pew Research Center. 21 November 2023
• Exploring the Spectrum of “Needfulness” in AI Products. Emily Campbull, The Shape of AI. 28 March 2024
• AI’s Impact On The Future Of Consumer Behavior And Expectations. Jean-Baptiste Hironde, Forbes. 31 August 2023.
• Is generative AI bad for the environment? A computer scientist explains the carbon footprint of ChatGPT and its cousins. The Conversation. 23 May 2023