The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
AI is already being used in marketing to develop campaign concepts, generate and adapt video, personalize creative, and respond to live events. Published case studies show how those workflows operate—and report cost, speed, and audience results—but their figures are campaign-specific, not reliable forecasts for other teams.
What marketing teams are using AI to do
The documented examples cover five distinct jobs: exploring ideas, producing synthetic images and video, adapting creative for markets or platforms, personalizing content, and drafting real-time social responses. They are not interchangeable. A workflow for generating a campaign film has different review and data needs from one that uses a customer’s photo or reacts to a live event.
- Concepting: Generate and refine possible campaign directions before committing to production.
- Synthetic production: Create or iterate on image and video assets.
- Localization and variation: Adapt a core idea for languages, markets, recipes, or platforms.
- Personalization: Use a person’s supplied material to create content relevant to them.
- Real-time engagement: Draft responses or experiences around live events or contextual data.
How the workflows look in practice
Lysol: concept selection followed by synthetic production
In an eight-week pilot for Laundry Sanitizer, Lysol used Gemini to examine consumer trends and competitor positioning and generate concepts. People selected three directions to develop. The team then used Veo and Imagen for synthetic production, iterated on imperfect outputs, and used consumer testing to optimize the final assets. The case study also describes a “digital twin” approach using consenting, compensated actors. Lysol VP Benoit Veryser cautioned that experimenting with generative tools is not the same as shipping AI-built creative. Think with Google’s January 2026 Lysol case study reports the company’s results.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Snapple, State Farm, and Lyft: creative tied to context
Three cases illustrate different ways AI can connect creative to an event or place. Snapple and agency Deutsch used Veo 3 to turn brand trivia into stylized short films; human creatives guided and approved frames, and the article says the social assets disclosed AI use. State Farm and Highdive used Gemini to track NBA Finals highlights and draft responses for “Stan the State Farm Stanchion Pad,” followed by team review and generated video. Lyft Local, developed by Lyft and SpecialGuest, used live route data with AI-generated localized narrative, audio, and visuals to turn a ride into a guided discovery experience. These are examples of creative patterns, not a measured comparison of the approaches. Think with Google’s June 2026 article describes the cases.
#1 Best Overall
Dept and Pit Viper: ideation feeding video production
Google’s case study reports that agency Dept used Gemini to develop and refine campaign concepts for Pit Viper, then Veo 2 to generate video scenes. Dept creative director Paul Bjork said the work still depended on curation: “The magic comes from the choices we make and how we use the tools, not the tool itself.” The campaign figures are reported by Google, not a controlled comparison with other production methods. Read the Dept/Pit Viper case study.
WPP’s T&Pm: pre-visualization, localization, and personalization
WPP agency T&Pm used Azure OpenAI and Sora to pre-visualize campaign ideas and create video variations for markets, languages, recipes, and platforms. One example adapted a rice-brand idea for audiences in the U.S., China, and India. Another used a dog’s photo and story to create a personalized animated video for a dog-food campaign. Microsoft describes these workflows as evolving and does not report a quantified performance result. Because personalization can involve user-supplied material, the story also emphasizes privacy, security, and responsible-AI guardrails. Microsoft’s June 3, 2025 customer story provides the details.
Rank #2
What the reported results do—and do not—show
| Case | Reported result | How to interpret it |
|---|---|---|
| Lysol Laundry Sanitizer | 80% lower cost per asset than Lysol’s traditional process; its best AI asset performed nearly identically to its top traditional asset in short-term sales-lift studies. | Figures reported by Lysol in a January 2026 Think with Google case study. They describe this pilot and its measures, not typical savings or a universal sales outcome. |
| Dept/Pit Viper | A 400% decrease in production timeline; 3.8% ad-recall lift in four days, 6.5% lift among people aged 18–24, and reach of 10.3 million users. | Figures reported in Google’s 2025 case study for this campaign. They are not directly comparable to Lysol’s cost and sales-lift measures. |
| Snapple, State Farm, and Lyft | No comparable quantified performance result stated in the June 2026 Think with Google article. | Useful as examples of event-linked and location-linked creative workflows, not as evidence of measured lift. |
| WPP T&Pm | No quantified performance result stated in Microsoft’s June 3, 2025 customer story. | The story describes workflows and guardrails rather than a measured campaign outcome. |
These cases use different metrics and methods, so they do not establish which platform or workflow is best. The available examples also do not support a representative industry-wide estimate of time or cost savings. Treat each published number as a case result tied to its stated company, campaign, and measurement.
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchWhere human direction and review matter
Across the examples, AI supports parts of a process; people make consequential decisions about what should be made and what is ready to publish. In practice, teams can assign human review at the points where risk or brand judgment is highest:
- Before generation: Set the audience, objective, brand constraints, and any limits on data or likeness use.
- During concepting: Select promising ideas rather than treating generated options as approved strategy.
- During production: Inspect outputs, reject weak or misleading material, and iterate where generation is imperfect.
- Before publication: Check the final asset for brand fit, factual accuracy, rights and consent, and any necessary disclosure.
- For live or personalized work: Review proposed responses and apply privacy and safety controls before using live signals or user-supplied content.
The examples show these roles in different forms: Lysol selected concepts and tested assets; Snapple’s creatives guided and approved frames; State Farm’s team reviewed responses; Dept curated ideas and outputs; and T&Pm’s described personalization work included privacy and responsible-AI guardrails. Human review is not an optional polish step when the workflow depends on consent, brand judgment, or contextual accuracy.
How to choose a workflow for your campaign
Start with the job to be done, not the tool. The cases suggest a practical way to scope a pilot and judge whether it helps:
- Name the task. Decide whether the bottleneck is concept exploration, video production, localization, personalization, or timely engagement.
- Define the input and boundaries. Identify what information the workflow needs, whether it includes personal material or likenesses, and what should never be generated or published.
- Specify human checkpoints. Assign who selects concepts, reviews intermediate outputs, approves the final asset, and handles errors or unsuitable content.
- Choose a campaign-specific measure. If the goal is production efficiency, track time or cost per asset; if it is audience response, choose an appropriate campaign measure such as ad recall. Do not treat reach, cost, and lift as interchangeable.
- Compare against your own baseline. Document the existing process and conditions so the pilot’s result can be interpreted in context rather than borrowed from another company’s case study.
A good pilot tests both output quality and the work required to make that output publishable. A fast first draft is not the same as a faster approved asset, and generating many variants is useful only if the team can review and use them responsibly.
Recommended Free Tools
How complete is the evidence?
The examples here are a selected set of published case studies, not an audited catalogue of every deployment suggested by the phrase “24 real deployments.” The source listing does not establish that all 24 cases have been verified. The cases do establish several concrete patterns and reported outcomes, but they do not prove that those outcomes will recur across industries, campaigns, or teams. For decisions, use the workflow examples to frame a test and use your own measurement to judge the result.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
