CASE 73 / IBM / Date not verified
Granite: Preview and Reveal
A technical promise becomes easier to judge when someone builds with it.
AI-assisted research and analysis for the SNS Creative Fieldbook
Sources reviewed 2026-11-05
Campaign credits belong to the original creators. They did not write or endorse this analysis.
Credits: The Goat Agency with IBM; award credits include IBM’s Ryan Bares and Goat’s Gregg Tobin. Seven developer or technology creators participated.
Recognition: Entered in Business to Business at the 2025 Shorty Awards. The entry is not labeled a finalist or winner.
Launch / edition: Date not verified. Award years: 2025.
The problem. The idea. The experience.
IBM wanted developers and enterprise decision-makers to understand practical applications of Granite models.
How the idea came to life
Creators introduced a project they planned to build, then returned to reveal the result and explain the model’s role.
How it reached people
Short-form content appeared on LinkedIn, TikTok, Instagram and YouTube, with paid amplification and links to IBM or developer destinations such as Hugging Face.
Supporting sources: Primary agency case → Primary award entry → Primary creator post →
Read the data in context.
The numbers below describe different kinds of outcomes. Their date, scope and reporting source are part of the result.
- Reporting period
- Campaign window not published
- Scope
- Separate distribution sources; not activated users
- Evidence
- Source-reported; not independently audited here
The entrant reports 43 million impressions, over 8,600 organic link clicks and over 59,000 paid link clicks. Exact campaign flight dates are not established in the checked case pages.
The creative mechanism.
Open project setup followed by a concrete result
Show the work product
Abstract technology claims often sound interchangeable. A small working example gives the audience something to inspect and question. The marketing task is to explain the system’s contribution without implying the tool did every part of the work.
Make the second episode necessary
A preview should create a meaningful question, and the reveal should answer it. If the result adds no new information, the sequence is only frequency in disguise. A good technical series can include what failed and what needed human correction.
Respect the audience’s preferred destination
A developer may want documentation or a model repository rather than a sales form. Give the next action the right level of commitment. Forcing every interested viewer through a lead gate can interrupt the very exploration the content created.
Separate popularity from adoption
A video can be admired without the tool being installed or used. Link clicks are closer to intent than impressions but still do not prove adoption. Define activated use, evaluation quality and sales qualification separately, with privacy-respecting tracking.
The intended behavior loop
- 01Recognize a practical problem
- 02Watch a creator propose a build
- 03Return to inspect the result
- 04Visit documentation or a repository
- 05Evaluate the tool in a real workflow
Growth job: Technical curiosity → observed application → developer exploration. This is an interpretation of the mechanism, not an audited attribution model.
A direction to take into the room.
Original editorial interpretation and proposed adaptation. This is a proposed adaptation, not historical campaign execution or a forecast.
The transferable principle
Commission a transparent two-part build that ends with an inspectable result and a destination where the audience can evaluate the product.
Keep these boundaries
Do not imply AI-generated output is reliable without review, or describe clicks as qualified enterprise buyers. Sponsorship and tool limitations should remain visible.
Turn the lesson into a test.
Illustrative adaptation, not a historical campaign fact
A small-team experiment
A workflow-software company sponsors two practitioners to automate one modest, non-sensitive task and publish their findings.
Build the test
- Choose a reproducible problem and protect confidential data.
- Record setup, limitations and human effort.
- Link to documentation and an optional demo rather than only a lead form.
Measure the right thing
Measure documented trials that reach a useful first outcome, not clicks alone.
Know when to stop
Stop if the demonstration requires undisclosed manual work or claims exceed what the system can reproduce.
How this reading fits your brief
Start with the useful part.
Best for: Technical curiosity → observed application → developer exploration
What to adapt: A technical promise becomes easier to judge when someone builds with it.
Prerequisites: Creator contracts and usage rights; Clear sponsorship disclosure; Product and factual review; Defined measurement window.
Smaller-team route: Commission a transparent two-part build that ends with an inspectable result and a destination where the audience can evaluate the product.
Budget: Not established by the cited public sources. No cost-fit claim is made.
Trade-offs and evidence: Read the limits below before choosing a test. There is no guaranteed reach, sales lift or ROAS.
Questions for your next brief
- What could a skeptical viewer inspect?
- What does the second installment prove that the first could not?
Sources & further reading.
Brand and award-entry results are attributed claims, not independent causal audits. Source links open the original third-party pages.
- The Goat Agency: IBM Granite →Primary agency case · Supports: Objective, sequential storytelling, channels and reported outcomes.
- Shorty Awards: IBM x The Goat Agency →Primary award entry · Supports: Entrant status, project mechanics, creator output and named credits.
- Tiffany Janzen: Sponsored Granite quiz project →Primary creator post · Supports: Direct example of disclosed creator demonstration; page shows relative rather than exact date.
Independent educational analysis of external campaigns. These are not SNS client projects, endorsements or claims of involvement. Campaign facts and reported outcomes are attributed to their sources. Explanations, adaptations and test plans are original analysis and hypotheses, not proven causal findings.