← Reading listOUTTHINK
Text

DEEPER STUDY / IBM / 2015

Outthink

Organise an emerging technology proposition around specific business questions rather than a list of abstract capabilities.

Enterprise AI / cognitive computingInternational5 min read
Published by SNS Media Tech

AI-assisted research and analysis for the SNS Creative Fieldbook

Sources reviewed 7 November 2026

The original campaign creators did not write or endorse this analysis.

Original campaign credits: Ogilvy New York; creator portfolios by Juliane Hadem and Sid Tomkins document the platform and print work. A complete individual production roster is not established.
Recognition: No award-level claim is made. Sources establish execution, chronology and creator contributions.
Launch / edition: 2015. Award years: No award claim.

The problem. The idea. The experience.

An unfamiliar technology category needed examples that business audiences could evaluate.

How the idea came to life

A recurring story platform connected cognitive-computing claims with business challenges.

How it reached people

Print, television, online and social stories.

Read the data in context.

The numbers below describe different kinds of outcomes. Their date, scope and reporting source are part of the result.

2015 platform and creator contributions documented; promotional effectiveness figures are not adopted.

Read the campaign story

IBM’s Outthink platform dates to 2015 in creator Juliane Hadem’s portfolio. Her account describes a recurring series of business stories across publications, television, online and social channels, framing IBM Watson within the language of cognitive computing. Sid Tomkins’ portfolio separately documents full-page Wall Street Journal advertisements produced at Ogilvy New York.

This case covers the platform as a family rather than counting every application or industry story separately. The creator accounts establish the communication structure and contributions. They do not independently verify all technical claims or isolate the campaign’s effect on preference and revenue. Historical descriptions also should not be read as a current product catalogue or as proof that an AI system can solve every problem shown in advertising.

Editorial interpretation: an emerging technology becomes easier to discuss when it is attached to a recognisable question. A series of specific problems can provide multiple entry points for different business audiences while retaining one organising idea. The risk is that a strong verb implies an ability broader than the examples support. Teams should make the boundary between a demonstrated capability, an experimental application and an ambition visible rather than allowing the platform to blur them.

Editorial interpretation: repeated stories can also become a disciplined evidence system. Each instalment has an opportunity to explain the input, task, human role and observed result. That is more useful than accumulating impressive category labels. An adaptation should resist the temptation to substitute a dramatic potential outcome for a validated workflow, particularly when the example concerns health, employment, finance or another area where mistakes can materially affect people.

Choose a question with an inspectable answer

Frame the story around a bounded task that can be evaluated, not a sweeping promise to transform an industry. Explain what the system receives and what it produces. If the answer depends on judgement or uncertain evidence, say so. This gives the audience a basis for assessing relevance and prevents the creative platform from turning an exploratory demonstration into an implied operational guarantee.

Show the human work around the technology

Identify who prepares information, checks outputs and decides what action is appropriate. Do not make people disappear simply to create a more magical demonstration. Their role can be part of the value proposition because it reveals how the tool fits into an actual process. A realistic account of supervision and limitations helps buyers understand the resources needed to use the system responsibly.

Maintain an evidence ledger for every story

Record the source, date, scope and status of each capability claim before adapting it into different formats. Distinguish a research prototype from a released product and a reported result from an independently verified one. Revisit the record when the campaign is reused. A continuing technology platform needs active factual maintenance because a compelling old example can become misleading as products and evidence change.

A small test for another brief

Proposed adaptation: demonstrate a low-risk document-sorting task using synthetic material, pairing a problem-led story with a transparent explanation of the tool’s limits and human review.

Measure whether viewers understand the task, can identify the human review step and distinguish demonstrated performance from wider claims. Evaluate the explanation separately from the tool’s actual accuracy on the limited test set.

Stop if the story implies autonomous expertise beyond the evidence, conceals errors or encourages deployment in consequential settings without appropriate validation and oversight.

  • Use a sandbox and invented records, with no sensitive customer or employee data.
  • Compare outputs against a predetermined reference set and disclose errors as well as correct results.
  • Avoid consequential eligibility, medical, legal or financial decisions in the demonstration.

Sources & further reading.

These are educational analyses of external campaigns, not SNS client projects. Source-backed campaign details, editorial interpretation and proposed tests are distinguished. Brand and campaign-entry reports are not independent causal audits.