Three-Shift Search Transformation Model
Assess search disruption across input, interface, and user role
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 5
- Confidence
- 99%
The Three-Shift Search Transformation Model evaluates AI search through three separate but interacting changes. First, how people search expands beyond typed keywords to voice, video, images, sketches, and mixed inputs. Second, the search experience moves from a standard list of links toward generated, query-specific interfaces that may look different for each person. Third, the role of the searcher changes: the person supplies an initial signal, while the AI clarifies, rewrites, expands, or infers the effective query. Looking at only rankings misses the larger disruption. Marketers should map each shift, determine how it affects content eligibility and user behavior, and revise measurement assumptions because individualized queries and interfaces make traffic less predictable than traditional keyword-based search.
Origin
Extracted from Marketing Against The Grain
Core principles
- 01Search disruption changes more than result ranking
- 02Input modalities determine what users can express
- 03Generated interfaces alter how results are consumed
- 04AI agents increasingly refine intent on the user’s behalf
- 05Strategy must account for all three shifts together
How to run it
- 1
Audit input change
Document every modality users may employ to express a need, including text, voice, video, images, and sketches.
Pro tip Test how the same intent appears in different modalities.
Watch out Do not assume typed keywords will remain the dominant input.
- 2
Audit interface change
Analyze whether the search system returns fixed result components or generates a custom experience around the query.
Pro tip List the actions users can complete without visiting an external site.
Watch out A generated interface can alter attribution and click behavior even when it cites third parties.
- 3
Audit role change
Determine whether the user constructs the final query or whether the AI clarifies and improves it on the user’s behalf.
Pro tip Track the inferred intent and follow-up questions, not only the initial prompt.
Watch out Keyword data may reveal less about the actual query executed by the agent.
- 4
Model the combined impact
Estimate how multimodal inputs, bespoke interfaces, and agent-refined intent jointly affect reach, traffic, and conversion.
Pro tip Use scenario ranges instead of a single traffic forecast.
Watch out Historical click-through assumptions may not survive the interface transition.
- 5
Adapt the acquisition strategy
Develop accessible content and assets that can be understood, cited, remixed, or embedded across the new search experience.
Pro tip Measure inclusion and influence as well as direct clicks.
Watch out Optimizing only traditional ranking factors leaves major parts of the new system unaddressed.
In the wild
A parent begins with an incomplete party-planning request. The AI asks about setting and interests, infers a richer query, and builds a custom interface mixing organized information and visual elements rather than returning ten standard links.
→ The searcher shifts from repeatedly rewriting keywords to guiding an agent that constructs both the query and the experience.
A shopper records a broken component and asks by voice where to find a replacement. The AI interprets the video, clarifies dimensions, and generates a comparison interface with compatible products and installation guidance.
→ The brand’s discoverability depends on machine-readable imagery, specifications, video, and data rather than text keywords alone.
Common mistakes
Treating AI search as another algorithm update
Ranking changes are only one component when inputs, interfaces, and query construction are also changing.
Forecasting from historical clicks
Custom interfaces and agent-completed tasks can materially change how often users visit source sites.
Tracking only the initial prompt
The AI may transform a poor initial prompt into a substantially different effective query.
Is it for you?
Best for
Organizations whose discovery, acquisition, or product usage depends heavily on search behavior.
Not ideal for
Teams treating AI search as a conventional ranking update with stable interfaces and predictable click-through rates.
From the transcript
“First of all, it's changes how we actually search.”
“The search UX, which I think is the biggest thing, is going to change, right?”
“The role of the searcher changes. The searcher is like an input, but not the driver of the search.”
From the episode
Google Launches Gemini AI (And It’s Better Than GPT-4)