Adaptive AI Moderation: 4 Dimensions of Real Qual Research

Adaptive AI moderation is the methodology that determines whether an AI-moderated interview produces genuine qualitative insight or simply collects responses through a scripted chatbot with a conversational veneer. The distinction rests on four dimensions: conversationally adaptive (non-deterministic probing that follows unexpected threads), contextually adaptive (tailoring every interview to the participant’s demographics, role, and history), value-adaptive (matching research depth to business impact by segment), and hypothesis-adaptive (the research itself getting smarter as interviews accumulate). These four dimensions separate real qualitative AI research from predetermined branching logic marketed under friendlier names. Most platforms that claim “adaptive” capabilities implement one dimension at best. Genuine adaptive moderation requires all four working simultaneously, and the difference in insight quality is not incremental. It is categorical.

This framework matters because the AI-moderated interview space is growing fast, and the terminology is growing faster than the methodology. Teams evaluating AI research platforms need a clear framework for distinguishing genuine adaptive intelligence from dynamic questioning with better branding.

What Is Adaptive AI Moderation?

Adaptive AI moderation is an interview methodology in which the AI moderator adjusts its behavior across multiple dimensions during each conversation based on real-time signals from the participant, cumulative learning from prior interviews, and the strategic priorities of the research study.

The critical word is “adaptive,” and it means something specific. It does not mean the AI selects from a menu of pre-written follow-up questions based on keyword matching. It does not mean branching logic where response A triggers question path B. It means the AI generates genuinely novel responses to what the participant says, informed by the full context of the conversation, the participant’s profile, the business value of their segment, and the evolving state of the research hypotheses.

This is a fundamentally different architecture from what most AI research platforms implement. The industry’s dominant approach is dynamic questioning — a system where researchers define a question tree with conditional branches, and the AI navigates that tree based on participant responses. Dynamic questioning is better than a static survey. It is not, however, adaptive. It is scripted navigation with natural language polish.

Adaptive AI moderation, by contrast, is non-deterministic. The moderator cannot predict in advance which questions it will ask any given participant, because those questions depend on what the participant actually says, how they say it, what their profile indicates about their context, and what the cumulative research has surfaced so far. This non-determinism is the feature, not the limitation. It is what allows AI-moderated interviews to reach the depth that critics like Nielsen Norman Group have argued AI cannot achieve.

The four dimensions of adaptive AI moderation are:

1. Conversationally Adaptive — Non-deterministic probing that follows the participant’s actual language, emotions, and reasoning rather than forcing them through a predetermined question sequence.

2. Contextually Adaptive — Adjusting tone, vocabulary, question framing, and probing depth based on who the participant is: their role, seniority, demographics, cultural context, and relationship with the product or brand.

3. Value-Adaptive — Allocating research depth proportionally to the business impact of each participant segment. Enterprise churners generating hundreds of thousands in ARR receive more exploratory interviews than trial users who never converted.

4. Hypothesis-Adaptive — Using cross-interview learning to reallocate research effort in real time. As early interviews confirm certain hypotheses, the AI spends less time on those areas and redirects probing toward open questions that remain unresolved.

Each dimension addresses a different failure mode of scripted AI moderation. Together, they produce research that is qualitatively different from what branching logic can deliver, regardless of how sophisticated that branching logic becomes.

Why Do Most AI Moderators Fall Short?

Nielsen Norman Group published research arguing that AI cannot conduct deep qualitative discovery. Researchers on Reddit regularly express skepticism that AI interviews produce anything beyond surface-level responses. These critiques deserve serious engagement, not dismissal, because they are largely correct about the AI moderation tools those critics have evaluated.

The problem is not that AI is incapable of depth. The problem is that most AI moderation platforms implemented the wrong architecture. They built sophisticated branching logic systems — dynamic questioning engines where researchers define question trees, the AI selects the appropriate branch based on participant responses, and the result looks conversational without being genuinely adaptive.

This architecture has a ceiling. A branching system, no matter how many branches it contains, can only explore the territory its designers mapped in advance. It cannot discover what the researchers did not anticipate. And the entire point of qualitative research is to discover what you did not anticipate. If you already knew what participants would say, you would not need to conduct the research.

Dynamic questioning platforms produce interviews that feel natural to participants. The conversational interface is polished. The transitions between topics are smooth. But the underlying logic is deterministic: this response triggers that follow-up. A participant who mentions an unexpected frustration that falls outside the question tree receives a generic “tell me more” prompt rather than a targeted probe that connects their frustration to the specific context they described two minutes earlier.

This is why critics are right to be skeptical. Most AI moderators they have encountered operate on branching logic. Branching logic cannot ladder 5-7 levels deep into emotional and identity-level motivations, because the researchers who designed the branches did not know where those levels would lead for each individual participant. The branches were written before the conversation started. Laddering methodology requires that each probe be generated from the specific content of the previous response, which is structurally impossible in a branching system.

The industry made a reasonable engineering decision — branching logic is simpler to build, easier to validate, and faster to ship — that created a methodological limitation most teams did not notice until researchers started asking hard questions about depth. The answer is not to abandon AI moderation. The answer is to build AI moderation that is genuinely adaptive.

Dimension 1: Conversationally Adaptive — Non-Deterministic Probing

The first and most fundamental dimension of adaptive AI moderation is conversational adaptiveness: the AI’s ability to generate follow-up questions that are genuinely novel responses to what the participant said, rather than selections from a pre-written menu.

In practice, this means the AI maintains a dynamic model of the conversation as it unfolds. When a participant says, “We switched providers because the reporting wasn’t flexible enough,” a branching system would route to the pre-written “reporting” follow-up path. An adaptive system recognizes that the word “flexible” is the operative term and generates a probe specifically about what flexibility means to this participant: “When you say the reporting wasn’t flexible enough, what specifically were you trying to do that you couldn’t?”

That follow-up was not pre-written. It was generated from the participant’s specific language. And the next follow-up will be generated from whatever the participant says in response to it.

This is how structured laddering methodology actually works in practice. The moderator does not have a script. The moderator has a methodology — a framework for deciding what to probe, when to probe deeper, and when to move on — and the specific questions emerge from the conversation itself. User Intuition’s AI moderation operates on this principle: the methodology is fixed, the questions are not.

The depth this produces is measurable. Adaptive conversational probing consistently reaches 5-7 levels of depth per topic area, moving from surface attributes through functional consequences to psychosocial consequences and eventually to identity-level values. A branching system typically plateaus at 2-3 levels because the question tree was not designed to explore the specific territory each participant’s responses reveal.

Consider the difference in a churn interview:

Branching system approach: “Why did you cancel?” followed by “Was it pricing, features, or support?” followed by “What about the features didn’t meet your needs?” — three levels, each constrained by the pre-defined response categories.

Adaptive approach: “What led to the decision to stop using the platform?” followed by a probe generated from the participant’s specific answer (perhaps they mentioned a failed implementation), followed by exploration of who was involved in the implementation failure, how that failure affected their standing with their team, what it meant for their ability to advocate for new tools in the future, and ultimately what it revealed about the participant’s professional identity and risk tolerance. Each question is novel, generated from the previous response, and could not have been scripted in advance.

The second path produces insight. The first produces data that confirms what the research team already suspected.

Non-deterministic probing also detects and pursues emotional signals that scripted systems miss entirely. When a participant says “the pricing was fine, I guess,” a branching system records a neutral response about pricing. An adaptive system recognizes the hedging language — “fine” qualified by “I guess” — and generates a targeted probe: “You said ‘I guess.’ What was the part that didn’t feel entirely fine?” That single follow-up, generated from a two-word signal, can unlock an entirely different understanding of why the customer churned.

Dimension 2: Contextually Adaptive — Every Participant Gets a Tailored Interview

The second dimension addresses a problem that most AI moderation platforms do not even acknowledge: every participant arrives with a different context, and the interview should adapt to that context before the first question is asked.

Contextual adaptation means the AI adjusts its tone, vocabulary, question framing, pacing, and probing depth based on what it knows about the participant — their role, seniority, demographics, cultural background, language, relationship with the product, purchase history, and any other relevant attributes from the research brief or CRM data.

A C-suite executive who has been a customer for three years receives a fundamentally different interview experience than a junior team member who signed up for a free trial last month. Not different questions about different topics — different calibration of how the same research objectives are explored.

For the executive, the AI uses strategic framing: “How does this tool fit into your broader operational priorities?” The probing is assertive and direct, matching executive communication norms. The AI can reference their tenure and spending history to make questions specific rather than generic.

For the junior team member, the AI uses experiential framing: “Walk me through how you typically use the tool during your workday.” The probing is more exploratory and open-ended, allowing the participant to reveal their context before the AI narrows the inquiry. The AI does not assume knowledge of organizational dynamics that a junior employee may not have visibility into.

This contextual calibration extends across cultures and languages. User Intuition supports research across 50+ languages, and contextual adaptation means more than translation. In high-context cultures, direct questions about dissatisfaction can produce unreliable data because social norms discourage explicit criticism. The AI adjusts its approach — using indirect probing, scenario-based questions, and third-person framing to elicit genuine responses within culturally appropriate parameters.

In low-context cultures, the same adjustment would waste time and frustrate participants who prefer directness. The AI calibrates accordingly.

The practical impact is data quality. Participants who feel that the interview is calibrated to their level of expertise and communication style engage more deeply. They disclose more candidly. They stay in the conversation longer. This is one of the reasons User Intuition maintains a 98% participant satisfaction rate — the interview feels like a conversation with someone who understands their context, not a questionnaire wearing a chatbot mask.

Contextual adaptation also means the AI allocates time within each interview based on participant-specific relevance. A product power user receives deeper exploration of feature experience and workflow friction. A recent churner receives deeper exploration of the decision process and competitive alternatives. A prospective buyer receives deeper exploration of evaluation criteria and perceived risk. The research objectives are the same across all participants, but the conversational pathway is optimized for each participant’s unique position.

This is not achievable with branching logic. A branching system would need a separate question tree for every permutation of participant role, seniority, tenure, cultural background, and product relationship. The combinatorial explosion makes it impractical. Adaptive moderation handles the same complexity through real-time generation, adjusting the conversation as it unfolds based on accumulated context.

Dimension 3: Value-Adaptive — Research Depth Matched to Business Impact

The third dimension is the most strategically provocative, and it is where adaptive AI moderation most clearly separates from every other approach to AI research. Value-adaptive moderation means the depth and exploratory scope of each interview is calibrated to the business value of the participant’s segment.

This is a resource allocation decision that traditional research has always made implicitly but never made well. When a research agency charges $15,000-$27,000 for 15-30 interviews, every participant gets roughly equal interview depth because the cost structure demands it. The enterprise customer generating $500,000 in ARR receives the same 45-minute interview as the SMB customer generating $5,000 in ARR. The agency cannot differentiate because each interview consumes the same amount of moderator time.

AI moderation removes this constraint. And removing it changes the research calculus.

Value-adaptive moderation segments participants by their business impact and adjusts interview parameters accordingly. Enterprise churners — customers whose departure represents significant revenue loss — receive interviews designed for maximum exploratory depth. The AI is configured to pursue more topics, probe more aggressively on emotional drivers, explore competitive alternatives in detail, and invest time in understanding organizational decision dynamics. These interviews may run longer, cover more ground, and generate substantially more data per participant.

Trial users who never converted receive efficient, focused interviews designed to identify the primary conversion barrier quickly. The AI still probes for depth on the barrier it identifies, but it does not conduct a 40-minute exploration of a user who spent 3 minutes on the platform before leaving. The research investment is proportional to the insight’s potential business impact.

Mid-market customers receive calibrated depth between these extremes. The AI adjusts the interview’s exploratory scope based on segment indicators: contract value, tenure, product usage patterns, support ticket history, and expansion potential.

This is not about treating some participants as less important. Every participant receives a genuine, methodologically rigorous interview. The difference is in how much exploratory territory the AI covers and how deep it probes into each area. A win-loss analysis that gives enterprise deals 50% more interview depth than SMB deals produces better strategic intelligence without increasing total research cost.

The concept is novel — there is virtually no competition for “value-adaptive moderation” in any search results — because it is only possible with AI moderation at scale. When you run 200 interviews in 48-72 hours at $20 per interview, the cost of allocating more depth to high-value segments is negligible compared to the incremental insight it produces.

Traditional research cannot do this. Every human moderator hour costs the same regardless of participant segment. The incentive structure is to standardize interview length and treat all participants equally, which sounds fair but is strategically wasteful. Value-adaptive moderation is honest about the reality that a churned enterprise customer’s motivations matter more to revenue outcomes than a casual free-trial user’s impressions, and it allocates research resources accordingly.

Dimension 4: Hypothesis-Adaptive — The Research Gets Smarter Mid-Study

The fourth dimension is where adaptive AI moderation becomes a fundamentally different category from any prior approach to qualitative research, including human-moderated research. Hypothesis-adaptive moderation means the research itself improves while it is running, using cross-interview learning to refine what subsequent interviews explore.

Every research study begins with hypotheses. A churn study might start with three hypotheses: pricing is the primary driver, competitor features are pulling customers away, and implementation friction is creating early-stage churn. In traditional research, the moderator guide is written around these hypotheses before the first interview, and the same guide is used for all 15-30 interviews.

In hypothesis-adaptive moderation, the AI tracks how each hypothesis is performing across the accumulating interview data. If the first 20 interviews strongly confirm that implementation friction is causing early-stage churn — with consistent, detailed testimony and strong emotional signals — the AI recognizes that this hypothesis requires less exploratory time in subsequent interviews. The evidence is already robust.

Simultaneously, if those same 20 interviews reveal an unexpected fourth theme — that customers are churning because their internal champion left the organization and no one inherited the relationship — the AI recognizes an open question that deserves deeper exploration. Subsequent interviews allocate more probing time to understanding champion dynamics and less to confirming what is already well-evidenced.

By interview 50, the research is substantially more targeted than it was at interview 1. By interview 100, the study has converged on the open questions that remain genuinely unresolved, with confirmed findings already backed by deep evidence trails. By interview 200, the research has effectively conducted its own iterative refinement — something that would require three or four separate traditional research phases conducted over months.

This cross-interview learning is possible because the AI has access to the cumulative data from all prior interviews in the study. It is not starting fresh with each participant. It carries forward the emerging patterns, the confirmed themes, the open questions, and the unexpected signals, and it uses that accumulated intelligence to make each subsequent interview more strategically targeted.

The implication is significant: a 200-interview adaptive study is not just a larger version of a 20-interview study. It is a qualitatively different research instrument. The later interviews are asking better questions than the early ones because they are informed by everything the earlier interviews revealed. The Customer Intelligence Hub makes this cumulative learning searchable and persistent, so it compounds across studies rather than resetting with each new project.

This dimension has zero competition in search results because it requires capabilities that most AI moderation architectures do not possess. Branching logic systems cannot learn across interviews because each interview follows a static question tree. Hypothesis-adaptive moderation requires a dynamic research model that updates in real time as evidence accumulates — an architecture that is substantially more complex to build but categorically more valuable to the research teams using it.

How Do the Four Dimensions Work Together?

The power of adaptive AI moderation emerges when all four dimensions activate simultaneously. Consider a practical example: a B2B SaaS company running a churn study through User Intuition.

The study includes 200 churned customers across three segments: enterprise ($200K+ ARR), mid-market ($20K-$100K ARR), and SMB (under $20K ARR). The research team has three initial hypotheses about why churn is occurring.

Interview 1: Enterprise churner, VP of Operations, 3-year customer.

Value-adaptive dimension activates: The AI recognizes this as a high-value segment and configures for maximum exploratory depth — more topics, deeper probing, longer conversation.

Contextual dimension activates: The AI adjusts to executive-level communication. Strategic framing, direct probing, reference to the participant’s tenure and likely organizational visibility.

Conversational dimension activates: The VP mentions that “the platform worked fine for our team, but leadership started asking questions we couldn’t answer with the data it provided.” The AI recognizes “questions we couldn’t answer” as the operative phrase and generates a targeted probe: “What specific questions was leadership asking that the data couldn’t address?” This leads to a 5-level exploration revealing that the company’s new CFO required ROI evidence that the platform’s reporting could not produce, making the VP’s continued advocacy for the tool politically untenable.

Hypothesis-adaptive dimension activates: None of the three initial hypotheses covered “internal political dynamics around ROI evidence.” The AI flags this as a new emergent theme and prepares to explore it in subsequent enterprise interviews.

Interview 37: Mid-market churner, Marketing Director, 8-month customer.

Hypothesis-adaptive dimension has evolved: By interview 37, the “internal champion loss / ROI evidence” theme has been confirmed across 6 enterprise interviews. The AI now proactively explores this theme in mid-market interviews to test whether it applies beyond the enterprise segment.

Contextual dimension activates: The AI adjusts for a marketing director who has been a customer for 8 months — different framing than a VP with 3 years of tenure.

Conversational dimension activates: The Marketing Director reveals that they personally loved the platform but their new CMO brought in a competitor from their previous company. The AI pursues this thread 5 levels deep, uncovering that “vendor relationships following executive hires” is a systematic churn driver that no survey would identify because the churner has no complaint about the product itself.

Interview 150: SMB churner, Founder, 2-month customer.

Value-adaptive dimension activates: SMB segment receives focused exploration — efficient identification of the primary barrier.

Hypothesis-adaptive dimension has matured: By interview 150, three hypotheses are well-confirmed and two unexpected themes have emerged with strong evidence. The AI focuses this interview on remaining open questions about early-stage onboarding friction in SMB accounts.

Conversational dimension activates: The Founder mentions they “couldn’t figure out where to start.” Five levels of adaptive probing reveal that the onboarding problem is not feature complexity but decision paralysis — the Founder had too many research options and no framework for choosing the right one.

This single study, conducted across 48-72 hours at $20 per interview, produces the strategic depth of what would traditionally require three separate research phases, multiple moderator guides, and months of elapsed time. The four dimensions working together create compounding intelligence that no single dimension could produce alone.

Adaptive AI Moderation vs. Dynamic Questioning: What’s the Difference?

The distinction between adaptive AI moderation and dynamic questioning is the most important evaluation criterion for any team selecting an AI research platform, and it is the distinction most vendor marketing actively obscures.

The practical consequence of this distinction shows up in research outcomes. Dynamic questioning produces well-organized data that confirms or disconfirms the hypotheses researchers started with. Adaptive AI moderation produces that same confirmation data plus unexpected discoveries that reshape the research team’s understanding of the problem space.

If your research objective is to validate assumptions, dynamic questioning is adequate. If your objective is genuine discovery — finding out what you did not know to ask about — adaptive AI moderation is the methodology that makes that possible at scale.

The complete guide to AI-moderated interviews covers the broader landscape of AI research platforms, but the adaptive versus dynamic distinction is the single most important architectural choice that determines research quality.

What Does the NN/g Critique Get Right — and Wrong?

Nielsen Norman Group’s critique of AI-moderated research deserves honest engagement. Their core argument is that AI moderators cannot achieve the depth of skilled human interviewers — that the AI follows scripts rather than pursuing genuine discovery, and that participants produce shallower, more rehearsed answers when they know they are talking to a machine.

Here is what they get right: most AI moderators they evaluated do follow scripts. Most AI research platforms implement branching logic systems where the depth of exploration is constrained by the question tree’s pre-mapped territory. When NN/g tested these platforms and found shallow, predictable responses, their methodology was sound and their conclusion was accurate for the tools they evaluated.

Here is what they get wrong: they generalize from the branching logic architecture to AI moderation as a category. The conclusion that AI “cannot” achieve deep qualitative discovery conflates a specific implementation limitation with an inherent technological constraint.

Adaptive AI moderation is the direct answer to the NN/g critique. The four dimensions — conversational, contextual, value-adaptive, and hypothesis-adaptive — address each specific failure mode that NN/g identified:

NN/g concern: “AI follows scripts.” Conversationally adaptive moderation is non-deterministic. There is no script to follow.

NN/g concern: “Participants give shallow answers.” Contextual adaptation creates conversations that feel calibrated to the participant’s level, producing engagement rather than compliance. This is why platforms implementing genuine adaptive moderation achieve participant satisfaction rates like User Intuition’s 98% — participants engage deeply because the conversation meets them where they are.

NN/g concern: “AI misses emotional signals.” Adaptive probing specifically detects hedging, contradiction, and emotional loading, generating targeted follow-ups that pursue those signals rather than routing to a generic “tell me more” prompt.

NN/g concern: “AI cannot discover the unexpected.” Hypothesis-adaptive moderation is specifically designed to discover the unexpected and reallocate research effort toward it.

The NN/g critique is a useful diagnostic of the AI research industry’s dominant architecture. It is not an accurate assessment of what adaptive AI moderation can achieve. The distinction matters for research teams making platform decisions: dismissing all AI moderation based on evaluations of branching-logic platforms is like dismissing all automobiles because you test-drove a golf cart.

How to Evaluate Whether an AI Moderator Is Truly Adaptive

Research teams evaluating AI moderation platforms need a practical framework for distinguishing genuine adaptive intelligence from dynamic questioning with better marketing. Here are the questions that expose the difference.

1. “Show me two interviews with the same research objective where the AI asked substantially different follow-up questions.”

In an adaptive system, two participants discussing the same topic will receive different follow-up questions because their responses were different. In a branching system, participants who give similar initial responses will receive identical follow-ups because they triggered the same branch. Ask the vendor to show you the transcripts side by side and identify where the AI generated novel probes rather than selecting pre-written ones.

2. “How does the AI handle a response that falls outside the question tree?”

This is the definitive test. In a branching system, off-tree responses get routed to generic follow-ups (“Can you tell me more about that?”) because no specific branch was designed for that territory. In an adaptive system, off-tree responses generate targeted probes that reference the participant’s specific language. Ask for examples of how the AI handled unexpected, off-script responses.

3. “Does the AI adjust interview approach based on participant segment or profile?”

If every participant receives the same interview regardless of their role, tenure, or business value, the system is not contextually or value-adaptive. Ask how a C-suite participant’s interview would differ from a junior team member’s, not in content but in calibration. Genuine adaptive moderation produces visibly different conversational approaches for different participant profiles.

4. “Can you show me how the research questions evolved between early and late interviews in a study?”

This tests hypothesis-adaptive capability. If early and late interviews in the same study are asking identical questions in identical patterns, there is no cross-interview learning. An adaptive system should show measurable sharpening of focus in later interviews, with more time spent on open questions and less on confirmed themes.

5. “What is the average probing depth per topic area across your last 10 studies?”

Branching systems rarely exceed 2-3 levels of follow-up per topic. Adaptive systems consistently achieve 5-7 levels. Ask for the data, not the marketing claim. If the platform can report this metric, it likely tracks conversation depth as a quality indicator. If it cannot, depth may not be part of its methodology.

Red flags that indicate scripted branching disguised as adaptive moderation: interviews that feel natural but follow predictable patterns, “AI-generated” questions that are suspiciously well-polished (suggesting they were pre-written), consistent interview length regardless of participant engagement, and an inability to show how the research evolved across interviews.

Getting Started with Adaptive AI-Moderated Research

The four-dimension framework is not theoretical. It is the operational methodology behind how User Intuition conducts every study on the AI-moderated interview platform — from 10-interview concept tests to 500-interview churn analyses.

For teams evaluating whether adaptive AI moderation fits their research needs, the practical entry point is straightforward:

Run a comparison study. Take a research question you recently studied with surveys, traditional qual, or a branching-logic AI platform. Run the same question through an adaptive AI-moderated study. Compare the depth, the unexpected discoveries, and the actionable specificity of the findings. The difference is immediately visible in the transcripts.

Start with a high-stakes use case. Adaptive moderation produces its most visible advantage on research questions where depth matters: churn analysis, competitive win-loss, brand perception, and purchase motivation. These are the domains where branching logic reaches its ceiling and adaptive probing breaks through to identity-level insight.

Evaluate the evidence trail. For teams that need to present findings to leadership or a board, the evidence chain matters. Adaptive AI moderation produces findings that trace directly to specific verbatim quotes from specific participants, with the full conversational context showing how the moderator reached that depth. This is the difference between “customers say pricing is an issue” and “VP-level buyers in the enterprise segment describe pricing anxiety rooted in their inability to demonstrate ROI to a new CFO who requires evidence-based budget justification.”

User Intuition delivers adaptive AI-moderated research at $20 per interview with results in 48-72 hours, drawing from a 4M+ vetted participant panel across 50+ languages. The platform holds a 5.0 rating on both G2 and Capterra and supports enterprise security requirements including ISO 27001 and GDPR compliance.

Explore the AI-moderated interview platform to see how the four dimensions work in practice, or book a demo to walk through a live adaptive interview with your own research questions. For teams that want to understand the full cost comparison, the pricing is transparent: from $200 for a 10-interview study to custom enterprise packages for continuous research programs.

The question is no longer whether AI can moderate qualitative research. The question is whether the AI moderation you are evaluating is genuinely adaptive across all four dimensions — or whether it is branching logic with a better name.