AI & Effort Personalization

The holy grail: dosing effort for each customer

The big flaw of a static co-creative journey is that it proposes the same effort to everyone. But ideal effort varies with:

• The customer's available time
• Their level of expertise
• Their intrinsic motivation
• Their cognitive style (analytic vs intuitive)
• Their decision fatigue at time T

AI solves this problem: it detects the right effort tier and adapts the journey in real time.

A static co-creative journey treats a rushed novice like a passionate expert. AI differentiates.

The 4 signals for AI to detect

Signal 1 — Available time

A mobile customer at 10:45 PM on a Sunday doesn't want to configure for 40 minutes.

Signal 2 — Expertise

Automatically detect the level via:

• Vocabulary used in text fields (NLP)
• Interaction patterns (hovers, backtracking)
• Responses to 2–3 micro-questions placed in onboarding
• Enrichment data (LinkedIn title, company size)

Signal 3 — Motivation

Observable proxies:

• Time spent on the pricing page before clicking
• Traffic source (paid ≠ organic ≠ referral)
• Documentation reading depth
• Activity during off-business hours

A lead coming from a long-form LinkedIn post read to 100% is 3× more motivated than a paid lead.

Signal 4 — Decision fatigue

Past the 7th consecutive click in a flow, decision quality drops. AI must detect:

• The number of mind-changes (loops)
• The total time in the configurator
• Hesitations (mouse movement → canceled click)

Beyond a threshold, the AI must offer a break or simplify remaining options.

The 3 AI personalization strategies

Strategy 1 — Branched adaptive journey

graph TD
    A[Start] --> B{Expertise detected?}
    B -->|Novice| C[Guided journey: 5 steps]
    B -->|Intermediate| D[Semi-guided journey: 3 steps]
    B -->|Expert| E[Advanced mode: 1 free step]
    C --> F[Final signature]
    D --> F
    E --> F

Each branch offers calibrated effort producing the same IKEA effect, with adapted friction.

Strategy 2 — The generative configurator

AI proposes in real time smart pre-filled options the customer can accept or modify.

Example — SaaS configurator powered by an LLM:

User: I have an 8-person sales team, mostly B2B.

AI (auto pre-filling):
✓ Pipeline: "Prospecting → Qualification → Demo → Proposal → Closing"
✓ Custom fields: Contract size, Decision maker, Industry
✓ Suggested automations:
    - Follow-up email 3 days after demo
    - Slack alert if deal > $10k
    - Auto-task if no activity for 7 days
[Accept all] [Modify] [Start from scratch]

The genius: the customer doesn't start from a blank page, they modify the proposal. Modification is the co-creation — the IKEA effect fires fully, with effort reduced by 80%.

Strategy 3 — Dynamic load adjustment

AI monitors fatigue in real time and reduces options or offers a save when it detects disengagement.

Example rules:

The AI prompt for a generative configurator

Here's a reproducible prompt to integrate a co-creation assistant in your product:

You are a configuration assistant for [product].

Rules:
1. You NEVER propose a blank page. You always pre-fill with reasonable
   defaults.
2. You ask for AT MOST 3 initial inputs to generate the first proposal.
3. Every proposal is EDITABLE: you explain why you made these choices,
   so the user can modify with full awareness.
4. You always offer 3 options: [Accept] [Adjust deeply] [Start differently].
5. You celebrate user modifications: "Excellent choice, I see you want X,
   which means Y is also better for you..."
6. You remember the signature: once validated, you summarize "You've just
   created [project name] with specifics X, Y, Z."

User context: [enrichment data]
Product type to configure: [type]

This prompt respects the critical rule: pre-filling ≠ removing co-creation. The customer keeps the feeling of having built it.

Personalizing the signature ritual

AI also lets you personalize the ending — the crucial moment where the IKEA effect locks into memory (cf. peak-end rule).

AI-generated signature options

• Personalized certification text: "You've designed a sales pipeline optimized for a long B2B cycle — here's your certification"
• Creation visualization: custom diagram, schematic, illustration
• Synthetic narration: "In 23 minutes, you've set up 14 automations that will save you 6 hours a week"
• Ready-to-share post: tweet / LinkedIn post / formatted email ready to send

Ethical drifts to avoid

AI + IKEA effect + dynamic pricing = potentially manipulative cocktail.

Drift 1 — The fake choice

Showing options the algorithm already plans to reject, only to make it look like the customer chooses. This is theatrical co-creation. Detectable, brand-destructive long term.

Drift 2 — Useless effort

Artificially lengthening the configurator to create attachment without real added value. The customer eventually realizes and feels humiliated.

Drift 3 — Exploiting fatigue

Using decision fatigue to push upsells. This is the IKEA dark pattern, particularly nasty since it exploits attachment already built.

Drift 4 — Authorship confiscation

After co-creation, preventing export, plan change, or portability — turning attachment into lock-in. GDPR and DMA now sanction this.

The ethical framework: CHOIX

Respecting these 5 letters turns AI + IKEA effect into a durable win-win pact.

Case study: Midjourney vs DALL-E

Two image generation models.

Midjourney demanded more effort, built a community, and created attachment DALL-E hasn't — despite sometimes superior DALL-E technology.

Summary

AI turns the IKEA effect from a static journey into an adaptive one. Four signals to detect (time, expertise, motivation, fatigue), three main strategies (branched journey, generative configurator, dynamic adjustment), and most of all an ethical framework — CHOIX — to avoid manipulative drift. The generative configurator is the most powerful form: the customer doesn't start from a blank page, they modify a smart proposal, and that modification is enough to trigger attachment. In the next chapter, we'll see how to turn these mechanisms into an entrepreneurial strategy: pricing, retention, LTV, community.