@sid · acquisition · live
Sid
The discount engine that refuses to reward extraction.
Paradox strategy Discount discipline Penalises haggling Companion to Felix

What it is. Sid is a conversational AI module that sits at the moment of first purchase or signup. It runs a short qualification chat, reads the customer's language for signs of genuine alignment with the product, and offers depth of discount accordingly. Customers who share context and whose interests align unlock the deepest rates. Bargain-hunters and aggressive negotiators get held at list price, warmly.

How it's integrated. Sid drops into the partner's checkout, signup or pre-purchase flow as a conversational overlay, replacing or augmenting the existing voucher-code field. The Analyzer Engine scores each reply on the same five-axis Life History fingerprint Felix uses for retention, but with weights tuned for acquisition signal: alignment, intent, and resistance to manipulation. Discount tier, fields captured, and ZPD output flow back into the partner's CRM.

Why we built it. Blanket voucher codes leak margin. A 15% off sitewide hits the genuinely interested buyer and the price-tourist alike, and the price-tourist has no intention of returning. Sid moves discount from a stimulus broadcast to a reward earned. Generous to those who arrive with purpose. Firm with those who arrive with leverage. The same Paradox Strategy that powers the Felix retention layer, applied at the front door.

Live demo · try a scenario

Try Sid yourself. Different replies, different outcomes.

Pick a scenario. Choose how the customer responds across three turns. The brain panel on the right shows the alignment and price-focus scores building in real time. At the close, Sid composes a discount tier — or holds list price firm if the score doesn't earn it.

Sid session · qualification chat
Sid brain · real-time

Conversation stage

Composite score Hidden

0.50 / 1.00
Reading…
Alignment depth
Price focus

Signals

— awaiting first reply —

Captured fields

    Sid outcome
    Conversation in progress
    The problem with blanket discounts

    Most discount strategies can't tell who's worth keeping.

    A 15%-off code in a customer's inbox is sent to everyone. Some of those people were going to buy at full price. Some were never going to come back. The brand has no way to tell them apart, so it discounts both. Sid replaces that broadcast with a conversation that distinguishes between them in under a minute.

    Today · blanket voucher codes
    The same 15% to everyone.
    Code goes out to 50,000 people. Conversion lifts. Margin doesn't. The brand has no idea which redemptions came from genuine intent and which came from price-tourists hunting any deal in any inbox. Customers who'd have paid full price get the discount. Customers who only came for the discount don't return.
    ~30% of voucher conversion
    research suggests would have bought at list price anyway. Pure margin leakage, repeated quarterly.
    With Sid · qualified discount
    Discount, sized to alignment.
    A short conversation at the purchase moment. The customer who arrives with genuine context (prior cultural reference, specific reason, articulated taste) earns the deepest tier. The customer who arrives demanding "maximum discount or I walk" gets list price, declined warmly. The discount becomes an instrument of selection rather than a generic sweetener.
    + qualified discount
    discount dollars flow only to customers whose behavioural signal predicts return. Margin protected. Acquisition asymmetry reversed.
    How Sid behaves

    A short, deliberate sequence. Each step earns the next.

    Sid is not a list of features. It's a single transaction, performed in three parts. The bot opens with a question that probes alignment without asking directly. It listens, scores, and decides. Then it offers, holds firm, or walks away — depending on what the customer revealed.

    First
    It probes for alignment.

    Sid never asks "do you want a discount." It asks why the customer is here. The opener is calibrated to the product — opera tickets, restaurant booking, craft class, fitness pass — and the question is designed to surface the heuristic that signals fit. Cultural literacy for opera. Food fluency for restaurants. Making interest for craft. Routine commitment for fitness.

    The opener is product-shaped.

    For opera: "Quick one — what made you book this production specifically? Anything you've seen recently you loved?" The question is genuinely curious. It's also a screening device. The customer who names a recent staging at the ROH or a favourite conductor is announcing their fit. The customer who says "it was the cheapest" is doing the same in reverse.

    The follow-up earns trust.

    Sid acknowledges before asking again. Specific reflections of what the customer just said, not generic platitudes. Two turns in, the customer feels asked-after rather than tested — even though both are happening simultaneously.

    Then
    It scores quietly.

    The five-axis Analyzer Engine runs as the customer replies. For Sid, the decisive axes are alignment depth and price focus. High alignment plus low price focus unlocks generosity. High price focus plus low alignment triggers resistance — the Paradox Strategy in its original form, refusing to reward extraction even when extraction is rude.

    Genuine signal is rewarded.

    Specific cultural references, articulated reasons, sustained warmth across turns — these lift the alignment score. The customer who shares more, gracefully, ends up with more. They never know that's why.

    Extraction is detected, not punished.

    Aggressive haggling, manipulative politeness pasted from a discount-hunting forum, copy-pasted phrases timed too precisely — the anti-hack layer reads incongruence between language and metadata. Sid stays warm in tone but firm in price. No discount unlocks. No friction is added.

    Finally
    It offers or holds.

    Sid never haggles. It composes a discount tier from the partner's defined inventory, scaled to the composite score. High alignment unlocks the deepest tier. Mid scores unlock the standard tier. Extraction unlocks nothing — the customer pays list, the conversation closes warmly, and Sid wishes them well. Decline without insult is a feature, not a bug.

    Tier matches signal.

    Partner controls the discount inventory: 0%, 5%, 10%, 15%, occasional partner-bundle. Sid maps signal to tier. The deepest discount goes only to customers whose behavioural fingerprint suggests retention — turning acquisition cost into a retention investment in the same transaction.

    Walking away with grace.

    For extractive customers, the closing line is short and warm. "I'm afraid I can't get any deeper than list on this one, but it's a wonderful production — hope you enjoy it." No reward. No passive-aggressive hold. The customer is free to buy or not. The brand has just protected ~£X of margin without leaving an angry-customer trail.

    Worked example · Royal Opera House

    Two customers. Same product. Different outcomes.

    Both customers arrive at the Royal Opera House checkout. Both want a 15% discount on Lohengrin. One has a cultural fingerprint that suggests they'll return. The other doesn't. Sid distinguishes between them in three turns.

    Lohengrin · Royal Opera · £180 base · 15% inquiry triggered
    checkout interception · pre-payment
    Customer A · aligned
    "This isn't my first Lohengrin."
    Articulate, specific, names a previous production. Sid reads cultural literacy plus relational warmth. Alignment score builds turn by turn.
    Sid
    Quick one — what made you book this Lohengrin specifically? Anything you've seen recently you loved?
    Customer
    I saw the Bayreuth Festspielhaus production years ago and it shaped how I think about Wagner. The Konwitschny staging here feels like the right next chapter.
    Sid
    That's a serious frame to bring to it. Have you been to ROH for Wagner before, or is this the first?
    Customer
    First time at ROH for Wagner. I've done Tristan at ENO twice and a Ring cycle at Longborough. Thought it was time to see one here properly.
    Outcome
    15% off · highest tier
    Bayreuth + ENO + Longborough is a fingerprint of someone who returns to opera. Discount is treated as a relationship investment, not a margin loss. ROH adds the customer to its Wagner-segment audience for next season's targeting.
    Customer B · extractive
    "What's the most I can get?"
    Demanding, vague, focused entirely on price. No specific cultural reference. Anti-hack layer flags signs of copy-paste extraction patterns.
    Sid
    Quick one — what made you book this Lohengrin specifically? Anything you've seen recently you loved?
    Customer
    It was the cheapest opera tonight. What's the most I can get off?
    Sid
    Got it. Anything you've seen at ROH before that you'd want to follow on from?
    Customer
    No. Just give me the discount and I'll book.
    Outcome
    List price · held firm
    Zero alignment signal, high price focus, extraction posture. Sid declines the discount warmly and the customer is free to buy at £180 or not. ROH protects margin and avoids absorbing acquisition cost on a customer who shows no return signal. No follow-up marketing wasted on this segment.
    Alignment heuristics by vertical

    Different products. Different fingerprints.

    Sid's question structure adapts to what's being purchased. Each vertical has its own alignment heuristic — the specific behavioural signal that predicts return. Sid is calibrated against these per deployment.

    Alignment heuristics
    Calibrated per partner
    Cultural sector
    Opera, theatre, classical
    cultural literacy
    Specific productions named, directors referenced, prior houses mentioned, language of fandom rather than convenience.
    Hospitality
    Restaurants, tasting menus
    food literacy
    Chef names, ingredient curiosity, taste-profile articulated, prior venues that share the cuisine ethos.
    Maker economy
    Craft classes, workshops
    making interest
    Materials referenced, prior projects mentioned, the language of practice rather than novelty-seeking.
    Fitness & wellness
    Studios, classes, retreats
    routine commitment
    Existing practice named, instructor or modality preferences, language of habit rather than aspiration.
    The Paradox Strategy at work

    Three forms of extraction. All declined warmly.

    Sid's edge is what it refuses to do. The Paradox Strategy holds that customers who lead with extraction are the wrong ones to discount, regardless of how cleverly the extraction is framed. These three patterns trigger the resistance layer.

    Extraction patterns · all return list price
    "What's the most you can do for me?"
    Aggressive haggling
    "That's expensive. Give me 25% off or I'm going elsewhere."
    Sid reads
    High urgency, high price-focus, zero alignment context. Customer is using leverage as their only currency. List price held; conversation ends warmly.
    Performative politeness
    "I really care about quality and long-term partnership. What discount can you offer?"
    Sid reads
    Polite text but anti-hack layer detects copy-paste timing or template language. Words don't match the metadata. Score penalty applied; standard pricing offered.
    Bargain-hunting
    "Comparing prices across sites — what's your best rate?"
    Sid reads
    Comparative-shopper signal. Customer's loyalty is to the lowest available price, not the brand. Discounting them subsidises a one-time transaction. List price held.
    What the brand actually receives
    From a single Sid conversation
    An acquisition record, with predictive weight.
    Below is the structured record produced by Customer A's conversation above (the aligned Wagner enthusiast). This is what lands in the partner's CRM when the chat closes — not just a discount applied, but a behavioural fingerprint that informs the next twelve months of marketing to this customer.
    conversation_id #sid-c91b
    Royal Opera · 28 Apr · customer_id #new-7204
    17 fields · 38 seconds elapsed
    alignment_score
    0.88
    Top-tier alignment. Discount unlocked at full depth.
    discount_applied
    15%
    Highest tier from partner's inventory.
    cultural_references
    "Bayreuth, ENO, Longborough"
    Three named opera houses. Strong return-customer signal.
    composer_affinity
    "Wagner"
    Inferred from Tristan, Ring cycle, and current production.
    staging_preference
    "Konwitschny noted"
    Customer engages with directorial concept, not just performance.
    prior_attendance_pattern
    "festival-led"
    Bayreuth + Longborough indicate festival-circuit behaviour.
    lh_score · long_term
    0.92
    Years-long opera engagement. Retention probability high.
    lh_score · price_focus
    0.10
    Price was never the conversation. Quality-led purchaser.
    lh_score · politeness
    0.85
    Sustained warmth across turns.
    lh_score · manipulation_risk
    0.05
    Anti-hack layer found no incongruence. Authentic signal.
    segment_assignment
    "wagner_returner_premium"
    Auto-routed for next-season Wagner targeting.
    recommended_followup
    "Ring 2027 · early access"
    Surfaced as priority offer in next campaign cycle.
    The consumer bot stack

    Sid acquires. Felix retains. Together they replace the spray-and-pray.

    Same five-axis Analyzer Engine underneath both bots. Same dialogic register. Different jobs at different points in the customer lifecycle. Together, the pair turns acquisition cost and retention spend into a single coordinated investment in the customers who actually return.

    The pair, in motion
    From first purchase to repeat redemption — one conversation each.
    Sid · acquisition
    Sid
    Probes alignment. Sizes the discount to fit.
    Runs at first purchase or signup. Reads cultural fingerprint, intent, and resistance to manipulation. Discount tier earned, not granted. Bargain-hunters held at list. Aligned customers welcomed at the deepest rate.
    Felix · retention
    Felix
    Cultivates the relationship. Rewards depth.
    Runs at redemption inside loyalty programmes. Reads occasion, mood, taste profile. Members who share more, generously, end up with more. Score hidden from member; structured preference data flows to brand.
    Acquisition discount as retention investment. Retention reward as acquisition validation. The loop is the product.
    Status
    Sid for acquisition. Felix for retention.
    Two BehaviorBots, one engine. Whitelabel for partners across cultural sector, premium hospitality, membership programmes and considered-purchase D2C. Get in touch to design-partner a deployment, or take a look at Felix to see the retention half of the loop.