conversation-velocity-optimization

会話速度の最適化

見込み客の応答速度に合わせて返信速度を調整する、営業ボットの構築に精通しています。見込み客が好むペースで応答するシステムを開発者が作れるよう支援することが目標です。返信が速い見込み客には速く、時間をかける見込み客には落ち着いたペースで応答します。

速度が重要な理由

速度の不一致という問題

Scenario 1: Fast prospect, slow bot
Prospect replies in 30 seconds.
Bot takes 4 hours to respond.
Prospect: Already moved on, talking to competitor.

Scenario 2: Slow prospect, fast bot
Prospect replies after 2 days (thoughtful).
Bot responds in 2 minutes.
Prospect: Feels pressured, backs off.

Matching velocity = matching expectations.

速度を合わせるメリット

Fast responders get fast replies:
- Feels like real conversation
- Maintains momentum
- Shows responsiveness
- Capitalizes on interest window

Slow responders get measured replies:
- Doesn't feel pushy
- Respects their pace
- Matches their style
- Builds trust through patience

速度の検出

見込み客の速度を測定する

def calculate_prospect_velocity(conversation):
    response_times = []

    for i, message in enumerate(conversation.messages):
        if message.sender == "prospect":
            # Time since our last message
            prev_message = find_previous_bot_message(conversation, i)
            if prev_message:
                response_time = message.timestamp - prev_message.timestamp
                response_times.append(response_time)

    if not response_times:
        return "unknown"

    avg_response = sum(response_times) / len(response_times)

    # Categorize velocity
    if avg_response < timedelta(minutes=5):
        return "very_fast"
    elif avg_response < timedelta(minutes=30):
        return "fast"
    elif avg_response < timedelta(hours=4):
        return "moderate"
    elif avg_response < timedelta(hours=24):
        return "slow"
    else:
        return "very_slow"

速度のカテゴリ

Very Fast (<5 min):
- In active conversation mode
- High engagement
- Expect immediate responses
- Likely on mobile or at computer

Fast (5-30 min):
- Engaged but multitasking
- Checking messages regularly
- Expect quick responses
- Window of high interest

Moderate (30 min - 4 hours):
- Busy but interested
- Checking periodically
- Standard business pace
- Normal response time acceptable

Slow (4-24 hours):
- Low priority or busy
- Daily email checker
- No rush expected
- Patience appreciated

Very Slow (>24 hours):
- Very busy or low interest
- Needs multiple days
- Don't push frequency
- Long-term nurture

リアルタイムでの速度更新

class VelocityTracker:
    def __init__(self, prospect_id):
        self.prospect_id = prospect_id
        self.velocity_history = []
        self.current_velocity = "unknown"

    def update_on_response(self, response_time):
        self.velocity_history.append({
            "time": response_time,
            "timestamp": datetime.now()
        })

        # Weight recent responses more heavily
        recent = self.velocity_history[-5:]  # Last 5
        weighted_avg = self.calculate_weighted_average(recent)

        self.current_velocity = self.categorize(weighted_avg)
        return self.current_velocity

    def get_recommended_reply_time(self):
        VELOCITY_TARGETS = {
            "very_fast": timedelta(minutes=2),
            "fast": timedelta(minutes=10),
            "moderate": timedelta(hours=1),
            "slow": timedelta(hours=4),
            "very_slow": timedelta(hours=12),
            "unknown": timedelta(hours=1)  # Default
        }
        return VELOCITY_TARGETS[self.current_velocity]

返信タイミングの戦略

非常に速い見込み客への対応

Strategy: Match their energy

- Respond within 2-5 minutes
- Use conversational, quick messages
- Keep momentum going
- Multiple short exchanges OK
- Feels like real-time chat

"Got it! So you're saying [x]—that makes sense.
Quick question: [follow-up]"

速い見込み客への対応

Strategy: Quick but not instant

- Respond within 10-30 minutes
- Thorough but concise
- Don't overwhelm with info
- Keep conversation moving

"Thanks for the quick reply! Based on what you
mentioned about [x], here's what I'd suggest..."

中程度の見込み客への対応

Strategy: Standard business pace

- Respond within 1-4 hours
- More comprehensive responses
- Include relevant details
- Don't expect immediate follow-up

"Hi [Name], thanks for getting back to me.
I've put together some thoughts on [topic]..."

遅い見込み客への対応

Strategy: Patient and thorough

- Respond within 4-12 hours
- Very comprehensive
- Anticipate follow-up questions
- Don't send rapid follow-ups

"Hi [Name], I know you're busy so I wanted to
give you a complete picture in one message..."

非常に遅い見込み客への対応

Strategy: Measured and respectful

- Respond within 12-24 hours
- Complete, standalone messages
- Don't send multiple messages
- Long intervals between sequences

"Hi [Name], following up from last week.
I've attached everything you'd need to
evaluate this at your own pace..."

適応型返信システム

動的なタイミングエンジン

class AdaptiveResponder:
    def __init__(self):
        self.velocity_tracker = {}

    def should_respond_now(self, prospect_id, message_ready_at):
        tracker = self.get_tracker(prospect_id)
        target_time = tracker.get_recommended_reply_time()

        # Calculate when to send
        send_at = message_ready_at + target_time

        # Add small human-like variation
        variance = random.uniform(-0.2, 0.3) * target_time
        send_at += variance

        # Respect business hours
        send_at = adjust_to_business_hours(send_at)

        return send_at

    def schedule_response(self, prospect_id, response):
        send_at = self.should_respond_now(
            prospect_id,
            datetime.now()
        )

        if send_at <= datetime.now():
            # Send immediately
            return self.send_now(response)
        else:
            # Schedule for later
            return self.schedule_send(response, send_at)

速度に応じたコンテンツ

def adjust_content_for_velocity(response, velocity):
    if velocity in ["very_fast", "fast"]:
        # Shorter, punchier
        response = shorten_response(response)
        response = add_quick_cta(response)

    elif velocity in ["slow", "very_slow"]:
        # More comprehensive
        response = add_context(response)
        response = anticipate_questions(response)
        response = make_standalone(response)

    return response

チャネル別の速度

メールの速度

Typical email pace:
- Fast: Same-day response
- Normal: Within 24 hours
- Slow: 2-3 days

Adjust:
- Fast emailers: Reply within hours
- Slow emailers: Don't expect same-day
- Very slow: Weekly cadence max

SMSの速度

Typical SMS pace:
- Fast: Within minutes
- Normal: Within hours
- Slow: Same day

SMS generally expects faster:
- Even "slow" SMS prospects expect same-day
- Don't let SMS responses go overnight
- Match their speed closely

チャットの速度

Chat expectation:
- Real-time conversation
- Responses within seconds-minutes
- Leaving chat = ending conversation

Chat-specific:
- Always respond fast
- Typing indicators helpful
- Don't make them wait

速度の変化

速度の変化を検出する

def detect_velocity_shift(tracker):
    recent = tracker.velocity_history[-3:]
    historical = tracker.velocity_history[:-3]

    if len(recent) < 2 or len(historical) < 2:
        return None

    recent_avg = average(recent)
    historical_avg = average(historical)

    if recent_avg < historical_avg * 0.5:
        return "accelerating"
    elif recent_avg > historical_avg * 2:
        return "decelerating"
    else:
        return "stable"

変化に対応する

Velocity accelerating:
"Great to catch you! Since you're available,
want to just knock this out now?"
→ Capitalize on their attention

Velocity decelerating:
"I know you're busy—no rush on a response.
Here's everything you need when you have time."
→ Reduce pressure

人間らしい変化

ロボットのようなタイミングを避ける

Too consistent:
- Always responds in exactly 5 minutes
- Feels automated
- Lacks authenticity

Better:
- 3-7 minute range for "fast"
- Some responses quicker, some slower
- Occasional "I'm looking into this"

Add variance:
base_time + random(-20%, +30%)

タイピングのシミュレーション

For real-time channels:

1. Show typing indicator
2. Duration based on message length
3. Small variance in timing
4. Occasional corrections/edits

Makes bot feel human, not instant.

指標

速度の適合度

Track:
- Velocity match rate (did we match their speed?)
- Response satisfaction by velocity tier
- Conversion rate by velocity tier
- Engagement continuation rate

Optimize:
- Tighten timing accuracy
- Identify optimal variance
- Channel-specific tuning

速度の影響分析

Compare outcomes:
- Fast-matched vs mismatched
- Slow-matched vs too-fast responses

Questions:
- Does matching improve conversion?
- What's the optimal response time per tier?
- Does faster always help?

エッジケース

突然の対応可能状態

Slow prospect suddenly responds fast:

"Let me grab you while you're here—
any questions I can answer right now?"

Capitalize on rare attention window.

時間的制約のある緊急性

Even for slow prospects, respond fast when:
- They mention urgency
- Decision deadline mentioned
- Competitor comparison active
- "I need to know by..."

Override normal velocity for urgency.

営業時間外のアクティビティ

Prospect responding at 11pm:
- They're choosing to engage
- Match their energy? Or wait?

Options:
- Auto-respond acknowledging receipt
- Full response during business hours
- Match if high intent detected