The Funnel That Doesn't Exist Anymore
The traditional marketing funnel — Awareness → Consideration → Purchase → Loyalty — was a useful mental model for a world where advertising was interruptive and linear. The brand broadcast a message, potential customers received it, a predictable percentage converted.
That world is gone.
In 2025, the customer journey to purchase involves non-linear discovery, peer research in channels the brand can't track, AI-assisted information synthesis, multiple device switches, varying attribution windows, and purchase decisions that happen after a sequence of touchpoints that last-click models attribute to the final step.
The funnel isn't wrong — it's incomplete. And teams that plan and measure as if the traditional funnel is an accurate model are making systematically bad resource allocation decisions.
The Paths That Aren't Being Measured
Dark social. A significant percentage of content discovery and sharing happens in places that analytics can't track: WhatsApp messages, private Slack channels, Discord servers, text messages, email forwards between individuals. When a prospect finds your article because someone in their industry Slack shared it, your analytics attributes the session to "direct" or "organic search" when they come back later to look it up. The dark social contribution is invisible.
Research from SparkToro suggests that for most B2B brands, a substantial portion of "direct" traffic is actually dark social attribution. The brands making decisions based purely on last-touch attribution are systematically undercounting the contribution of channels that produce dark social sharing.
AI-assisted discovery. An increasing percentage of prospects are using ChatGPT, Perplexity, Claude, and similar tools to synthesize information in the early research phase. The prospect who asks an AI assistant "what are the best marketing agencies in Tampa for SEO?" and then visits your site based on the AI's response shows up in analytics as "direct" — but the AI was the discovery mechanism.
This path is growing and currently invisible in standard attribution models.
Podcast and long-form content. Attribution for audio content is notoriously difficult. A prospect who hears your name mentioned on an industry podcast, thinks about it for two weeks, searches for you directly, and then purchases attributes in most models as branded search or direct. The podcast's contribution is invisible.
Extended consideration cycles. B2B purchase decisions routinely involve 6-18 month consideration cycles. A first touchpoint that happened fourteen months ago is treated as unrelated to the eventual purchase in standard attribution windows. The long-form influence of brand awareness and thought leadership content is systematically under-credited in short-window attribution models.
The Rebuilt Funnel Model
The model I actually use when planning programs:
Layer 1: Awareness / Reach (building the universe)
The goal: be known by people who fit your ideal customer profile, even before they have active need. Brand content, thought leadership, organic social, PR and press, podcast appearances, speaking — these activities build awareness in a universe of potential future buyers. The ROI is long-term and attribution is imprecise.
This layer is systematically underfunded in performance-marketing-dominated organizations because the ROI isn't attributable in standard models. It's also the layer that makes every downstream conversion cheaper, because brand-aware prospects convert at higher rates and lower CPCs than cold prospects.
Layer 2: Interest / Discovery (capturing active consideration)
The goal: when a prospect enters active consideration mode, show up where they're looking. SEO for the research queries they're running. Content that matches the specific questions they're asking. Presence on the platforms where peers share recommendations.
The key behavior signal: a prospect moving from passive awareness to active research. This is detectable through behavioral signals (visiting the pricing page, returning to the site multiple times, engaging with specific solution content), but it's often happening partly in channels you can't track.
Layer 3: Evaluation / Trust Building (reducing friction to decision)
The goal: make the decision easy for a prospect who is genuinely considering you. This is where case studies, testimonials, transparent pricing, process explanation, and competitive differentiation do their work. The friction to decision is highest at this layer because the perceived risk is real.
Most landing page and CRO optimization happens here. The mistake: trying to optimize this layer before fixing attribution and top-funnel investment.
Layer 4: Decision / Conversion (the measurable event)
The goal: facilitate the specific action that converts prospect to customer. This is the layer that standard attribution models over-credit because it's the measurable endpoint. The conversion is the output of every layer above it; treating it as the input misunderstands where the work happened.
Layer 5: Retention / Expansion (the durable revenue)
For most businesses, the majority of revenue comes from existing customers. The funnel ends at purchase only if you're thinking about acquisition, not if you're thinking about business economics. Customer success, post-purchase communication, and expansion within accounts are the highest-ROI marketing activities in businesses with recurring revenue models.
The Attribution Approach That's More Honest
Given the tracking limitations, the most honest attribution approach combines:
Multi-touch attribution with appropriate time windows. Minimum 180-day attribution window for considered purchases; 90-day for faster cycles. Distribute credit across the touchpoints in the journey rather than giving 100% to the last.
Self-reported discovery tracking. "How did you hear about us?" on inquiry forms and in onboarding surveys captures dark social, AI discovery, and long-form content attribution that no tracking system can capture. The data is imperfect but captures paths that analytics entirely misses.
Branded search volume as a proxy. Growth in branded search volume over time indicates increasing awareness and consideration even when specific touchpoints aren't attributable. A 30% increase in branded searches following a PR campaign or thought leadership push is the footprint of activity that doesn't appear in last-click data.
Cohort analysis for long-cycle evaluation. For B2B with long consideration cycles, evaluating which content a cohort of customers engaged with in the 6-18 months before purchase reveals the actual contribution of awareness-layer content to eventual revenue.
Key Takeaways
- The traditional linear funnel is incomplete, not wrong — the real customer journey is non-linear and partially invisible to standard tracking
- Dark social, AI-assisted discovery, and extended consideration cycles are untracked paths that represent significant discovery volume
- The rebuilt funnel model: Awareness (be known) → Interest (show up in active research) → Evaluation (reduce decision friction) → Conversion (measurable event) → Retention (durable revenue)
- Standard attribution over-credits bottom-funnel and under-credits top-funnel brand investment — systematically producing bad resource allocation
- More honest attribution: multi-touch with long windows + self-reported discovery + branded search trends + cohort analysis
- Awareness layer underfunding makes every downstream conversion more expensive — brand awareness subsidizes performance marketing efficiency
- The funnel ends at purchase only in acquisition thinking; business economics require extending it through retention and expansion