Dark Intent, Pre-Purchase Signals, and Buyer Readiness

Competitive advantage is moving upstream. That’s the core finding from Skift’s research on relevance in the AI era, and it should make every performance marketer uncomfortable. Waiting for a product search or a site visit isn’t a strategy anymore. It’s a concession. The real window of influence now sits in a phase most teams never see: the pre-purchase period where buyers quietly form opinions, compare options, and build mental shortlists before a single pixel fires. That phase has a name. It’s called dark intent, and it’s already reshaping how serious brands compete.

The Invisible Decision Process (And Why Your Stack Can’t See It)

Here’s a useful analogy. Astronomers detect planets they can’t directly observe by watching how surrounding objects move. Dark intent works the same way. Someone reads three comparison articles about CRM platforms over two days. They ask a generative AI assistant about integration options. Their LinkedIn comments shift from general frustration about their current tools toward very specific questions about feature gaps. None of those actions trigger a retargeting pixel. None of them populate a keyword auction. But the purchase decision? It’s already forming.

Most marketing stacks weren’t built for this. They depend on explicit signals: branded searches, product page visits, cart additions, demo form submissions. Those signals are real and valuable. They’re also the final 10-15% of the journey. Gartner’s research on B2B buying behavior puts the number bluntly: buyers complete roughly 70% of their decision process before they ever talk to a sales team. That 70% is where dark intent lives. If your campaigns only activate on explicit signals, you’re competing for scraps while the actual decision happened somewhere else entirely.

Three Signal Categories. One Scoring Model.

Pre-purchase intent detection runs on three overlapping signal categories. Generative AI models process each and assign probabilistic purchase-readiness scores. Most teams get the signal categories right but ignore how they interact. That’s the mistake.

Content consumption arcs. A single blog post about “best project management tools” means almost nothing. Five pieces of content consumed over three days, moving from general overviews to feature comparisons to pricing breakdowns, means everything. That progression arc is the signal. Generative agents track these arcs across publishers and content platforms using contextual classification, not individual tracking pixels. Frequency matters less than trajectory.

Conversational AI queries. ChatGPT, Gemini, Perplexity, and enterprise copilots are now handling millions of purchase-related questions that used to flow through Google. A query like “which email platform handles transactional and marketing sends from one dashboard” carries serious purchase intent. As agentic referral traffic keeps growing, brands that can interpret and score these conversational signals will have a genuine structural advantage. Right now, almost nobody’s doing it.

Social comment sentiment shifts. Passive liking is noise. Active commenting with specific objections or feature requests is a signal. When someone moves from scrolling a vendor’s LinkedIn posts to posting a pointed question about implementation complexity, that behavioral shift marks movement along the consideration curve. Platforms like Brandwatch and native social APIs can capture this evolution. No single comment tells you much. The trajectory over time tells you everything.

Vibe-coding frameworks let marketers and growth engineers synthesize all three categories into a unified purchase-readiness score. Expressed as a confidence threshold between 0 and 1, not a binary. The approach extends how intent-based lead scoring already works for service businesses, but it pushes the scoring window dramatically further upstream.

Funnel Stage Mapping: Which Signal Goes Where

Not every signal carries equal weight. Running a demo ad against someone at a 0.2 confidence score burns budget. Serving educational content to someone at 0.8 wastes their time. The fix is a practical mapping of signal types to funnel stages, so campaigns activate with the right message at the right moment. This matters more than people think, because the mismatch between message and stage is one of the most expensive mistakes in performance marketing.

When to Actually Start Spending: The Threshold Question

Set the activation threshold too low and you fund curiosity. Set it too high and the buyer already has a shortlist that doesn’t include you. Both errors are expensive in different ways. Based on early data from brands testing contextual AI scoring, here’s a practical starting framework.

• Content syndication and awareness: Activate at 0.2 confidence. Low cost per touch, broad educational messaging.
• Paid social consideration: Activate at 0.4. Use paid social strategy built for mid-funnel engagement, not direct response.
• Direct response and demo offers: Activate at 0.65. The prospect has shown sustained, multi-signal purchase behavior across multiple touchpoints.
• Sales outreach and high-touch personalization: Activate at 0.8. Human time is finite. Only deploy it when signal density genuinely justifies the cost.

Treat those numbers as hypotheses. Run two-week sprints. Measure downstream conversion rates at each threshold and adjust. If your 0.4 cohort converts at nearly the same rate as your 0.65 cohort, pull the activation threshold down for direct response. If your 0.2 cohort shows zero progression after 30 days, raise the floor for content syndication spend. The data tells you what to do. Most teams ignore it and keep the original thresholds forever.

Does It Actually Beat Retargeting?

Early indicators say yes. Particularly on lifetime value. Here’s the logic worth understanding: retargeting captures people who already visited your site. They already know you exist. You’re either reinforcing a decision they’ve already made or chasing one they’ve already rejected. Pre-purchase intent audiences are a fundamentally different population. You’re reaching people during the formative stage of their evaluation, which means you shape preference instead of chasing it.

To measure this properly, run a clean A/B test with strict audience separation. One campaign targets site visitors using standard pixel-based retargeting. A parallel campaign targets a dark intent audience built from contextual AI signals, specifically people who score 0.5 or above but have never visited your site. Zero overlap between cohorts. Run both for at least 30 days, then measure cost per acquisition, first-purchase average order value, 90-day repeat purchase rate, and 12-month LTV.

Early-mover brands testing contextual pre-purchase scoring through platforms like The Trade Desk and custom generative agent pipelines are reporting 20-35% higher LTV from pre-purchase audiences compared to retargeting pools. The likely reason is anchor positioning. When you reach a buyer before they’ve formed a shortlist, your brand becomes the reference point. Every competitor gets evaluated against you, not the other way around.

A Six-Step Playbook for Your First Dark Intent Campaign

Concrete steps. No fluff.

The brands that get serious about dark intent detection won’t just improve their ROAS. They’ll change when they enter the competitive conversation. The decision is moving upstream. Your campaigns should follow it there.

Frequently Asked Questions

What is dark intent in marketing?

Dark intent refers to purchase-readiness signals that buyers emit before performing traditional intent actions like product searches or site visits. These signals include content consumption patterns, conversational queries to AI assistants, and shifts in social comment sentiment. They’re “dark” because most marketing stacks can’t detect them with standard pixels and keyword-based tools.

How does pre-purchase intent detection differ from retargeting?

Retargeting reaches people who already visited your site or interacted with your brand. Pre-purchase intent detection identifies prospects who are actively considering a purchase in your category but haven’t yet visited any vendor’s site. It captures buyers earlier in the decision process, often before they’ve formed a shortlist.

What tools are needed to detect dark intent signals?

You need a combination of social listening APIs, contextual content classification tools, and a scoring model (often built with vibe-coding platforms or lightweight ML frameworks). Platforms like The Trade Desk, Brandwatch, and custom generative agent pipelines can serve as the foundation. The key requirement is the ability to synthesize multiple signal types into a unified purchase-readiness score.

What confidence threshold should I use to activate campaigns?

Start with awareness campaigns at a 0.2 confidence score, paid social consideration at 0.4, direct response at 0.65, and sales outreach at 0.8. Treat these as hypotheses and adjust based on measured conversion rates across each threshold during two-week testing sprints.

Do pre-purchase intent audiences outperform retargeting on LTV?

Early data from brands testing contextual AI scoring suggests pre-purchase audiences deliver 20-35% higher lifetime value than traditional retargeting pools. This is likely because reaching buyers before they form a shortlist allows your brand to become the anchor option in their evaluation process.