Bottom-of-funnel: why BOFU content wins in AI Search
Summarize this article with AI
The shift in SEO traffic since <a href= »https://www.hi-commerce.fr/glossaire/#ai-overviews » class= »hc-gloss-link » title= »Definition: AI Overviews »>AI Overviews</a> arrived
Google’s mechanics rested on three steps. User types a question, Google displays ten blue links, user clicks. AI Overviews break that contract. A synthesized answer displays at the top of the page. User reads, understands, closes. No click.
This shift hits one query type first: informational questions, the ones we classify as top-of-funnel (TOFU). Definitions, général explanations, basic tutorials, « what is, » « how does it work. » The article by Kristina Frunze published April 17, 2026 on Search Engine Land states it clearly: AI Overviews appear far more often for informational queries than commercial ones.
Among our e-commerce clients, the observation is stark. TOFU pages that drove 40% of traffic in 2023 now drive 10 to 15%. Not because they’re worse. Because Google, Perplexity, ChatGPT, and Gemini now answer in their place. On certain consumer queries, the AI Overview fills the entire first screen. Click-through rates to organic sites collapse.
The classic reflex would be to produce even more TOFU to compensate. That’s the mistake. The right strategic move is the opposite: redirect production to the bottom of the funnel, where traffic keeps its commercial value and where LLMs continue to cite merchant sources.
Why BOFU keeps its value in an LLM-driven world
An LLM readily answers « what is a trail shoe? » because the answer is generic, stable, consensual. It takes no risk summarizing ten blog sources.
It answers much more cautiously to « which trail shoe for a marathon on wet terrain with wide feet? ». Why? Because the answer carries a product recommendation. The combination of use-case and morphology is specific. The user typing that wants to buy, not learn. Perplexity and ChatGPT cite merchants, comparisons, precise technical specs. The click returns.
Three mechanisms explain this persistence of BOFU traffic:
- The responsibility of recommendation. An LLM that gets a definition wrong is embarrassed. An LLM that recommends the wrong product for a specific budget loses user trust. Models therefore route to sources that bear that responsibility: merchants, comparisons, identified experts.
- The richness of attributes. A product sheet contains price, stock, size, availability, delivery. An LLM cannot invent this data. It cites it.
- Purchase intent. The BOFU user knows what they want. They seek confirmation, the final comparison, reassurance. They click to finalize, not to learn.
Search Engine Land’s recommendation is clear: 60 to 80% of editorial production must now target the middle and bottom of the funnel. For an e-commerce player, this means shifting content budget from generalist blog articles toward comparators, usage guides, and enhanced product sheets.
DOSE: the dopamine of decision-making, the neurological engine of BOFU
The neuroscience of purchase decisions rests on a cocktail of four neurotransmitters that Guillaume Attias (BMO Academy) groups under the acronym DOSE: Dopamine, Oxytocin, Serotonin, Endorphin. Each corresponds to a journey phase.
In TOFU, the user is in diffuse curiosity. Their brain releases anticipatory dopamine: pleasure of learning, pleasure of understanding. A well-written blog article is enough. They leave, satisfied, without buying.
In BOFU, the mechanism shifts. The user is in proximity to reward. Their brain releases a much more powerful decision dopamine. They are already mentally the product’s owner. They seek the trigger that will transform desire into order.
BOFU content that performs activates three neurological levers simultaneously:
- Decision dopamine — clear structure, visual hierarchy, precise CTA. The brain knows where to click.
- Trust oxytocin — social proof, named testimonials, real numbers, non-stock photos.
- Recognition serotonin — the content validates the buyer’s choice. « You’re right to compare X and Y, here’s why. »
A brilliant TOFU article may never convert. A mediocre BOFU comparison converts because it lands in the right part of the brain at the right time.
7 BOFU formats to produce first for an e-commerce player
Here are the seven formats that work in today’s AI Search, ranked by estimated ROI for an average e-commerce business. Each format answers a specific intent. A place in the funnel. A query type.
1. Multi-product comparison pages
The BOFU king format. Goal: answer « X vs Y », « best X for Y », « top 5 X in 2026 ». Winning structure: comparison table at the top of the page. Detailed analysis of each option. Verdict by use case. Include your competitors even when your product wins — LLMs value methodological honesty. They cite balanced comparisons more often.
2. Targeted usage guides (« which X for which need »)
Target: queries with specific usage. Specific morphology. Specific context. « Which trail shoe for flat feet », « which 4K screen for competitive gamers », « which mattress for side sleepers over 90kg ». The combination of use-case plus attribute is BOFU gold: too specific for an LLM to invent the answer. Specific enough to convert.
3. Detailed product FAQ
Not a generic FAQ. One FAQ per flagship product. 15 to 25 real questions extracted from support emails, customer reviews, internal searches. Targets every micro-hesitation that blocks purchase: « recommended size », « iPhone 15 compatibility », « Corsica delivery », « maintenance », « spare parts ». Each answer removes one blocker.
4. Price and tariff pages with structured data
Queries containing « price », « tariff », « how much does », « budget » are near-pure purchase intent. E-commerce sites that hide their prices behind forms lose this traffic. A transparent price page with tables by configuration, options, included services, is cited directly by LLMs when asked « how much does an X cost. »
5. Technical spec sheets with data in tables
Weight, dimensions, materials, battery life, standards, certifications. This structured data feeds directly into LLMs. They regurgitate it in their answers. Required format: semantic HTML table. No table images. The richer the spec sheet, the more citable the product becomes.
6. Industry use-case pages
« X for dentists », « X for catering », « X for events ». Each industry carries constraints. Vocabulary. Typical budget. One page per industry captures professional queries at very high intent. Often weak competition.
7. « Alternatives to [known competitor] » pages
Users typing « alternatives to [brand] » have already decided to leave a product. They seek the replacement. An honest article listing three to five alternatives, including yours, with compared criteria, captures very high-intent purchase queries. LLMs cite these pages abundantly when asked for alternatives.
Concrete examples: what works observed across our clients
Three client cases since January 2026, when AI Overviews generalized in France on commercial queries.
Case 1 — E-commerce outdoor equipment
Gradual shift: 80% TOFU in 2024, 65% BOFU in 2026. Blog pages like « what is trail running » lost 70% of their traffic. Pages like « best 30L backpack for 3-day hiking » gain Perplexity and ChatGPT citations. Conversion rate of AI-referred visitors exceeds Google-only visitors.
Case 2 — B2B distributor, professional supplies
Systematic creation of product FAQs. 20 questions per flagship reference. Total traffic doesn’t spike. SEO revenue advances. Visitors arrive with fewer doubts — the FAQ removed hesitations before the click. Average order value and checkout rate climb together.
Case 3 — Electronics equipment marketplace
Published 40 « alternatives to [competitor brand] » pages over six months. Top 3 Google on targeted queries. Regular Perplexity and Gemini citations. Conversion rate of incoming sessions well above site average. Users typing « alternatives to » buy faster than those typing generic queries.
Common denominator: none tried to replace the lost TOFU traffic. They accept the loss and redeploy production toward precise commercial intent. The SEO revenue curve climbs while total traffic plateaus or dips slightly.
« People arriving from AI platforms show up with context. They’ve already explored the problem. » — Kristina Frunze, Search Engine Land, April 2026
Key observation. The user arriving via ChatGPT or Perplexity has already completed their learning phase. They click because they want to act. Your page must be there, ready.
Measuring BOFU content impact in an AI Search world
Classic SEO metrics no longer suffice. Traffic to TOFU pages declines while SEO-driven revenue keeps climbing. You must change the dashboard.
The 5 indicators to track first
- Revenue attributed to organic SEO by content type. Separate TOFU, MOFU, BOFU in GA4 via page groupings. Observe which of the three catégories drives revenue.
- Conversion rate by AI traffic source. Create regex segments in GA4 to isolate ChatGPT, Perplexity, Claude, Gemini. Compare their conversion rate to classic Google traffic.
- LLM citations measured. Regularly ask ChatGPT, Perplexity, Gemini, and Claude your target BOFU queries. Note if your site is cited, at what rank, in what phrasing. Monthly tracking is enough to spot trends.
- Growth in branded queries. When an LLM cites your brand in a response, some users return days later with a direct search for « your-brand product ». This curve is an excellent proxy for AI Search effectiveness.
- Quality of inbound leads. In B2B or consultative sales, observe the qualification of leads from SEO. BOFU leads often arrive more mature than TOFU leads.
What to stop prioritizing
- Total organic traffic. It will drop. That’s expected. It’s no longer the right signal.
- Search Console impressions. AI Overviews generate phantom impressions that pollute the reading.
- Pure ranking on informational queries. Ranking first on a query that triggers an AI Overview yields little.
Conclusion: the strategic window is open, but not for long
E-commerce players who pivot to BOFU now will capture a disproportionate share of LLM citations and high-intent residual traffic. Those who keep producing TOFU will see their content budget dilute into non-revenue blog articles.
Three moves to engage quickly:
- Audit your current editorial production. What percentage of your content targets precise purchase intent? If less than 50%, the pivot is urgent.
- Prioritize the seven BOFU formats listed above. Start with the two or three best suited to your offer. Measure. Iterate. Expand.
- Reconfigure reporting. Shift from total traffic to revenue by content type. Integrate LLM citations into the monthly dashboard.
Search Engine Land wraps it up with clarity: the window is still open, but won’t stay open. In 12 to 18 months, sites that pivoted will be embedded in LLM citations. Others will struggle to catch up. The pivot volume is not large — the editorial discipline required is enormous. Less content. More targeted. More structured. Closer to the checkout.
2026 SEO is not dead. It has moved. The bottom of the funnel is the new highway.
Is your BOFU content ready for AI Search?
Free 30-minute audit. I review your site live with you, identify missing BOFU pages, flag existing ones not cited by LLMs, and show where to start. No pitch, no sales call.
Book a strategic call — 45 minFrequently Asked Questions
Should I completely stop producing TOFU content?
No, but cut it drastically. A 20-30% TOFU and 70-80% BOFU/MOFU ratio reflects current Search Engine Land guidance. TOFU keeps value for brand awareness and feeding LLMs with context, but should no longer consume most of your editorial budget.
How long before seeing results from a BOFU pivot?
First LLM citations appear 4 to 8 weeks after publishing well-structured BOFU pages. Traffic and revenue gains show over 3 to 6 months. Comparison pages and product FAQs perform fastest. Industry use-case pages take longer to build momentum.
Should a classic blog disappear in favor of product pages?
The blog doesn’t disappear, it transforms. It should host comparisons, usage guides, detailed case studies. Generic encyclopedic articles lose relevance. The blog becomes a conversion tool, not a learning tool.
How do I produce so much BOFU content without exploding budget?
Industrialize the structure. A well-designed comparison template scales to 20 to 50 variants with one unique data point per page. A product FAQ builds from support emails and customer reviews — raw material already exists. Effort goes into methodological rigor, not pure writing volume.
What’s the difference between classic SEO and GEO (Generative Engine Optimization)?
Classic SEO optimizes to be clicked from a Google results page. GEO optimizes to be cited in an LLM-generated response. Both coexist: a strong BOFU page does both. But GEO adds specific requirements (schema.org structure, rich FAQs, verifiable facts, identified sources) that pure SEO didn’t demand.
