Boutique SEO Consulting, Automation & Entity Salience
The one with Tommy Delorme and Ismael Girard
What we talked about
- Boutique consulting model vs agency
- Pricing SEO: fixed price vs hourly vs value-based
- 3-month SEO sprints
- Referral-based lead generation and the 'Go-Giver' approach
- ICP: high-value B2B/B2C service businesses
- Automation with n8n, Make, Moonlit and DataForSEO
- Keyword research workflow with psychographic research
- SERP-based vs semantic clustering
- Google Natural Language API and entity salience
- Embeddings and cosine similarity for chunk optimization
- GEO as semantic SEO; Common Crawl
- Fear-based marketing around 'SEO is dead'
The questions we dug into
Q1How does your consulting/boutique model work and who do you serve?
A two-person consultancy with a boutique-agency feel and no project managers — direct contact with every client. They work mostly with service-based businesses (nationwide/worldwide) and some e-commerce stores, keeping a small, close client list.
Asked by Nik · answered by Tommy Delorme
Q2How do you price your work — by the hour?
They moved from fixed per-deliverable prices (which caused underpricing because no site or client is the same) to a monthly hour allocation (e.g., ~23–30 hours) spent on valued deliverables plus consulting, reporting and meetings. Conversations focus on value brought, not tasks. It isn't scalable to 50 clients, but it supports higher retainers with fewer clients.
Asked by Sara · answered by Tommy Delorme & Ismael Girard
Q3How do you get clients and choose your ICP?
Almost entirely referrals and word of mouth, rooted in Tommy's previous consulting clients and LinkedIn visibility, following the 'Go-Giver' principle of giving value first. ICP: white-collar service firms (insurance, law, accounting) with 50–500 employees, where each lead is valuable, so fewer clicks justify SEO investment.
Asked by Nik · answered by Tommy Delorme & Ismael Girard
Q4How do two people handle all the client work — which automation tools do you use?
They use n8n, Make and Moonlit, with workflows organized in Notion. Automated: keyword research via DataForSEO APIs, SERP-based clustering scripts, content briefs, SERP scraping and QA. The rule is to never sacrifice quality for speed: push automation until quality drops, then stop at that ceiling.
Asked by Nik · answered by Tommy Delorme & Ismael Girard
Q5Can you walk through your keyword research process in more depth?
They start with a client interview and personas, then psychographic research (even via a ChatGPT/Gemini prompt) to find pain points, turned into 30–100 seed terms. Pull People Also Ask, related searches, autocomplete and keyword ideas from multiple sources/APIs to build a keyword universe, then SERP-based clustering and SERP analysis. They treat it as market research, including competitors, backlinks and a content audit.
Asked by Sara · answered by Tommy Delorme & Ismael Girard
Q6Can you demo the internal content tool you built?
It uses Google's Natural Language API to show the most salient entities Google detects in text. Salience depends on placement and phrasing, not keyword frequency (e.g., 'At Fujifilm, we make cameras' makes the brand most salient). A second step uses embeddings (e.g., Gemini) and cosine similarity to check how well chunks align with target and fan-out queries.
Asked by Sara · answered by Tommy Delorme & Ismael Girard
Q7What are your predictions for SEO, GEO and LLMs in 2026, and how do you explain it to clients?
Ismael: fear sells, and AI hype is scaring CEOs; SEOs must learn how LLMs work to explain it transparently, since there's no 'Search Console for LLMs'. Every client asked for a 2026 GEO plan. Tommy: GEO is essentially semantic SEO — associate your brand with target entities. Tests show guest posts naming a 'best X' can influence LLMs; being in Common Crawl matters. ~85% remains the same as SEO.
Asked by Nik · answered by Ismael Girard & Tommy Delorme
Lines that stuck with us
- Ismael: 'We never want to sacrifice quality for speed. Never.'
- Tommy: 'SERP-based clustering should be the only way to cluster.'
- Tommy: 'Machines are dumb and cheap' — they spend as little energy as possible understanding content.
- Ismael: salience is 'not about occurrence but more about placement.'
- Tommy: GEO 'means semantic SEO basically'; ~85% remains the same as SEO.
Note: This page was AI-generated from the podcast transcript and manually reviewed for accuracy and context. While we’ve made every effort to ensure it reflects the original conversation, mistakes or omissions are still possible. If you spot something that doesn’t look right, please reach out and let us know.