How We Work

Most agencies pick keywords by search volume and write pages by feel. We don't. Before we write a single word of your website, we model the gross profit every keyword would generate if you ranked for it, then build your entire site around the terms that actually pay. Here is the full process, start to finish. No black box.

Phase 1: Pre-campaign data analysis

The goal of this phase is a keyword list ranked by dollars, not guesses: a spreadsheet that tells us the gross profit each search term is worth to your business if we land you in the top three results.

1. Identify your primary business category.
2. Use that category as the root keyword in Google Keyword Planner.
3. Use the initial search to surface every relevant keyword cluster and sub-cluster in your market.
4. Take the top 10 clusters and sub-clusters and feed them back into Keyword Planner as new seed keywords.
5. Export the full keyword set.
6. Remove keywords that are irrelevant to your business.
7. Remove top-of-funnel and low-intent middle-of-funnel keywords, searches that don't turn into booked jobs.
8. Pull industry benchmarks: average top-3 click-through rate, website conversion rate, and close rate.
9. Determine your average gross profit per service.
10. Replace Keyword Planner's volume ranges with exact monthly search volume from DataForSEO, using a custom Python script we run in Google Colab.
11. Calculate the gross profit each keyword would generate if ranked in the top three.
12. Strip out every field that isn't needed to model gross profit per keyword, leaving a clean, revenue-ranked target list.

This is why our recommendations aren't opinions. By the end of Phase 1 we can tell you which pages will make you money and roughly how much, before we've built anything. See a worked example in how to forecast SEO revenue.

Phase 2: Website

Now we turn that model into a site. Every structural and content decision traces back to the intent and profit data from Phase 1.

1. Map the intent clusters into a site structure, so each page targets a distinct search intent. (More on this in how to structure a website for SEO.)
2. Plan the title, meta description, slug, H1, and intro copy for every page.
3. Interview you to pull specific, non-generic details we can actually use: real numbers, real jobs, real differentiators.
4. Write page content that satisfies the search intent using your interview data.
5. Check every page for thin content.
6. Check every page for generic content.
7. Confirm each page satisfies the search intent it's intercepting.
8. Set up GA4 tracking so we can measure conversion rate from day one.

Phase 3: Post-launch data analysis

The launch isn't the finish line: it's where the model starts getting real data. Once traffic and leads flow, we close the loop.

1. Enrich the keyword model with your actual click-through rate, conversion rate, and close rate from GA4 and your CRM, replacing industry averages with your real numbers.
2. Adjust the site structure strategy based on what the data shows.
3. Adjust keyword targeting to chase the terms that are actually converting to revenue.

Systems-first: how we diagnose and stay update-agnostic

Everything above rests on a bigger principle: we optimize for how Google's ranking system actually works, not for exploits. Google grades pages offline and then serves them through a staged funnel (retrieval, then hand-built signals, then click-based re-ranking, then a little AI on the very top), and we break the whole machine down in how Google search actually works. That matters for two reasons.

Diagnostics. When a client's rankings move, we don't guess. We can reason about which stage changed (did the page fall out of the retrieval set, did a site-quality signal drop, did the click behavior shift?) and fix the actual cause instead of throwing tactics at a wall. Optimizing the system, rather than a trick, is what gives us that visibility in the first place.

Update-agnostic clients. A core update isn't weather to us; it's a change to a specific part of the system. Because we know the architecture, we can tell which part moved and adjust deliberately, so our clients ride out the updates that wipe out competitors chasing shortcuts. Steady, compounding growth beats a spike that a single update can erase.

Marketing Mix Modeling: for multi-location and enterprise accounts

For our top accounts we go further and run Marketing Mix Modeling (MMM), the same statistical modeling large corporate teams use to prove which channels and spend levels actually drive revenue. MMM only works when there is enough data to model: it requires ongoing data collection across multiple locations and multiple markets. Most businesses simply can't generate that volume, which is why full MMM is reserved for our largest clients.

For that reason, we don't do heavy reporting or Marketing Mix Modeling for accounts under $50,000 in monthly spend. Below that threshold, the honest answer is that the data isn't rich enough to model reliably, so we focus your budget on the profit-mapped search strategy above, which is where it earns the most at that stage. If you're a multi-location or multi-market operator at or above that level, MMM is where we pull decisively ahead of most in-house teams.

That's the whole system: model the profit, build to the intent, then let real data rewrite the plan. If you want to see what a data-driven plan looks like for your business, start with a free audit or read our case studies to see the process in action.