How I Diagnose Why Organic Growth Stalled

What a growth plateau looks like

I hear a version of the same sentence every few weeks. “We’ve hit a ceiling.”

The offer works, clients are happy, and the team is good at what it does. But organic traffic has flattened or dropped. Organic leads, especially good ones, have stopped climbing. And nobody can say why with any confidence.

That’s a plateau. Growth that came easily for a few years stops coming, even though nothing obvious broke. It’s different from a crash, where something clearly failed or changed. But a plateau is quieter, which is what makes it expensive.

So the team does what teams do. They produce more content, optimize for more keywords, maybe hire a new agency or commission a site redesign. Sometimes all of it at once. And the numbers still barely move, because nobody stopped to ask what was actually holding them down.

First, I confirm it’s actually a stall

I always diagnose before I prescribe. But even before I diagnose anything, I check whether the stall is real and where it lives. Plenty of “traffic problems” turn out to be reporting problems. Analytics scripts fail to load. Channels get misclassified. A tool update breaks something in the workflow.

I measure organic growth in three layers, using my Triple-P Model. Presence asks whether you show up for the questions your buyers ask. Prominence asks whether you get cited or recommended when you do. Performance asks whether you get chosen, which turns into brand searches, direct traffic, and pipeline growth.

Then I look for the pattern. In Search Console, I compare impressions to clicks over 12 to 16 months, and I split branded queries from non-branded ones. Impressions holding while clicks fall tells a different story than both falling together.

In analytics, I separate visits from AI assistants like ChatGPT, Perplexity, Gemini, and Claude from the rest of organic traffic. And in the CRM, I check whether lead quality changed, not just lead volume.

A stall usually shows up in one of four different ways:

• Impressions and clicks both down.
• Impressions up but clicks down.
• Traffic steady but leads flat.
• Leads steady but the wrong kind.

Each shape points somewhere different.

Why traffic can fall while rankings hold

This one confuses a lot of teams right now, so it’s worth its own section.

Your rankings can look fine while your traffic slides, and it isn’t necessarily your fault. SparkToro’s research found that 68% of U.S. Google searches in early 2026 ended without a click to anywhere. When an AI Overview sits at the top of the page, a lot of buyers get their answer and leave.

Seer Interactive’s study of 3,119 search terms across 42 organizations shows how big the gap is. In Q3 2025, organic click-through rate was 1.45% on searches without an AI Overview. With one, it dropped to 0.70% when the brand was cited in the overview, and to 0.52% when it wasn’t.

Two things jump out. Clicks fall either way. But being cited softens the fall, which is why the citation itself is now worth measuring. Seer is careful to note that this doesn’t prove citation causes the higher click rate, since stronger brands may simply get cited more.

The buying journey moved too. Bain & Company found that 85% of B2B buyers end up buying from their “day one” list, the vendors they had in mind before they searched. If you’re not on that list, a good ranking later in the journey does less than it used to.

None of this means organic is dead. In many cases, organic visibility has simply shifted. But if there’s a drop in traffic as a result, it’s a symptom, and the cause can sit almost anywhere.

Symptoms that look like causes

Most stalls arrive with a symptom attached, and the symptom gets treated as the cause. That’s where the money or the intended results (or both) go sideways. Here’s how I read the common ones.

What you see	What it often points to	Where I look first
Traffic down, rankings steady	Buyers getting answers from AI and not clicking	Triple-P (Prominence)
Rankings fine, pipeline flat	Ranking for the wrong questions	Bullseye Method (Fit)
Leads coming in, poor fit	Reaching the wrong buyers, or the right ones too early	Bullseye Method, OATH
Traffic steady, conversion slipping	Message pitched at the wrong awareness stage	OATH
Claims ignored, sales cycle long	Weak or generic proof	FORCEPS
AI names competitors, not you	Missing from the sources AI trusts	Triple-P, Four Doors
Everything a little soft	No clear position to rank, cite, or remember	Power Positioning

Same flat line on the dashboard. Very different fixes.

How I diagnose a plateau, in four phases

I use the same loop on every diagnostic. I call it IDEAL: Investigate, Decide, Execute, Analyze, Learn. It’s a loop, not a checklist. Each stage feeds the next, and the last stage feeds the first one of the next lap. Skip a stage and the loop breaks, which is why so many audits end in a report nobody uses.

In a diagnostic, I run it in four phases. I Investigate the problem, Decide on the solution, then Execute it. Analyze and Learn travel together at the end, because checking the result and learning from it happen in the same review. Here’s a breakdown of the four phases.

Phase 1: Investigate buyer questions, not keywords

Investigate is intelligence gathering without premature conclusions. The job is to read the system as it actually works, not as the team believes it works. Most often, the team or the C-suite will tell me what they think the problem is, and often they’re looking in the wrong direction.

So I pull wide. Search Console and analytics, with AI referrals separated out. The CRM, to see which leads turned into revenue and where they came from. Sales call recordings and support tickets, because that’s where the buyer’s real questions live.

The biggest shift in how I investigate is mapping real buyer questions, not just keywords or question-based queries. I collect the long, conversational questions buyers actually ask, then run them through ChatGPT, Gemini, Perplexity, and Google to see who gets named and which sources get cited. The full method is in how I run an AI visibility audit, and The AI Shortlist Study shows it at scale.

Then I talk to people. The leadership team, a few people doing the work (particularly team members who work closely with clients), and a few clients. Each group sees a different part of the stall.

The hard part is restraint. As I said earlier, leadership usually arrives with a diagnosis already in mind. Some are looking for confirmation. But I treat it as one data point in the read, not the thing that leads it.

Phase 2: Decide on the one constraint

Decide is where most audits fall apart or produce nothing valuable. It’s easy to hand over a long list of findings and let the room sort it out. That’s description, not diagnosis. Of course, the problem can sometimes be the result of a combination of issues. But my clients look to me, as their strategic advisor, for a prescription that moves the needle.

So before I commit, I trace the stall back through five lenses. I call them PULSE: Position, User, Legibility, Stage, and Evidence. Each one is tied to a framework I use and a check I run against the Phase 1 data, and I read them in a deliberate order.

Position. Can the team describe what makes the firm different in one sentence, and does the website say the same thing?

User. Do the people finding the firm match the buyers who actually close, and is the firm present where those buyers look, including the surfaces AI reads?

Stage. Does the content meet buyers where they are on the awareness spectrum, or where the firm wishes they were?

Evidence. Which kind of proof is missing, and where does the doubt set in?

Legibility. When buyers ask AI who to hire, does the firm show up, get cited, and get recommended?

I walk through each lens, and the framework behind it, in the PULSE Diagnostic. It’s simply a mnemonic that highlights the key areas that tend to stall organic growth. Plus, “taking the pulse” of a business’s organic marketing makes sense.

While PULSE is simply an acronym, it’s often not the exact order I follow. I read Legibility last, because visibility amplifies whatever the first four already are. Often a stall touches more than one lens. But one of them is usually holding the others down.

So I commit. Out of everything Investigate surfaced, I name the one constraint I believe is holding the rest down. And I make it specific enough to be wrong. In other words, if there’s more than one, I prioritize and name the one that will move the needle the most in the long term.

Stated differently, deciding is also choosing what not to fix yet. A firm with a fuzzy position doesn’t need 40 new blog posts. It needs a position first, or the 40 posts will be fuzzy too.

Finally, a diagnosis is clear, specific, and measurable. “Your content needs work” isn’t a diagnosis. “You rank for questions your buyers stopped asking two years ago, and you’re missing from the five sources AI cites for your category” is. One of those can be tested. The other can’t.

Phase 3: Execute citation-first fixes

Execute turns the decision into work. The trap here isn’t sloppy work. It’s drift, where the team ships something that looks like the diagnosis from a distance but behaves like the old plan up close.

To prevent that, I write the Decide sentence at the top of every brief. Every piece of work has to trace back to it. The work that can’t gets cut, even when the team is attached to it.

What gets prescribed depends on the constraint. A position problem gets a positioning rebuild before any content. An audience problem gets a reset of who the content is for. A proof problem gets real evidence: client results, original data, named methods.

When the constraint is visibility, the fix is citation-first content. I rebuild key pages with SOME, the four moves I use to make content AI can quote: Summarize, Organize, Modularize, Emphasize. Each page opens with a direct answer a model can lift, headings mirror the questions buyers ask, and the claims carry specifics worth citing.

Then the work moves off-site. The firm’s facts need to match across its site, schema, LinkedIn, and directories, which is the work behind brand memory. And the firm needs to earn mentions on the lists and publications AI already cites. That part is slow, and it’s the part most firms skip.

Phase 4: Analyze and Learn from real data

You tested a hypothesis. Analyze is where you find out if you were right.

The question isn’t whether some numbers moved. It’s whether the specific outcome Decide predicted actually happened. For example, if I said non-branded citations would climb once the category pages were rebuilt, that’s what I measure. I try not to get distracted by other metrics.

So I go back to the same measures from the start, read through the Triple-P layers. Search Console for impressions and clicks on the target questions. Analytics, with a filter that isolates referrals from chatgpt.com, perplexity.ai, gemini.google.com, and claude.ai, for AI traffic and conversions. The same buyer questions re-run across AI engines, more than once, because AI answers change from run to run. And the CRM, because traffic that doesn’t turn into pipeline isn’t the goal.

In my experience, this is the stage most consulting work skips. The report shipped, everyone moved on, and nobody checked whether the diagnosis was right. I’ve seen many situations where clients deployed an action plan but failed to measure whether the plan worked or worked as intended.

Sometimes, knowing when to Analyze is just as important as knowing what to analyze. Early results can look promising. But the problem is, it’s easy to make some premature assumptions, or arrive at the wrong conclusion because something else changed that skewed the results, whether favourably or not.

Just as Decide follows Investigate, Learn follows Analyze. It turns the result into something the firm can apply, and it’s where the next lap starts. What did this teach us that we didn’t know going in? It creates a feedback loop that informs the next steps and improves incrementally.

For the client, that means updating the buyer question list, the benchmarks, and the reporting so the next stall gets caught earlier. For me, it means updating the templates I bring to the next diagnostic, whether it’s with the same client working on retainer or a new one. That’s how the loop compounds.

A lesson that lives in someone’s head fades. A lesson written into the next lap’s checklist sticks.

How I recover from an SEO decline

A plateau can be a slow levelling off or a gradual slide. A sudden drop is different. Still, recovery follows a similar order. I work from the inside out.

• First, the technical floor. Did something break? A migration, a redirect gone wrong, pages accidentally blocked from indexing, or AI crawlers blocked in robots.txt. These are the cheapest fixes and the most embarrassing to miss.
• Second, the pages. Are the pages that used to win still answering the question buyers ask today? Often they answer last year’s question.
• Third, the footprint. Is the firm present and consistent in the places AI and buyers check before they ever reach the site? This is where most declines that look like “SEO” actually live now.

The order matters. Building off-site mentions for pages that can’t be indexed is wasted effort. So is rewriting pages when the real issue is that nobody outside the site vouches for you.

A plateau I diagnosed from the inside

I’ve done this as an operator, not just as an outside advisor. When I joined Musora Media, a digital music education platform, as VP of Growth, I led their organic marketing team made up of SEO analysts and content creators. At the time, the company had hit a growth plateau, and part of my mandate was to find out why.

I ran a 360-degree diagnostic before changing anything. The constraint wasn’t what it looked like from the dashboard. It was gaps in commercial-intent capture and technical SEO that were quietly bleeding opportunity, plus international markets nobody was going after.

So we shifted toward user-first, entity-based SEO and more credentialed content.

A year into my work, traffic grew 244% year over year, search visibility 79%, and organic leads 115%. None of that came from doing more. It came from fixing the right thing first.

What a good diagnosis hands you

An effective diagnosis is not only descriptive but also prescriptive. A diagnosis shouldn’t end with a 40-slide deck. Decks describe what’s happening. They rarely explain why and what to do about it.

When you get a diagnosis, what you should walk away with is short. You get the root cause stated clearly and concisely. You get a prioritized list of fixes, in order, with the first one clear enough to start on Monday. And you get a prediction of what should move, and how you’ll know. Plus, you get a list of what to stop doing, which is often where the quickest savings hide.

If a diagnosis can’t tell you what to stop, it probably didn’t find the constraint.

When to bring in outside help

Internal teams can run a lot of this themselves. But a few signs tell me an outside read is worth it. The team has tried three or four fixes and none of them stuck. Everyone has a different theory about the cause. Or the people closest to the work built the current strategy, which makes it hard for them to see past it.

That’s not a talent problem. It’s one of distance. Familiarity comes with inside knowledge. But it also comes with assumptions, biases, and a certain “marketing myopia” that creates blind spots. A third party brings in a fresh pair of eyes, their experience working in the same situations, and no baggage that can influence their work.

That’s what The Gauge is for. It’s my paid diagnostic, usually four to eight weeks, and every engagement I take on starts there. You can see how it fits with my other work on the services page. The five lenses get their own deep dive in the PULSE Diagnostic.

Flat numbers are a signal. The fix starts with knowing what they’re signalling.

---

Frequently Asked Questions