Someone will say, almost verbatim, some version of this: "I think I need a new website — or maybe just better SEO — I'm not totally sure which." It comes up constantly, and it's usually said with a kind of apology in it, like they should already know the answer and don't.
Here's the thing I've stopped doing in response, even though it would be easier and would probably close the conversation faster: I don't answer that question. Not because I don't have an opinion — I always have one within the first ten minutes of looking at the actual site — but because the question, as asked, is almost never the real one. It's a symptom wearing the costume of a diagnosis. And the fastest way to waste someone's budget is to treat a symptom as if it were the disease.
This piece is an attempt to actually show the reasoning underneath that, not hand over a framework. The honest version of how I get from "I'm not sure which" to "here's what's actually happening and here's what we're doing about it" — including the parts where I'm genuinely uncertain, and the parts where I've been wrong before the data corrected me.
The trap of fixing the thing you can see
Websites are visible. You can look at one, wince at it, and feel like you've diagnosed something. SEO, by contrast, is mostly invisible — rankings, impressions, click-through rates buried in a dashboard nobody checks. That asymmetry does something predictable to how business owners frame their own problem: the visible thing gets blamed, because it's the thing sitting right in front of them looking dated.
That instinct isn't stupid. A site that looks like it was built in 2014 probably is contributing to the problem. But "contributing to" and "causing" are different claims, and conflating them is how a business spends real money on a redesign that doesn't move a single number that actually mattered, because the leak was somewhere else the whole time.
The pattern I see most often: traffic is fine, sometimes even good. The site gets found. People land on it. And then nothing happens — no calls, no form fills, or the calls happen but never get answered, or they get answered badly. None of that is a website problem in the design sense, and none of it is an SEO problem in the ranking sense. It's a gap between two systems that were each built separately and never actually tested together as one path from stranger to client.
That's the actual reason I refuse to answer the "website or SEO" question on day one. Answering it early means picking a side of a false binary before finding out whether the real problem is even inside that binary at all.
What actually gets studied first
Before I ask the client anything, I study the thing itself — the live site, the search presence, whatever analytics exist. This isn't a courtesy. It's because a client's own description of their problem is filtered through whatever they already believe is wrong, and that belief is exactly the thing I'm trying not to inherit uncritically.
Two checks happen almost immediately, before anything else, because they tend to reveal more than an hour of conversation would. First: mobile. I open the site on a phone, not a desktop, because that's how most real visitors are actually arriving — a pattern that shows up clearly in the local search data. BrightLocal-sourced research cited by biziq found that the average mobile page loads in roughly 8.6 seconds, well past Google's own stated threshold of under three seconds for what counts as acceptable. If a site is slow or broken on mobile, everything downstream of that — rankings, ad spend, SEO content — is being fed into a leaking bucket, and no amount of additional traffic fixes a leaking bucket.
Second: the hero, the first section a visitor sees. If I can't tell what the business does and why I should care within five to ten seconds of landing, that's a finding on its own, independent of whatever the client thinks the problem is. A slow site and a confusing hero are both website problems in the literal sense, but they're not the same website problem as "this looks dated," which is usually what a client actually means when they ask for a redesign. One is cosmetic. The other is structural. They get treated completely differently, and confusing them is the single most common mistake I see other people make in this exact conversation.
Alongside that, I look at what's already ranking, what isn't, and whether there's any existing measurement — analytics, call tracking, a CRM — that could actually show where visitors are dropping off, rather than guessing. This is the part that determines how much of the next phase is diagnosis versus how much of it is building the ability to diagnose, because a shocking number of small businesses have neither.
Asking, not telling
Once the study is done, I have a working theory. I don't lead with it. This isn't a performance of humility — it's a specific, deliberate choice about how the actual problem gets surfaced, and it comes from a simple observation: a client who's told what's wrong with their business gets defensive, even when the diagnosis is accurate. A client who's asked the right questions and arrives at the same conclusion themselves owns it, and owning it is what makes them actually willing to fund the fix instead of negotiating it down.
So the conversation becomes questions. What's the actual goal — more calls, more form fills, more foot traffic, something else entirely? What's already been tried, and what happened when it was tried? What's the real timeline, and is that timeline coming from a genuine deadline or from impatience? Where do most inquiries currently come from, when they do come — and does anyone track that, or is it a guess?
That last question does more diagnostic work than almost anything else, because the honest answer is very often "I don't actually know." That answer isn't a failure on the client's part. It's a finding. It means part of the actual project, before a single design decision gets made, is building the measurement that would let anyone answer that question with data instead of a guess. Fixing a system nobody can see clearly is how a lot of money gets spent solving the wrong problem convincingly.
Breaking the funnel into pieces that can each be blamed individually
Here's the move that actually does the work, and it's the same move that shows up in basically everything I do, not just this: instead of treating "growth is stalled" as one big, unsolvable problem, it gets broken into smaller pieces, each of which can be checked independently.
Is there a traffic problem? Is the site actually being found for the searches that matter, or is it invisible? This is checkable directly — search console data, ranking tools, straightforward comparison against competitors.
Is there a first-impression problem? Do people land on the page and leave immediately, measured by bounce rate and time on page? This is a website-design question in the literal sense — pacing, clarity, the hero test above — and it's separable from traffic entirely. A page can rank perfectly and still fail this check.
Is there a conversion-mechanism problem? Of the people who stay, how many actually take the action the page asks for? This is where industry context matters more than most people assume. Ruler Analytics' 2026 conversion benchmark report, tracking more than five million conversions across thirteen industries, found that legal, automotive, and software all convert above 7.5 percent, while travel, retail, and health and social care sit under 3 percent — and the report's own read on why is that industries with higher-value, more considered purchases tend to have lower measured conversion rates, not because the marketing is failing, but because the buyer's journey involves more touchpoints before anything gets formally captured as a conversion. That distinction matters enormously here: a 2.5 percent conversion rate might be a real problem for one business and a perfectly normal number for another, and treating them the same is how a client gets sold a redesign they didn't actually need.
Is there a fulfillment problem? This is the piece almost nobody checks, and it's often where the actual leak is hiding.
The leak that's almost never where anyone looks
This is the part of the funnel that lives entirely outside the website and entirely outside SEO, which is exactly why it gets ignored by anyone whose service happens to be websites or SEO — and it's often the actual answer.
A visitor does everything right. They find the business, they land on a page that makes sense, they decide to reach out, and they pick up the phone — which, for local and service businesses, is still the dominant channel by a wide margin. Research summarized by Numa, drawing on BrightLocal's survey data, found that 60 percent of consumers who looked a business up on Google chose to call rather than use any other contact method. That's the majority outcome, not an edge case — which means a website that treats the contact form as the primary conversion point, while the phone number sits small in the corner, is optimizing for the minority path.
And then the call goes unanswered. A study cited by Phonely, tracking 85 businesses across 58 industries over 30 days, found those businesses answered only 37.8 percent of their incoming calls. That number is worth sitting with. It means that even in a world where the website and the search visibility are both doing their job perfectly, nearly two out of three of the calls they successfully generate never get picked up at all.
This is the piece of the puzzle a pure design-or-SEO framing simply can't see, because it's neither. It's an operations problem wearing a marketing costume. And it explains something that used to genuinely confuse me early on: a client with a decent site and reasonable rankings, insisting something's broken, and being completely right — just not about the thing they thought was broken.
There's a related wrinkle worth naming honestly, because YouGov's research on channel preference complicates the obvious fix. Nearly 70 percent of Americans use phone support, but only 35 percent actually prefer it — and website contact forms show the same gap in miniature, used by 34 percent but preferred by only a small fraction of that. People default to whatever channel is available, not whatever channel they'd choose if every option worked equally well. That means the fix isn't automatically "make the phone number bigger." Sometimes it's building a faster, better form specifically because a form doesn't require anyone to be standing by a phone at the exact moment it rings — and knowing which of those two is the actual leverage point requires the funnel breakdown above, not a guess.
When two fixes would both work: how I actually choose
Sometimes the diagnosis comes back genuinely ambiguous — not because the process failed, but because two different fixes would both move the number that matters. A faster site would help. So would a call-answering system. So would a clearer hero. When that happens, I don't default to whichever is fastest to ship, and I don't default to whichever is the usual move for this kind of problem. Fast and usual both optimize for looking productive this week, and neither one is actually the tie-breaker I care about.
The tie-breaker is the long-run effect of each option. A quick fix to the hero copy might lift the immediate conversion number, but if the underlying structure of the page still buries the actual proof of quality three scrolls down, that fix has a ceiling, and it's a low one. A slower fix — restructuring the page's actual argument, in the order information gets presented — costs more time up front and keeps paying off after the project is technically finished, because it isn't dependent on one clever line of copy staying clever forever.
This is also where I'm genuinely uncertain more often than I'd like to admit out loud. Long-run effect is a judgment call, not a number I can pull from a dashboard, and I've been wrong about which fix would compound and which wouldn't. The honest mitigation for that isn't confidence — it's building in a checkpoint early enough that being wrong is cheap to correct, rather than discovering it eight months in.
Custom build or existing tool: the real deciding factor
A related decision shows up constantly inside this same diagnosis: once the actual leak is identified, do you build something custom to fix it, or do you bolt on an existing tool — a call-tracking service, a form plugin, a booking widget?
Cost gets discussed first in almost every conversation about this, and it's the wrong lens. The actual deciding factor is flexibility — specifically, whether the workflow can still be adjusted six months from now without the tool itself becoming the reason it can't be. An existing tool is often the right call precisely because it's fast and proven. But some existing tools lock a business into a rigid structure that made sense for whoever built the tool's average customer, and not necessarily for this specific business's actual workflow. When that's the case, the tool becomes the bottleneck it was supposed to remove.
That's the actual principle underneath this, stated plainly: the system should never be the bottleneck. Time is the real constraint in almost every business — not process, not tooling — and the entire point of a good workflow or a well-chosen tool is to protect time, not to become one more thing that has to be worked around later. A custom piece built with room to adjust is sometimes worth the extra upfront cost specifically because it removes a future bottleneck before it exists. A rigid off-the-shelf tool is sometimes the better call specifically because it's already solved, tested, and not worth reinventing. There's no fixed rule that resolves this in advance — it comes back to studying the specific workflow and asking where it's actually likely to need to bend later.
Where the honest uncertainty actually sits
It would be tidy to end this with a clean framework — a flowchart that always produces the right answer. That's not honest, so it's worth naming where the real uncertainty lives instead.
The biggest one: without existing measurement — analytics, call tracking, some record of where inquiries actually come from — a lot of this diagnosis is educated inference, not proof. I can tell a slow mobile load is a problem in general, because the data on load time and bounce behavior is well established. I can't always tell, without real numbers from that specific business, exactly how much of their specific stall is design versus visibility versus fulfillment versus something outside either — a sales process that loses warm leads, pricing that doesn't match the market, a step in between the call and the close that never gets looked at at all. Sometimes the honest answer, after all the questions and all the study, is that the real fix is outside what a website or search visibility can touch, and saying that plainly is worth more than pretending otherwise.
The second uncertainty is about timing. The long-run-effect tie-breaker described above assumes there's time for the long run to actually play out, and sometimes there isn't — a business with three months of runway needs the fastest real lever pulled now, even if it's not the most durable one. Judging when "durable" has to yield to "urgent" is a real trade-off, not a solved problem, and I'd rather say that directly than pretend every decision resolves cleanly to the theoretically best answer.
What actually changes my mind mid-project
Diagnosis isn't a single event at the start that then gets executed on faith for months. The plan built after that first round of questions is a hypothesis, and hypotheses get tested against what actually happens once changes go live.
If a redesigned hero ships and the bounce rate barely moves, that's information, not a failure to be defended around — it means the actual friction was somewhere else, and the next round of questions gets aimed there instead. If call-answering gets fixed and the phone still doesn't ring more, that's information too — it points back toward visibility or the page itself. This is the same decomposition instinct from the start of the project, just running continuously instead of once: break the thing into pieces small enough that each piece can prove itself right or wrong, and let the ones that fail to prove themselves get replaced rather than defended.
That continuous checking is also, honestly, the part of this work I find genuinely enjoyable rather than tedious — re-testing an assumption, finding out it was half-right, adjusting it, and doing that again. Nothing about a site or a search strategy is ever actually finished in a permanent sense. There's always a next question worth asking about whether the current setup is still the right one.
Where this actually leaves you
If you're the business owner asking "do I need a new website or better SEO," the honest answer is: probably neither is the full answer, and definitely not yet. The question worth asking first isn't which fix to buy — it's where, specifically, in the path from a stranger finding you to that stranger becoming a client, the thing that should be happening isn't happening. That path is checkable, piece by piece, with real evidence at each step, rather than guessed at from whichever part of the business happens to be visible and easy to blame.
Sometimes that process does end with a new website. Sometimes it ends with better visibility. More often than either camp would like to admit, it ends somewhere neither a designer nor an SEO consultant was set up to look — a phone that goes unanswered, a form that nobody follows up on within the hour, a measurement gap that meant nobody could actually see the leak until someone went looking for it directly. Finding out which one it actually is, rather than assuming, is most of the real work. Everything after that is just building the fix.