Before you fix your conversion rate, find out if the number is real.
Most conversion work starts with heatmaps and button colors. Ours starts with a harder question: is the number you're trying to raise measuring anything true? Then we rebuild the journey your buyers actually walk, in a test that runs beside your site without touching it. Three months. A start and a finish. Everything we build stays yours.
By the time you ask about conversion, you've usually been burned.
Owners rarely arrive curious about conversion. They arrive having already paid for the buzzword version: a redesign that changed nothing, a testing tool nobody drove, an agency with a glowing dashboard.
The suspicion you walked in with is earned, so here's what these first conversations actually sound like.
We never optimize a number before proving it's real. Raising an unverified number is how you end up optimizing a lie.
Conversion isn't a button problem. It's a journey problem.
You were right to distrust the buzzword. Most of what sells as conversion rate optimization is tool output: heatmaps, color tests, a calendar of experiments that dutifully raise whatever number was agreed on, whether or not that number matters.
Here's the version we practice. Your buyer walks a path from first doubt to signed deal, and at each step something has to shift in how they see you before the next step feels like the obvious move.
Conversion work is finding where that walk breaks, then rebuilding the experience so each next step feels logical instead of demanded.
You've stood inside the broken version yourself: a door marked pizza that opens into a maze of gates. You wanted the pizza. You left. Your buyers do the same thing, and no heatmap explains why, because the why lives in what they believed when they walked up to the door.
Seeing those shifts takes more than talent, and that's no insult to anyone who's worked on your site.
It takes experience across many businesses and the room to do real research: sitting with your best customers, and with the ones who came right to the edge and didn't buy, until the real reasons surface.
The journey gets rebuilt in five moves.
No mystery, and no drawer of secret tricks. The same sequence runs every time, because the sequence is the skill.
Verify the number.
The analytics audit comes before anything else: every sale or lead counted once and credited to the right source, as far as the data allows. If the number underneath is fiction, everything built on top of it inherits the lie.
Interview the people who already decided.
We sit with your best customers, and with the ones who got close and walked away. Every conversation is recorded, analyzed, and shared with you. On one engagement that meant about 25 interviews over about four weeks.
We're not listening for what they bought. We're listening for what the purchase was supposed to make possible, because that's what they were actually buying.
Build for the buyer closest to the finish line.
Five kinds of strangers land on the same page, and each needs a different first sentence. We build the experience for the readiest one first, then create variations for the others, matched to the ad or the search that brought each of them in.
Run versions, not opinions.
Versions B, C, and D compete against your current page, which stays live as the control. The whole test runs in a slice built beside your site, isolated from it, and the slice is yours.
End every test in a written rule.
Each rule records what we tried, when, and what happened, dated and kept. Wins compound into a playbook that's yours. So do the misses, because a documented miss never gets paid for twice.
Five different strangers walk up to the same door.
In a perfect world, you'd build a separate funnel for each of them. Very few brands can afford to optimize five funnels at once, each specific to its own audience, so most sites build one page for an average visitor who doesn't exist.
We start where the money is closest and work backward.
And the rebuild doesn't have to stop at the page. When the build calls for it, we carry a buyer through all five stages, working with whoever runs your traffic to segment campaigns, ads, and channels by intent, with copy and creative matched to each level of awareness.
One rebuild ran this way for a business-software company whose page had been built by engineers, for engineers, while the people signing the checks were the ones actually reading it.
Rebuilding the journey for each reader did more than lift the page: the same map ended up inside their cold email, their demo agendas, and their sales calls, so every channel finally told one story. The demos started booking, and the demos turned into deals.
Three months. A start, a finish, and no waiting until the end to know.
Conversion work here is a project, not a subscription that renews on hope. The runway is three months, typically around four tests, and the pace is set so you're never a quarter away from learning something.
The quick wins
Every business has a few leaks that need a slight adjustment, not a rebuild. We find them, rank them, and fix them first: measurable movement at no added cost while the real tests are being built.
Read in weeks, not quarters
A test that wins becomes a written rule and the next version builds on it. A test that loses becomes a written rule too, and it sharpens the next version.
A test that sits flat pulls our meeting forward, because flat means the conversation happens now, not at the next check-in.
The handoff
The plan gets reprioritized the whole way based on what the tests reveal.
By the end, your team has the written rules, the research, and the training to run the loop without us. None of that depends on our tools.
Our hands, your keys.
We do the building: the research, the versions, the edits, the measurement. All of it happens in the test slice beside your site, so your week doesn't change, and your live pages are never the experiment.
By the time the project starts, the diagnosis has already taught us your positioning, what you sell, and your buyers, which means we're not learning on your invoice.
Your half is small but real. If your calls are recorded and your pipeline is logged honestly, we need very little of your time.
The one thing we keep asking for is the thing no system captures: what your sales team is sensing. Are the leads asking different questions? Pushing back in new ways? Feeling stronger or weaker than last month? Patterns hide in those answers, and your closers hear them first.
And on day one we agree on one rule that saves projects: when we're waiting on you and the work needs to move, do we push forward on our best judgment, or stop and wait?
Either answer works. Deciding it up front is what keeps three months meaning three months. Anything significant still gets a call before we act, whichever rule you chose.
The other ways to buy this, no spin.
You have four other doors. Depending on your situation, one of them might be the honest winner, so open each one before you decide.
A conversion-only agency They'll hit the number you agreed on. That's the danger. Optimizing one metric, blind to what it costs the rest of the business.
A specialist shop optimizes the numbers in the contract without owning the repercussions outside it.
We once took over an account where the previous conversion agency had done exactly what it promised: small, cheap orders through the roof.
What nobody connected was that those orders never grew into the large ones the business lived on, and all that traffic had taught the ad platform's tracking profile, the pixel, to hunt for exactly the wrong buyer. The pixel was so badly trained it had to be scrapped and rebuilt from zero.
The agency called it a win the whole way down, because on the agreed number, it was.
A testing tool, or AI Anyone can run a test. The sequence is the skill. The tools are fine. What they lack is the context around the test.
A competent owner can absolutely run these tests. The break happens in the handoff. Delegate the testing to someone with less experience, and the context doesn't survive the relay: what to test first, what to leave alone, how a change on the page ripples into the follow-up emails, the checkout, the sales call.
Even an owner briefing an AI tool rarely gives it the full picture twice, and without a proper system underneath, the tool fills the gaps with assumptions that sound general and safe. The results still feel like wins. They just don't build the best path to growth, or knowledge that compounds.
Here's the tell: we've deliberately lowered a conversion rate, adding friction so the wrong buyers select themselves out and the end result gets better. No tool would recommend that, and it was the right call.
Your web person or designer Talented and loyal, but missing the latitude to work at that level. The gap isn't ability. It's experience and room to operate.
In over 25 years, we've rarely seen a designer given what this work actually requires. The first half is experience across enough businesses to recognize the patterns.
The other half is latitude: the time, the budget, and the mandate to sit with real buyers until the shifts make sense at the depth the rebuild needs.
So your web person does the honest thing with what they have: they execute your self-diagnosis to the best of their ability, to your standards, aiming to satisfy you.
The page gets better-looking. The path stays broken. And everyone did their job.
Another redesign You already know how this one ends. One client had burned through five or six before we met.
A redesign is the safe purchase: everyone has bought one, and nobody gets fired for proposing another. It's also the most expensive way to change nothing.
The business-software company you met above came to us after five or six redesigns and rebuilds, the full nine yards, and the demos still weren't booking, because a redesign repaints the door without changing what's behind it.
When we rebuilt a custom home builder's page around the path its best clients already took before buying, the design was almost beside the point: conversions rose more than fourfold. If your last redesign changed nothing, the design was never the disease.
When conversion work is the wrong prescription.
Sometimes the page is innocent. If your page suddenly got slow, you don't need conversion work, you need a faster server, and we've told owners exactly that: fix the hosting, come back in a month or so, we'll look again, no invoice.
If your page converts under about 1%, the mismatch usually lives upstream: the targeting is off, the thing you sell has lost its edge, or the ad platform's profile of your buyer was trained on the wrong data.
One clothing brand ran a $10 promotion on a $50-and-up product line to move old inventory. It felt like a winning campaign because customers came.
But those customers were price shoppers who almost never bought again, and repeating that promotion taught the platform to go find more people just like them. No page rebuild fixes traffic aimed at the wrong buyer.
And sometimes the honest answer is to stop pulling the lever entirely.
A brand that sold everything through Amazon asked us to make its own website sell. We rebuilt the pages and the funnel, and they worked, but not well enough to beat what a sale on Amazon already cost them.
So we stopped selling them conversion work, even though that's what the engagement started with, and shifted the effort to reaching their Amazon customers directly, done carefully and by the platform's rules. When the data says conversion isn't the lever, we put the lever down.
The last gate is arithmetic. The lift has to be worth buying:
A $5,000 product selling once a week. A real lift pays for the work quickly.
Worth itA $50 product selling at volume. Small gains multiply across every order.
Worth itA $5 product selling a couple of times a day. The math doesn't clear, and we'll say so.
Not yetWhen the real disease turns out to be traffic, the product, positioning, or measurement, the fix doesn't require a new vendor search.
The diagnosis and every one of those treatments live under this roof, so shifting the effort takes a conversation, not a new round of interviews and onboarding.
Three rebuilds, told in full where they live.
De-identified, like everything we publish. Each number below is carried by its full record, so you can read the whole story and judge the frame yourself.
The website needed a redesign. It had been redesigned before.
Their best clients all took the same path before buying, and the homepage was hiding it.
The page was rebuilt around that path. Website conversion rose 442%, and every team finally shared one map of how a buyer is made.
Paid search was too expensive. A single click ran $100 to $280.
The expensive traffic they already paid for was landing on a page that let it leave.
The page was rebuilt, and the scoreboard moved to signed cases. Conversion rose 1,175%.
The dashboard said 2% of visitors booked. The phone stayed quiet anyway.
Page refreshes and phone-number clicks were being counted as appointments. The 2% was fiction.
Honest tracking first, then the rebuild: a real 6.25%, 247 booked appointments, about $80K in first-visit revenue.
Results from specific engagements, measured by us and told with their context intact. They are records, not promises of what your account will do.
This project is prescribed, not sold.
Conversion work is a treatment, and we don't prescribe treatments before the examination. Everything lives on a document, in this order.
Conversion Rate Optimization
The journey rebuilt and tested beside your live site. Your pages untouched. Everything we build, yours.
Where every engagement starts. A senior diagnosis on your real numbers that confirms where the biggest opportunity actually is, conversion or not. If the pages turn out to be innocent, you leave knowing what the real disease is, and the findings are yours either way: act on them with us, with your own team, or with nobody.
Only if the diagnosis prescribes it. Typically around four tests against your current page as the control. The first month carries the heavy build, $12,500 to $15,000; months two and three run $5,000 to $7,500 each.
A simple lead form sits at the low end. A store with upsell flows and configuration wizards sits higher. The exact number is set by the prescription, in writing, before anything starts.
Already running other work with us? Conversion folds into the monthly engagement instead of a separate project.
- The test slice and everything in it is yours, whatever happens.
- Every test's written rule stays with you: what ran, when, what happened.
- The research is yours: recordings, analysis, and what your buyers actually said.
- Your team leaves trained: how to test, measure, and roll out without us.
The losing test What happens when a test loses? It will happen. Here's why that isn't the same as losing money.
Some tests lose. A losing test is capped by design: it ran in a slice beside your site, against a control that kept converting the whole time, and your live pages never carried the risk.
What a loss buys is information, and it gets banked the same way a win does: a written rule, dated, with the results attached, so the next version starts smarter and the same mistake never gets funded twice.
What you'll never get from us is a month of defending a plan the numbers have already called. Flat or falling numbers pull the meeting forward: you see what we see, and the plan changes while there's still runway to use what we learned.
The questions owners in your position actually ask.
What is conversion rate optimization?
Conversion rate optimization is the work of turning more of your existing visitors into customers, without buying more traffic. The shallow version tests button colors and layouts. The real version studies the journey your buyers take, finds where it breaks, verifies that your conversion numbers are measuring something true, and rebuilds the experience so the next step feels obvious to the person taking it.
What does conversion rate optimization cost?
With us, it starts with a $2,500 diagnosis that decides whether conversion work is even the right treatment. If it is, the three-month project runs $12,500 to $15,000 for the heavy first month, then $5,000 to $7,500 for each of the two months after. Where you land in those ranges depends on your setup: a simple lead form costs less to rebuild and test than a store with upsell flows and product configuration steps.
We already tried CRO and it didn't work. Why would this be different?
Ask what the last attempt actually did. If it tested layouts and colors against a number nobody verified, it was optimizing blind, and no amount of testing fixes that.
This process starts by proving the number is real, then does the research most conversion work skips entirely: interviews with your actual buyers and the ones who walked away. The tests come after the understanding, not instead of it.
Do I need a new website instead?
Usually not, and you should be suspicious of anyone whose answer is always yes. The tests run in a slice built beside your site, so nothing gets torn down to find out what works. If the diagnosis genuinely shows the site itself is the disease, you'll hear that before you spend a dollar on testing, with the reasoning on paper.
How long before we see results?
The first movement usually comes from the quick wins in the opening weeks: small fixes that need an adjustment, not a rebuild. Each test after that should show leading signals within two to three weeks, with an outcome inside a month. The whole project is paced so you're never waiting a quarter to learn whether something worked.
Would you ever lower my conversion rate on purpose?
Yes, and we have. When the sales team is drowning in wrong-fit leads, adding friction, a qualifying question or a step that makes the buyer earn the conversation, drops the rate and raises the revenue. The conversion rate is one instrument on the panel, not the prize, and flying by a single instrument is how funnels end up looking like spam.
My dashboard says our conversion is fine. Why would I touch it?
First question: is the number real? We've seen dashboards count page refreshes as bookings and credit sales to the wrong source entirely. Second question: fine at which step? A healthy-looking rate on leads can hide an expensive collapse between the lead and the signed deal, especially when your sale happens on a call. The audit answers both before anyone optimizes anything.
What if my traffic is the real problem?
Then conversion work is the wrong prescription, and the diagnosis exists to catch that before you buy it. A page converting under about 1% usually points upstream: targeting, an ad profile trained on the wrong buyers, or a product that lost its edge. Because the other treatments live under this roof, shifting the effort is a conversation, not a new vendor search.
Is my business too small for this?
The gate is arithmetic, not headcount. What does a real lift earn you in a year, and does that comfortably beat the cost of the work? A $5,000 product selling weekly clears it easily. A $5 product selling a couple of times a day doesn't, and we'll tell you so and tell you what to fix first.
The math is basic, we run it inside the diagnosis, and it's honest in both directions.
The lift is the receipt. The habit is the asset.
A year from now, the rebuilt page will just be your page. What lasts is what your team learned while it was built: test before you argue, measure what you roll out, trust the number because you checked it yourself.
When the project ends, it ends on a high note, with you equipped to keep going. That's the plan from day one.
One question decides everything on this page, and it's the first one the Second Opinion settles: is your number real? The diagnosis is yours to keep either way, and if conversion isn't your real problem, you'll hear it from us first.