Web design
A new website should not cost you the rankings the old one earned.
Most practice rebuilds start the same way. Somebody says the site looks dated. Eight weeks later it looks current and earns less than it did.
We have watched a migration take a year of search traffic off a site, in an account we run today. The numbers are further down this page, including the part that looks bad.
So we build in a fixed order. What the current site earns gets inventoried before anybody opens a design file. Design comes after that, and it gets a real budget, because a practice site also has to make a stranger trust you in about four seconds.
What web design means once a site already ranks.
It is two jobs sharing one budget: build a site a patient can actually use, and carry across to it whatever search equity the old one had earned.
The second job is invisible until it fails, because nobody ever sees a redirect working and everybody sees the homepage. So in the accounts we run, the redirect work is the first thing compressed when a timeline slips. The bill for that arrives about six weeks after launch, as a ranking chart nobody wants to open.
A design studio will hand you a better-looking site. That is a real skill and we are not dismissing it. But ask who on that team owns the question of whether the site still ranks in November, and the answer is usually nobody.
This page sits downstream of the rest of what we do. Organic search compounds on the pages a site actually has, so the rebuild decides its ceiling. Ranking in the map needs a page per location that a patient could use, not an address block. Getting quoted inside an assistant’s answer depends on page structure more than on anything else. And knowing what a patient cost depends on tracking that survives the launch, which is another thing rebuilds routinely break. The services overview shows how the pieces sit together.
What we do
What we build, and the order we build it in.
The inventory comes before the wireframe
Every URL on the current site that earns something: a ranking, a link, traffic, a lead. We pull that list before anybody opens a design file, because it decides which pages the new site has to keep, which ones merge, and which can go. Run it the other way round and the design settles the architecture. That is how a rebuild quietly drops the twelve pages that were doing the work.
Every page gets one of four decisions
Remove it, improve it, consolidate it into a stronger page, or keep it as it is. Before founding Dweeso, Michael built that audit in-house and it is still how we do this. One decision per page, written down, with a reason, and the page count usually falls. A practice site carrying two hundred pages rarely has two hundred pages of patient demand sitting behind it.
Page types before templates
A location page is a different object from a procedure page, and both are different from a post. Each needs its own structure, its own internal links and its own conversion path. Design the templates first and you get a site that looks consistent and answers nothing, so we settle the page types and then draw them.
The booking step is part of the page, not bolted to the end of it
This is where most practice sites lose the patient. The form asks eleven questions when four would do, and the number is not tappable on the device most people are holding. The scheduler loads in its own window, slowly, and half the people who get that far give up before it finishes. We treat the path from the page to the appointment as part of the build, and we test it on a phone before launch rather than after.
Redirects written, launched, then watched
Old URL to new URL, one line each, tested on a staging build before anything goes live. Then somebody checks Search Console daily for the first month. Search will tell you what broke within days of a launch. It only helps if a person is reading it, and the launch week is the week everyone is tired.
The order is the opinion. Most rebuilds run it backwards: design first, content poured in, redirects on the Friday. Every step above exists because we saw that sequence cost a practice most of a year.
What a rebuild cost one network, and what carried it through.
We run search for a 17-location behavioral health network. No client gets named on this site, which is a standing rule and not a courtesy in this one instance. The figures come from a report we prepared in May 2026.
Start with the damage, because leaving it until the end would be a trick, and a page selling caution about rebuilds has no standing to hide its own worst number.
−57%
That migration happened, the traffic reset, and it has not come back. Branded paid search is part of it as well, capturing terms the network already held first position for organically. Either way, the site move is the larger half of the answer.
This is the condition we were working inside, not the result of our work. It is also the single best argument on this page. A rebuild is capable of doing that to a seventeen-location business, and nobody sold it as a risk beforehand.
And the half that recovered first
- Organic conversion rate, May 2026
- 3.43%
- Change year over year
- Up about 97.5%
- Change on the month before
- Up 11.4%
Conversion nearly doubled while the traffic was resetting. When we measured the same site a year apart, fewer visits were turning into inquiries at roughly twice the rate.
The reason is the part that belongs on a web design page. What survived the rebuild in usable shape were the facility pages, and those pages were 8% of organic visits and half of the leads. The blog carried the volume. The page type attached to a place carried the business. The page-type argument is set out in full on its own page.
Both halves belong in the same sentence and we keep them there. Conversion nearly doubled while traffic reset. Quote the first without the second and you have an agency case study. Quote the second without the first and you have a horror story. The account is neither.
A rebuild that went the other way
Before founding Dweeso, Michael led the rebuild of a medical-ID nonprofit’s website. It had fallen from roughly 100,000 monthly visits to 5,000, carrying more than 1,300 pages of thin content.
Every page got one of four decisions. Remove, improve, consolidate, keep, and nothing else. First-position keywords went from 59 to about 1,200. Organic traffic doubled inside the first month and settled near 40,000. Membership and jewelry sales came back to where they had been before the decline.
Three things to be plain about. That work was in-house, it was before the agency existed, and a nonprofit is not a medical practice. We are not offering it as a Dweeso client result. It is where the method in step two above comes from, and the sites that need it most are the ones carrying a decade of pages nobody has read since.
The client accounts this page draws on are told whole on the results page.
Who it's for
Practices whose website is the constraint.
- Med spas
The site and the booking software are usually two products from two vendors, and the appointment gets lost in the gap between them.
- Cosmetic dentists
Two hundred pages, nine of them about veneers. This is the rebuild that removes more than it adds.
- Behavioral health
A page per facility, or seventeen locations competing in search as though they were one address.
- Dermatology
Medical and cosmetic are two audiences arriving at one homepage. Most of these sites speak clearly to neither.
- Urgent care
Hours, wait, address, insurance. The patient decides on a phone, standing up, in well under a minute.
- Plastic surgeons
Procedure pages that answer the price question, and a bilingual structure where the patient is searching from another country.
Six different structural problems, and one of them is probably yours. The diagnosis in the free analysis says which, and whether the fix is a rebuild at all.
The boundary
What a redesign doesn’t do.
- It does not create demand
- A new site converts the people who already find you. It does not add searchers. If nobody in your area is searching for what you do, a better website turns a small number into a slightly larger small number, and the money would have been better spent finding out why the searches are not there.
- It does not guarantee the traffic holds
- A redesign is the most common way a practice loses its rankings, and that stays true when a careful agency runs it. We have never seen a migration cost nothing at all. Planning changes how deep the dip goes and how long it lasts, which is worth paying for, and it is a different promise from the one most redesign pitches make.
- It does not fix a review problem or a phone nobody answers
- The site can do its whole job and the patient still calls, waits on hold, and books somewhere else. We saw that in an account we run, sitting in the call recordings the whole time. Fixing it is a front-desk decision rather than a design one, and no amount of build budget substitutes for it.
- Looking good is not why a page ranks
- A site can be beautiful and invisible at the same time. Design earns the trust that turns a visit into a call, but it does not earn the visit. Any agency selling a rebuild as an SEO strategy in itself is selling you the container and calling it the contents.
- We do not build portals or touch patient records
- No patient portals, no records systems, no clinical tools, no intake software. This is a marketing site. Where it has to hand off to something that holds patient data, we hand off and stop at the boundary, and your vendor for that system stays your vendor.
We measure this work on two questions, in this order. Does the new site convert better than the old one? And did it keep what the old one had? An agency reporting only the first is reporting half the job, and it is the half that is easier to make look good.
Questions
What practices ask before they rebuild.
- Will we lose traffic when the new site launches?
- Expect a dip. Even a clean migration moves URLs, and search takes weeks to settle again. In the network we run, organic clicks were down about 57% year over year after a migration, which is the bad end of that range. Planning shrinks the dip but does not remove it, and anyone promising otherwise has not sat through one and measured it.
- Can you fix our current site instead of rebuilding it?
- Often, and we will tell you when that is the answer. If the structure underneath is sound and the problem is thin pages, slow loading or a broken booking path, then that is repair work and it costs a fraction of a rebuild. A rebuild earns its price when the page inventory itself is wrong. We say which of the two you have before quoting either.
- What happens to our old blog posts?
- Each one gets a decision rather than a bulk import. Posts that rank for nothing and earn nothing come off, and their URLs redirect to the closest live page. Posts with links or rankings behind them are kept or merged into something stronger. In most rebuilds we run the page count goes down and the traffic does not.
- Who writes the copy?
- We do, for the marketing pages. Clinical copy is not ours to write, so it goes to your clinicians for review, and we will not put medical advice under your name without it. If you already have a writer, we give them the page types, the structure and the search intent for each page, then review what comes back before it publishes.
- How will we know the new site is better than the old one?
- Conversion rate by page type, same channel, before against after. Alongside that, rankings and organic clicks watched weekly through the launch window and the two months after it. We agree both measures before the build starts, because it is easy to pick a flattering one afterwards. If the new site converts worse than the old one, the report says so.
Send us the site and the Search Console access.
You get back which URLs earn something today, which of your pages are competing with each other, where the booking path loses people, and whether what you need is a rebuild or a repair. In writing, and it commits you to nothing.
If the question is whether the spend is worth it at all, the ROI calculator runs the arithmetic first. A patient target goes in, the spend it implies comes out.