Quick answer: our own website has run on three platforms. WordPress through Oct 6, 2025. Webflow from Oct 7, 2025 to Jun 7, 2026. Hydra OS since Jun 8, 2026. Measured over equal 46-day windows, sessions went 11,369 to 21,156 to 38,808. Organic search sessions went 2,117 to 5,110 to 16,245. Tracked leads went 88 to 93 to 203. Every chart below names its source and its date window, and the honest caveats are at the bottom.
We sell websites. That makes it fair to ask what our own site does. So this post is not about a client. It is about ciwebgroup.com, and it includes the parts that do not flatter us.
One note on the charts. Every one of them is inline SVG or plain CSS. There is no charting library and no client JavaScript on this page. An article about page weight should not be heavy.
How we measured this
Three platforms means three eras, and eras of different lengths cannot be compared directly. So we cut a 46-day window out of each one:
- WordPress: Aug 22 to Oct 6, 2025. The last 46 days before the Webflow cutover.
- Webflow: Apr 23 to Jun 7, 2026. The last 46 days before the Hydra OS cutover.
- Hydra OS: Jun 8 to Jul 23, 2026. The first 46 days after cutover.
Traffic figures come from GA4 property 308087083. Lead figures come from our WhatConverts profile. Search figures come from Search Console. Lighthouse figures come from lab runs we committed to our own repository. Every GA4 figure in this post was pulled twice, once for the original analysis and once again on Jul 24, 2026 before publishing, and the two pulls matched.
The cutover dates are not guesses. Archived copies of the site show a WordPress 6.8.2 and Elementor generator tag on Sep 11, 2025, a Webflow generator tag on Oct 7, 2025, and still Webflow on Mar 20, 2026. The Hydra OS date comes from our production release #105 on Jun 8, 2026, corroborated by GA4 landing paths shifting from /contact-us and /about-us to /contact and /about that month.
Here is the whole story in one picture before we break it into pieces.
Source: GA4 property 308087083, dimension yearMonth, pulled 2026-07-24. Jul 2026 is a partial month (Jul 1 to Jul 23) and is lower than a full month would be. The Sep 2025 jump happens on WordPress, about four weeks before the Webflow cutover, so it is not a platform effect.
Look at the September 2025 jump. Sessions roughly doubled, and it happened while the site was still on WordPress, about four weeks before we moved to Webflow. Our first pass at this analysis assumed that jump was a relaunch. The archive evidence says it was not. We are leaving it visible rather than quietly re-dating the migration to make the chart tell a better story.
Era one: WordPress
In its last 46 days, WordPress delivered 11,369 sessions and 7,621 users. That is about 247 sessions a day.
The number that mattered most was organic. Only 2,117 of those sessions came from organic search, 19 percent of traffic. Direct traffic carried the site at 6,680 sessions. In Search Console, our top 1,000 queries returned 609 clicks against 38,785 impressions, a 1.6 percent click-through rate at an average position of 37.1.
Position 37 is page four. We were being shown to people and almost never chosen.
Leads were 88 over the window, about 1.9 a day. Of those, 85 were phone calls and zero were web forms. Our website was not a lead form. It was a phone number with pages around it.
Era two: Webflow
Moving to Webflow roughly doubled the traffic. In the last 46 days on Webflow we saw 21,156 sessions and 14,809 users, about 460 sessions a day. Organic more than doubled to 5,110 sessions.
Source: GA4 property 308087083. Window totals 11,369 / 21,156 / 38,808 sessions divided by 46 days.
Search quality improved sharply. Head-query clicks went from 609 to 985 while impressions actually fell from 38,785 to 23,122. Fewer impressions and more clicks is usually a good sign. It means you stopped ranking badly for things nobody wanted and started ranking properly for things people did.
Source: Search Console, top 1,000 queries summed, not property-wide totals. Hydra window runs through 2026-07-21.
Source: Search Console head queries. WordPress 609 clicks on 38,785 impressions. Webflow 985 on 23,122. Hydra OS 731 on 23,855.
The first 90 days after the Webflow cutover, compared against the last 90 days of WordPress, showed sessions up 50 percent (16,779 to 25,180), pageviews up 73 percent (52,656 to 91,365), and leads up 95 percent (150 to 292). Across the full 244-day Webflow era we recorded 88,835 sessions, an average of about 364 sessions a day.
Here is the part we do not skip. In the matched 46-day windows, leads went from 88 on WordPress to 93 on Webflow. That is a 6 percent increase on more than double the traffic.
Webflow made our website faster to look at and easier to find. It did not make it convert. We had roughly twice the visitors and about the same number of customers.
Era three: Hydra OS
In the first 46 days on Hydra OS, sessions reached 38,808 and users reached 27,417. That is about 844 sessions a day, up 83 percent from late Webflow and up 241 percent from late WordPress.
Organic is the real story.
Source: GA4 sessionDefaultChannelGroup, Organic Search. Window totals 2,117 / 5,110 / 16,245.
Organic search sessions hit 16,245, up 218 percent from Webflow. More important than the raw count is what share of our traffic it represents.
Source: GA4 sessionDefaultChannelGroup.
Organic went from 19 percent of traffic on WordPress, to 24 percent on Webflow, to 42 percent on Hydra OS. For the first time, search is close to being the largest single source of traffic to our own site rather than a rounding error next to direct.
Source: WhatConverts profile 124049. Window totals 88 / 93 / 203 leads.
Leads went to 203 over the window, about 4.4 a day, up 118 percent from Webflow and 131 percent from WordPress. The composition changed more than the total did.
Source: WhatConverts profile 124049. Some recent Hydra-window form submissions resolve to client development hostnames on the shared tracking profile, so the 91 is a ceiling rather than a clean figure.
Web forms went from 1 to 91. Phone calls went from 91 to 112. The site started doing work that previously only a phone call could do.
One more thing showed up that had never appeared before. GA4 recorded 84 sessions from the AI Assistant channel in the Hydra window, and zero in either prior era. That is small. We are reporting it because it is new and directional, not because it is big.
The three eras side by side
Equal 46-day windows. GA4 and WhatConverts.
| Metric | WordPress | Webflow | Hydra OS |
|---|---|---|---|
| Sessions | 11,369 | 21,156 | 38,808 |
| Users | 7,621 | 14,809 | 27,417 |
| Organic sessions | 2,117 | 5,110 | 16,245 |
| Organic share | 19% | 24% | 42% |
| Pageviews | 45,024 | 58,177 | 107,796 |
| Pages per session | 3.96 | 2.75 | 2.78 |
| Tracked leads | 88 | 93 | 203 |
| Phone calls | 85 | 91 | 112 |
| Web forms | 0 | 1 | 91 |
| Key events | 132 | 1,241 | 774 |
Traffic by channel, same windows:
| Channel | WordPress | Webflow | Hydra OS |
|---|---|---|---|
| Direct | 6,680 | 11,527 | 17,096 |
| Organic Search | 2,117 | 5,110 | 16,245 |
| Referral | 1,442 | 1,770 | 2,302 |
| 441 | 1,529 | 1,569 | |
| Organic Social | 274 | 77 | 490 |
| AI Assistant | 0 | 0 | 84 |
What about speed
This is where we have to be careful, because speed is the claim people most expect us to make and it is the claim we can least support across eras.
We did not run Lighthouse against ciwebgroup.com while it was on WordPress or Webflow. We have no saved lab data from those eras. So we are not going to tell you the site got any particular number of seconds faster. We do not know, and any figure we published would be invented. You will notice there is no load-time-by-platform chart in this post. That is why.
What we do have is Hydra-era lab data, and we are publishing it including the bad parts.
Source: Lighthouse 12.8.2 lab runs, overrides/ciwebgroup/content/performance/raw/2026-07-13.
Desktop was excellent. Mobile was not. That is a real finding on our own site, on the platform we sell.
We found the cause. Our navigation island loaded on every page and was pulling in a full market dataset and a release feed to render a handful of links and one version string. That shipped about 99KB gzipped of pure data JavaScript into every page. The largest contentful paint element was hero text waiting on the main thread, not a slow image.
We split those imports. The nav bundle dropped to about 19KB gzipped. Here is what two days of that work did.
Source: Lighthouse 12.8.2 lab runs committed at overrides/ciwebgroup/content/performance/raw/2026-07-13 and /2026-07-15. There is no equivalent lab data for the WordPress or Webflow eras, so this chart is Hydra OS only and is not a cross-platform comparison.
Two pages went from failing to near perfect. The homepage moved from 59 to 66 and stopped there, and a re-run on Jul 18 held it at 66 with a 7.4 second mobile largest contentful paint. Server response time is consistently around 200 to 220 milliseconds, which is the part the architecture gives you for free. The rest is our own page weight, and it is our own work to finish. If you want the method behind these runs, it is in our Lighthouse and page speed guide.
Where this comparison is not clean
A performance article that does not name its own weaknesses is an advertisement. Here are ours.
- The Hydra window is a launch window. We compared the first 46 days of Hydra OS against the last 46 days of the two platforms it replaced. New sites get attention. That framing favors Hydra.
- The windows sit in different seasons and different years. WordPress is late summer 2025. Webflow is late spring 2026. Hydra is early summer 2026. The company also grew substantially across that span, and a bigger company gets more traffic regardless of platform.
- The September 2025 jump is unexplained by platform. It happened on WordPress, four weeks before the Webflow move. Something else drove it. That means our WordPress window is measured at an unusually strong point for WordPress, which if anything understates the Webflow gain.
- Key events went down, not up. Hydra recorded 774 key events against Webflow's 1,241, a 38 percent drop, while sessions, organic, and leads all rose. We believe that is an event-definition difference between the two stacks rather than a conversion collapse, because the lead data moves the other way. We have not finished proving it. Treat that row as unreliable until we do.
- We changed more than the platform each time. Every move came with new content, new information architecture, and new internal linking. You cannot separate the platform from the redesign here. Nobody who migrates a real site ever can.
- Email scaled independently. Email sessions went from 441 to 1,529 between the WordPress and Webflow windows because we built an email program, not because of a platform.
- Search Console figures are head-query sums. Those clicks, impressions, and positions are the top 1,000 queries added up, not property-wide totals. They are directionally right and absolutely incomplete.
- Lead tracking has a known impurity. Some recent form submissions in our WhatConverts profile resolve to client development hostnames on the shared tracking setup. The Hydra lead count may be slightly generous.
- No cross-era speed data exists. The Lighthouse charts above cover the Hydra era only. Any WordPress or Webflow speed number we quoted would be fabricated.
- The Hydra cutover date is pinned to a release, not an archive capture. Public archives have no usable capture of our site between Apr 16 and Jun 26, 2026. We anchored Jun 8 on our own production release and the GA4 landing-path shift.
What we actually take from this
Three honest conclusions.
First, leaving WordPress was worth it, and it was mostly a findability win. Average position went from 37 to 13. That is the difference between existing and being found.
Second, Webflow doubled our traffic and did almost nothing for our leads in matched windows. Traffic that does not convert is a vanity number. We had it for eight months.
Third, the Hydra OS window is the first time traffic, organic share, and leads all moved together, and the first time our own site produced web-form leads at any volume. It is also the era where we can finally see our own mobile performance debt, because we finally instrumented it.
We will re-run this comparison at 180 days on Hydra OS, when the launch window is well behind us and the seasonal comparison is fairer. If the numbers regress, we will publish that too.
Related reading
- Hydra OS, the platform our site now runs on.
- Decoupled architecture, why the server response time above is a property of the stack.
- Lighthouse and page speed, the method behind the lab numbers in this post.
- Website design, how we approach a migration like this for a client.
- Hydra OS launches, the same kind of cutover on customer sites.
- Get Started if you want this path for your own site.
