Speed advice usually arrives wrapped in consumer statistics. Shoppers abandon carts, retailers lose a percentage of revenue per second of delay, and so on. All true, and none of it quite lands for a company selling filtration systems to food plants.
The B2B version of the argument is different and, if anything, sharper. You get fewer visitors than a retailer, and each one is worth vastly more. A maintenance manager who gives up on a slow page isn’t an abandoned cart. They’re a contract you’ll never know you were in the running for.
Speed is also one of the few things on this list that Google measures directly and tells you about for free.
The three metrics, in plain language
Google measures page experience through three numbers, collectively called Core Web Vitals. They’ve been stable since Interaction to Next Paint replaced an older responsiveness metric in March 2024, and Google has announced no changes for 2026.
- Largest Contentful Paint (LCP) measures how long until the main content appears. Usually that’s your hero image or headline. Good is 2.5 seconds or less. Poor begins above 4 seconds.
- Interaction to Next Paint (INP) measures how quickly the page responds when someone taps or clicks. Good is 200 milliseconds or less. Poor begins above 500.
- Cumulative Layout Shift (CLS) measures how much the page jumps around while loading. The button that moves just as you go to press it. Good is 0.1 or less. Poor begins above 0.25.
You need all three in the good band to pass. There’s no combined score and no averaging, which is why a site can be quick to load and still fail.
LCP
- Largest Contentful Paint
- How fast the main content shows
- Good: 2.5 seconds or less
INP
- Interaction to Next Paint
- How fast it responds to a tap
- Good: 200 milliseconds or less
CLS
- Cumulative Layout Shift
- How much the page jumps
- Good: 0.1 or less
The part most people get wrong
Here’s the misunderstanding that wastes the most effort. Core Web Vitals are measured from real visitors, not from a test you run.
Google grades you at the 75th percentile of actual Chrome users over a rolling 28-day window. That means three out of four real page views have to hit the good threshold. Not your laptop on running on you office’s 1 Gbps fibre connection. It’s the person on a three-year-old Android phone with two bars of signal that also counts.
Two practical consequences follow.
The first is that a perfect score in a testing tool means very little on its own. Those tools run a simulation on a machine you configured. They’re useful for diagnosis, and useless as proof.
The second is patience. Because the data covers 28 days, a fix you deploy today won’t show its full effect for a month. Judging a change after a week will mislead you.
For real numbers, use the Core Web Vitals report in Google Search Console, which draws on field data. PageSpeed Insights shows both, with the field data at the top and the lab simulation beneath.
Field Data
- Real visitors, real devices
- 75th percentile, 28-day window
- What Google actually grades
- Search Console and PageSpeed
Lab Data
- A simulation you configured
- Useful for diagnosis only
- Not proof of anything
- Flatters desktop badly
What actually slows a B2B site down
In our experience the same handful of causes account for most failures on B2B marketing sites.
- Oversized images. A hero photo exported at 4000 pixels wide and 3MB. This is the single most common cause of a failing LCP, and the easiest to fix.
- Too many plugins. Each one loads its own scripts and styles, often on every page whether needed or not.
- Third-party scripts. Chat widgets, analytics, tracking pixels, embedded videos, and marketing tags. Each one is code from someone else’s server running on your page.
- Cheap or overloaded hosting. Shared hosting where hundreds of sites compete for the same resources shows up as slow server response, which delays everything after it.
- Layout that isn’t reserved. Images and ads without declared dimensions push content around as they load, which is most CLS problems in one sentence.
- Heavy page builders. Some generate a great deal of markup for a simple layout, and that weight has to be downloaded and parsed.
None of that is exotic. It’s mostly accumulation, which is why speed drifts downward over time on sites nobody is actively maintaining.
What Slows a B2B Site Down
Mobile is where B2B sites actually fail
Desktop scores flatter almost everyone, and B2B companies are especially prone to trusting them, because the team building the site works on good machines with fast connections.
Your buyers frequently don’t. In B2B the phone is often the worst-case device in the worst-case conditions. A maintenance manager checking suppliers from a plant floor. A site foreman between meetings. A procurement lead on a train commute.
The thresholds are identical on mobile and desktop, but mobile is far harder to pass, because the same page has to render on a slower processor over a weaker connection. Google’s own field data shows a large share of mobile sites missing the LCP threshold, and it’s the metric most sites fail.
So when you check your numbers, check mobile first and treat desktop as the easy half. If you only have time to fix one, fix the one your buyers are actually holding.
Where to start
If your site is failing, work in this order rather than starting with whatever a tool listed first.
- Fix whichever metric is in the poor band. Poor hurts more than needs-improvement, so clear the worst first.
- Compress and correctly size your images. Serve modern formats, size them to their display dimensions, and lazy-load anything below the fold.
- Audit your plugins. Remove what nothing uses. This helps speed and security at the same time.
- Review third-party scripts. Ask what each one earns. A chat widget nobody answers is pure cost.
- Reserve space for images and embeds by declaring dimensions, which usually clears CLS quickly.
- Check your hosting. If server response time is slow, everything downstream inherits the delay.
Work at template level rather than page by page. Fixing your service page template fixes every service page, which is a better use of an afternoon than optimizing one URL.
A worked example
A B2B site comes to us failing on mobile. The owner has run a testing tool, seen a score in the nineties, and can’t understand the problem.
The field data tells a different story. LCP sits at 4.1 seconds at the 75th percentile, well into the poor band, while the lab test on a simulated desktop connection returned 1.2 seconds. Both numbers are accurate. Only one of them describes actual customers.
The diagnosis takes about twenty minutes. The homepage hero is a 2.8MB photograph, served at full resolution to every device including phones. A chat widget nobody has answered in a year loads on every page. Three separate analytics tags are running, two of them left over from campaigns that ended. The theme’s slider library is loading on pages that have no slider.
The fixes are unglamorous. Compress and resize the hero, which alone takes more than a second off LCP. Remove the chat widget and the two dead tags. Load the slider only where it’s used.
Four weeks later (because that’s how long the rolling window takes to catch up) the site passes on mobile. Nobody rebuilt anything, and the total work was under a day.
That’s the usual shape of it. Speed problems on B2B marketing sites are rarely architectural. They’re accumulated weight that nobody has taken responsibility for removing.
What not to obsess over
A perfect score is not the goal, and chasing one has a real cost.
Google has been clear that good Core Web Vitals don’t guarantee good rankings. They’re one input among many, and content and relevance still matter far more, as we cover in our guide to B2B SEO. A site with excellent scores and nothing to say will lose to a slower site that answers the question.
It’s also worth knowing that most sites fail somewhere. Google’s own field data shows a substantial share of mobile sites missing the LCP threshold. Being in the good band puts you ahead of a lot of your competitors, and squeezing from good to perfect rarely earns back the hours.
The practical target is passing all three on mobile, on your important templates, and staying there.
It also should be mentioned that speed is not everything. It has to be balanced with your marketing message and the aesthetics of your website. For example, if you offer videography services, there is a very real chance that your homepage hero section will have a background video playing. In such a scenario, it may be acceptable that the visual impact of the video outweighs the small performance penalty.
Speed is a feature, not a score
The reason to care about this isn’t the badge in a report. It’s that a slow site quietly removes people from your pipeline before you know they existed, and it does it disproportionately to visitors on phones and weaker connections, which in B2B often means someone checking you out from a plant floor or a job site.
Get the three numbers into the good band, keep them there through ordinary maintenance, and then go back to worrying about what the pages actually say.
If you’d like to know how your site currently performs for real visitors rather than in a test, book a discovery call and we’ll pull the field data together.
