Most website reporting is a wall of numbers nobody acts on. Sessions up, bounce rate down, average time on page holding steady, and no clearer sense of whether the site is producing customers. The problem isn't the reporting tool. It's that the website KPIs being tracked measure activity rather than outcomes.
For an established B2B company, a small number of metrics answer the only question that matters: is the site turning the people who arrive into people who contact you.
The test for whether a metric is worth tracking
Before the list, the filter. A metric earns its place if a change in it would cause you to do something different. If a number moves and nobody knows what to do about it, you're collecting data, not measuring performance.
That single test eliminates most of what appears in a standard monthly report.
The four worth tracking
Conversion rate
The share of visitors who contact you. This is the closest thing to a single measure of whether the site works, and it's the one most companies never set up.
Track it as contacts divided by visitors, monthly. Watch the trend on your own numbers rather than an industry benchmark, because published benchmarks blend ecommerce, consumer and B2B traffic and tell you almost nothing about a manufacturer with referral-heavy traffic.
What makes it useful is that it isolates the site. Traffic can swing for reasons outside your control; conversion is mostly about whether the page does its job. If conversion rises on flat traffic, the site got better. The term is defined in our glossary entry on website conversion rate.
Contacts and quote requests, by source
Raw counts, split by how people arrived: referral, organic search, direct, paid, email.
This is the number your sales team already cares about, and splitting by source is what makes it actionable. Twenty contacts is a fact. Twenty contacts where fourteen came from people typing your web address and two came from search tells you the site isn't yet reaching anyone new, which is a completely different problem from "we need more traffic."
Track quote requests separately from general contacts if you can. They convert differently and they're the ones worth optimizing for.
Search impressions and striking-distance queries
In Google Search Console, two things matter more than rank.
Impressions show whether Google is surfacing you at all. This moves months before clicks do, which makes it the earliest honest signal that SEO work is having an effect. A program that looks dead on traffic is often clearly alive on impressions. If a new agency is running that program, how long to give a new marketing agency covers what each channel can honestly show by day 30, 60 and 90.
Striking-distance queries, meaning queries where you sit in positions five to twenty, show what's close to breaking through. Those are the pages worth improving next, because moving from position eleven to position six produces far more than starting something new.
Also worth watching: the split between branded and non-branded queries. If nearly everything is your company name, you're being found by people who already knew you existed.
Page speed on mobile
Not because speed is a goal in itself, but because it caps everything above it. A slow page loses buyers before the message loads, and B2B buyers check from shop floors and trucks more than most owners assume.
We measured 56,005 US B2B homepages against the three-second rule: 70.5% missed it, and the median PageSpeed score was 59 out of 100. The detail is in our speed benchmarks. If your mobile score is poor, fix it before optimizing anything downstream of it.
The ones that waste your time
Sessions on their own. Traffic without conversion is a cost, not an achievement. A site that gets more visitors and the same number of contacts got worse, not better. Sessions are only meaningful next to conversion.
Bounce rate, mostly. It's widely misread. In GA4 a bounce is a visit of ten seconds or less with one page view and no key event, so a buyer who spots your phone number and calls straight away counts as one unless the call click is tracked as a key event. It's only useful when compared against itself over time on a page you've deliberately changed, and even then it's weak. Our bounce rate entry covers the nuance.
Average time on page. Longer isn't better. A buyer spending four minutes hunting for your phone number is a failure, not engagement. Time on page can just as easily indicate confusion.
Third-party SEO scores. The graded numbers various tools produce are estimates of an estimate. They're useful for spotting broken technical items and useless as a performance measure. Google Search Console is what actually happened.
Rankings alone. Position one for a term nobody searches, or for a term buyers never use, produces nothing. Rank is only meaningful paired with impressions and clicks.
Pageviews per session. For a B2B site this is close to meaningless. A buyer who found what they needed on one page and called you is the outcome you want.
Setting it up properly
Two steps do most of the work, and without them everything above is guesswork.
Mark form submissions as key events in GA4. Every quote form, contact form and any tracked phone click. Until this exists, conversion is estimated rather than measured, and most companies are estimating.
Get Google Search Console access for the domain. It's free, it takes minutes, and it's the only place you see the real query data. If you don't have access, whoever built the site can add you.
Two optional additions worth it for most manufacturers: tag your quote form separately from your general contact form so you can see which is producing, and set up call tracking if a meaningful share of your inbound arrives by phone, because otherwise your best channel looks empty in analytics.
Setting up a key event, concretely
"Mark form submissions as key events" is the sort of instruction that gets nodded at and never done, so here is what it involves.
In most analytics setups you're doing one of two things. Either you fire an event when the form submits, or you count arrivals at a thank-you page. The thank-you page approach is simpler and more reliable: send every successful submission to a dedicated URL, then count views of that URL as the conversion. It also keeps rejected submissions out of the count, which matters once spam starts reaching the form.
That means each form needs its own confirmation page. Quote requests to one, general contact to another, capability downloads to a third. It sounds fussy and it's the difference between knowing you got 22 contacts and knowing you got 14 quote requests and 8 people asking about careers.
If a meaningful share of your inbound arrives by phone, add call tracking. Otherwise your strongest channel shows up as zero in every report, and you'll under-invest in it for years without knowing why.
Which channels actually bring in quote requests?
To tell which channels bring in quote requests, record the source on every request as it arrives, then count requests by channel each month. Visits by channel won't answer it. A form submission shows up in analytics by channel once it's marked as a key event, but a call, an email or a buyer you met at a show often reaches you with no source at all, so the count has to come from three places: analytics for forms, call tracking for calls, and the conversation for everything else.
- Forms. In GA4, the Traffic acquisition report lists key events by channel, so once the quote form is a key event you can read form requests by source there. That's the source GA4 credits to the visit that converted (an earlier source if they came back direct), though, so a buyer a peer referred, or one you met at a show, who later searches your name shows up as search or direct. When you follow up a form request, ask how they found you too. It also pays to store the source on the request itself, the page it came from and the channel that sent the visitor, so sales sees it on the record and not only in a report. Our guide to CRM website integration calls this the source level.
- Calls. On the website, the call tracking above only tells you which channel rang if the number on the page changes with how the visitor arrived. Most call-tracking services do that by swapping the number in when the page loads, with a number per channel or one per visitor. Where that's more setup than you want, ask on the call and write the answer down.
- Everything with no click. Referrals, trade shows, a distributor's rep. These come in by phone or email, or as direct traffic with nothing on record about where it started. Ask how they found you on the first call and log it on the request. Keep that question in the conversation rather than adding it to the form, because a longer form tends to get fewer requests.
Tag the links you control, such as email signatures, links inside PDFs and quotes, and show handouts, with campaign tags. Untagged, those visits can land in direct traffic or be credited to whatever brought that visitor last time, so the link that actually sent them gets no credit.
Then keep one tally: quote requests by channel, each month, and a quarter later, which of them became orders. A channel that sends plenty of requests that never turn into work is telling you something different from one that sends three a month and wins two.
A monthly review that takes fifteen minutes
Reporting fails when it's long. Here is a review that fits on one page:
- Contacts this month, by source. Up, down, flat, and which source moved.
- Conversion rate. This month against the last three.
- Search impressions. The trend line, not the number.
- Top five striking-distance queries. Which pages are close to breaking through.
- Branded versus non-branded split. Are we reaching anyone new.
- One action. What we're changing before the next review.
That last line is the one that matters. A report that ends without a decision is a report nobody needed.
Review it on the right cadence
These move slowly, and reviewing them weekly mostly produces noise plus the temptation to react to normal variation.
Monthly: search impressions, striking-distance queries, contacts by source. Quarterly: conversion rate trend, branded versus non-branded split, mobile speed. Annually: whether the site still describes the business, which is a judgment call rather than a metric.
For a company with a long sales cycle, a quarter is often the shortest window where a change in conversion is distinguishable from ordinary variation.
What to do when a number moves
The point of the short list is that each metric implies an action.
Conversion falls: something on the page or the form broke, or traffic mix shifted toward less qualified visitors. Impressions rise but clicks don't: your titles and descriptions aren't earning the click, or you're ranking for the wrong queries. Contacts from search stay flat while impressions climb: you're being surfaced for terms that aren't buyer-intent. Mobile speed drops: something was added to the page.
If the answer you keep arriving at is "our conversion rate is low," that's the most valuable thing to fix before spending anything on more traffic.
