Analytics Setup Guide: What to Track and How to Use It
Strategy · 5 min read ·
By Heath Daniel, Founder & AI Platform Architect, HD Connex · Published:
Direct answer: A useful analytics setup tracks a handful of actions that indicate real business outcomes — form submissions, calls, bookings — rather than pageviews and bounce rate. Configure conversion events first, then traffic sources, and ignore vanity metrics entirely.
Most analytics setups collect enormous amounts of data and answer no useful question. The fix is not more tracking — it is deciding what a good outcome looks like before you measure anything.
What should you actually track?
Start from the business, not the tool. For nearly every local or service business, the questions worth answering are:
- How many people contacted us this month?
- Where did they come from?
- What did they look at before contacting us?
- Which of those channels produced work worth having?
Everything you configure should serve one of those. If a metric does not, it is decoration.
Conversion events come first
An analytics install without conversion tracking is a traffic counter. The single highest-value configuration step is defining what counts as a conversion:
- Contact form submission.
- Phone call initiated by tapping the number.
- Booking or appointment completed.
- Quote request submitted.
- Optionally, a meaningful engagement such as reaching a pricing page.
Until these exist, you cannot answer whether any channel is working, because you can only see visits.
Phone tracking is the one most often skipped and most often the largest channel for local businesses. If your number is a tappable link, the tap can be tracked as an event, and suddenly a big blind spot closes.
Which metrics mislead people?
- Pageviews. More pages viewed can mean engagement or it can mean the visitor could not find what they wanted.
- Bounce rate. A visitor who lands on your contact page, reads the phone number, and calls you registers as a bounce. That is a success being recorded as a failure.
- Time on page. High duration can mean interest or confusion. It cannot distinguish between them.
- Average position in search. An average across all queries hides the ones that matter.
None of these are useless, but none should be optimised directly. They describe behaviour; they do not measure outcomes.
Traffic sources: where enquiries actually originate
Once conversions exist, segment them by source. The common and uncomfortable finding is that the channel producing the most traffic is not the one producing the most work.
Watch for:
- Organic search, split between brand and non-brand queries. Brand traffic was going to find you anyway; non-brand is what marketing earned.
- Direct, which is a mix of genuine repeat visitors and untagged links.
- Referral, useful for spotting citations and partner links that actually send people.
- Paid, which should be judged on cost per acquired customer rather than clicks.
Tag your own campaign links properly with UTM parameters, or that data becomes guesswork.
Search Console tells you what Analytics cannot
Analytics shows what happened on your site. Search Console shows what people searched to reach it, which pages appeared for which queries, and where you sit in results.
That combination is where content decisions come from. A page ranking eighth for a query with real volume is usually a better investment than writing something new, because it is already close.
How do you avoid drowning in dashboards?
Build one view with five numbers and look at it monthly:
- Total conversions.
- Conversions by channel.
- Conversion rate.
- Top five landing pages by conversions, not by traffic.
- Non-brand organic clicks.
If a metric is not in that list, it is diagnostic — something you consult when one of the five moves, not something you monitor.
How often should you look?
Monthly. Weekly review of a small-business dataset produces decisions driven by noise, because ordinary variation is large relative to the sample. Judge changes over quarters.
The exception is after a launch or a campaign start, when daily checking for a week is justified to catch something broken.
Common setup mistakes
- Tracking installed on some pages but not all, producing phantom traffic patterns.
- Internal traffic not excluded, so your own team inflates the numbers.
- Form submissions counted on page load rather than on successful submit.
- Multiple tracking scripts double-counting everything.
- No goal values, making it impossible to compare channels by worth rather than volume.
Where to start
Check whether your contact form submissions are being recorded as conversions right now. On a surprising number of sites they are not, which means every report produced so far has been measuring traffic and calling it performance.
If you would rather have measurement configured as part of a build, it is included in HD Connex website packages — and where enquiries are handled by an AI agent, conversation outcomes can be tracked alongside form conversions for a complete picture.
Frequently Asked Questions
What should a small business actually track?
Conversion events above all: form submissions, phone calls, bookings, and quote requests. Then which channels produce them and which pages precede them. Pageviews, sessions, and bounce rate are context, not outcomes, and optimising for them frequently makes the business worse.
Is Google Analytics enough on its own?
For most small businesses yes, provided conversion events are configured. Analytics without conversion tracking is just a traffic counter. Pair it with Search Console for query data, since Analytics does not show what people searched to find you.
What is a good conversion rate?
It depends far more on traffic quality and intent than on any benchmark. For local service businesses, 2 to 5% of visitors taking a contact action is common. Your own trend over time is more useful than any published average.
How often should analytics be reviewed?
Monthly for most small businesses. Weekly checking produces noise-driven decisions, because normal week-to-week variation looks like signal when the sample is small. Look at quarters to judge whether anything actually changed.
Topics: Analytics, Data, Strategy
Related: ROI Audits · Lead Generation · Industry solutions (Daytona Beach) · Website design packages