If your Google Analytics 4 property shows sessions climbing and conversions stuck at zero, the problem is usually not your marketing. It is the install. The GA4 setup mistakes small business teams live with are the silent kind: nothing breaks, no error appears, and the numbers look plausible enough to paste into a monthly slide.
That silence is expensive. You keep funding the channel that looks good in a broken report and you cut the one that quietly produced enquiries. What follows is the audit we run before we touch any digital marketing programme, in the order we run it. Most of it needs no developer at all.
Why does GA4 report traffic but no conversions?
Because GA4 collects events by default but marks almost nothing as a key event. A form submit can fire correctly for months and still never appear in a conversion report. The tag is working. The configuration step that tells GA4 which events matter was never done.
Google now calls these key events. Until you flag an event as one, GA4 treats a quote request exactly like a scroll. Your reports then compare channels on sessions, which is the metric with the weakest link to revenue.
The six GA4 setup mistakes small business accounts repeat
We see the same six failures, roughly in this order of damage. Published 2026 analyses suggest around 40 percent of GA4 setups have at least one event configured incorrectly. The properties nobody has looked at since installation day are the ones worth checking first.
1. Events fire but are never marked as key events
Enhanced measurement hands you page views, scrolls, outbound clicks and file downloads for free. None of those is an outcome. You have to decide which events represent a real business result, then mark exactly those. A property can only hold so many key events, so mark outcomes only, not every button on the site.
This matters more for paid media than it used to. Industry reporting in 2026 puts Performance Max as the dominant Google Ads campaign type for Indian advertisers, and that campaign type optimises against whatever conversion signal you send it. If you are running ads, check how conversion tracking should be wired end to end before you spend another rupee.
Feeding a weak signal is worse than feeding none, because the system will confidently buy more of the wrong traffic. That mechanism sits behind most complaints about Performance Max burning budget without results.
2. Event names are inconsistent, so the data splits
One developer sends form_submit, the plugin sends contact_form_submitted, and the WhatsApp button sends Whatsapp_Click. GA4 treats those as three unrelated things. Every total you report is understated, and no two reports reconcile. Pick lower case with underscores, write the convention down, and make every new tag follow it.
3. Internal and vendor traffic is never filtered
Your team opens the site all day. So do your developer, your agency and the reception desk. Without an internal traffic filter set to Active, that activity sits inside your session and engagement numbers, and a low traffic site feels it hardest. Add your office IP, your work-from-home IPs and any vendor you can identify.
4. Bots and referral spam inflate sessions
Industry reporting in 2026 puts bots and referral spam at roughly 10 to 30 percent of GA4 traffic on small sites. You will recognise it: referrers you have never heard of, sessions with no engagement, visits from places you do not serve. GA4 already filters known bots and spiders on its own, so there is no switch to find; your work is maintaining an unwanted referrals list and excluding the junk sources you can identify.
5. No cross-domain or payment gateway handling
If checkout runs through Razorpay, or bookings sit on a separate scheduling tool, the return visit is logged as a fresh session from a referral. Your own site loses credit for the sale it created. Configure cross-domain measurement for every domain you own, and add gateways to unwanted referrals so a UPI payment does not become a new traffic source.
6. Consent and tag firing are never thought through
Cookie banners installed from a plugin often block the GA4 tag completely, or load it twice. Neither problem is visible from the interface. Decide what your banner should do, test it, and confirm that events still arrive after a visitor accepts.
How do you fix a broken GA4 setup, in order?
Fix it in dependency order, not by whatever annoys you most. Collection first, then naming, then filters, then key events, then integrations. If you mark key events before you clean the naming, you will mark the wrong ones. Budget one focused block of work for a small site, then wait for fresh data before trusting the report.
- Confirm the GA4 tag loads once on every page, thank you pages included.
- Write the event list on paper first: the action, the event name, the parameters it carries.
- Rename and consolidate existing events onto one naming convention.
- Mark only genuine outcomes as key events, then confirm those definitions with whoever answers the phone.
- Add internal traffic filters for office, home and vendor IPs, and set them to Active rather than Testing.
- List unwanted referrals, including payment gateways and booking tools, and rely on GA4's automatic bot filtering for the rest.
- Set up cross-domain measurement, then link Google Ads and Search Console.
One more setting deserves two minutes. The data retention control defaults to the shorter window, and once event data ages out of it, no later fix brings it back. Change it before you need the history.
Two of these steps are structural. Cross-domain measurement and consent behaviour usually need whoever built the site. If nobody owns that work, ask us to scope it rather than rebuilding your reports on a foundation you cannot see.
How do you verify events with DebugView?
DebugView is the live event stream inside GA4 admin. Turn on debug mode, load your own site, complete the action, and watch the event arrive with its parameters. It is the only cheap way to prove a tag fires once, with the right name, at the right moment. Test every key event this way.
Enable debug mode through Tag Assistant, the Tag Manager preview, or the debug option your plugin exposes. Then behave like a customer rather than a tester. Fill the form, tap the WhatsApp button, run a test transaction if you can do it safely, and watch each event land.
- Does the event fire at all when the action completes?
- Does it fire once, not three times?
- Is the event name exactly what your convention says it should be?
- Are the parameters present and readable?
- Is any personal data leaking into a parameter or a URL?
Then wait and check the same events in your standard reports once processing catches up. DebugView proving an event fires is necessary but not sufficient. The report has to show it too, in the right channel, under the right name.
What each mistake does to your numbers
It helps to see the failure and the consequence side by side, because owners rarely connect a wrong decision back to a wrong report. This is the pattern we describe in most audit calls.
| Setup mistake | What the report shows | What it costs you |
|---|---|---|
| Events not marked as key events | Conversions read zero or close to it | Channels get judged on sessions, so the loudest channel wins the budget |
| Inconsistent event naming | One action split across several event names | Every total is understated and no two reports agree |
| Internal traffic not filtered | Inflated sessions and engagement time | Your own team looks like demand, worst on low traffic sites |
| Bots and referral spam | Unknown referrers and sessions with no engagement | Conversion rate looks worse than it is, and good campaigns get cut |
| No cross-domain handling | Payment or booking domain appears as a top source | Your best channel loses credit for the sale it created |
Consent, privacy and what you should never send to GA4
Two rules keep you out of trouble. Never push personal data into GA4: no names, phone numbers, email addresses or health details in event parameters, page paths or query strings. If your thank you page URL carries the phone number a visitor typed, that number is now sitting in your analytics.
Second, treat consent as a decision rather than a plugin default. If you serve visitors in the EU or UK, consent mode changes how tags behave and needs testing. In India, take the banner and data handling questions to your own legal advisor, then implement what they tell you. We can build it; we are not the right people to decide it.
Clinics and healthcare businesses need more care again. Do not send anything that could identify a patient or the condition they searched for, and keep appointment details out of analytics parameters entirely.
Why weekly reporting is meaningless without this
We work in weekly checkpoints, and a checkpoint only earns its place if the numbers can change a decision. When conversions read zero, the meeting becomes a review of traffic. Traffic rewards whoever published the most content, not whoever produced the most business.
Sessions are a weaker base than they used to be anyway. Published 2026 analyses suggest Google AI Overviews appear in roughly 20 percent of searches, and that about 60 to 63 percent of Google searches end without a click. Fewer people arrive before they decide, so the visits you do get carry more weight and deserve accurate measurement.
Once key events are clean, the weekly conversation changes shape. You compare cost per enquiry across channels instead of arguing about sessions, and you can see which pages produce qualified leads. That is the point where bringing cost per lead down becomes an engineering exercise instead of a guess.
It also changes what you look at. Keep a short list of numbers, hold it stable for a quarter, and ignore the rest; we set out that shortlist in our guide to the numbers growth teams actually review. Our performance marketing team will not scale a campaign until GA4, Google Ads and the client's own enquiry log agree with each other.
Frequently asked questions
Will fixing GA4 recover the conversion data I already lost?
No. GA4 does not backfill. Once you mark a key event it counts from that moment forward, and data that has aged out of your retention setting is gone. Fix the setup, note the date you changed it, and compare like with like from then on.
Do I need Google Tag Manager to fix this?
Not always. On a WordPress site with a well maintained analytics plugin, you can get clean events without it. Tag Manager earns its place when you run several tags, custom form flows, a payment gateway, or when more than one person touches the site.
How long does a GA4 audit and rebuild take?
The work on a small service or brochure site is short. The waiting is the long part, because you need fresh data after the changes before any comparison means anything. We schedule the audit first, then hold the first honest review once clean data has accumulated.
Is GA4 enough, or do I need another analytics tool?
For most small businesses in India, a correctly configured GA4 plus Search Console plus your CRM or WhatsApp enquiry log is enough. Extra tools do not repair a bad measurement plan. Add one only when you have a question GA4 genuinely cannot answer.
Where to start this week
Open your property and check one thing. Does the conversions number for last month match the enquiries you actually received? If it does not, stop reading dashboards until the property is fixed, because every budget decision after that is guesswork wearing a chart.
We run this audit for businesses across Pune and PCMC, from the manufacturing belt around Akurdi and Chakan to the clinics and retail brands in Baner and Wakad, and for Vadodara clients served from the same Pune office. If you want a second pair of eyes on your setup, our team in Pune will take a look. Send the property access and your list of actions that count as a lead, and we reply within one business day: start with a short brief.
