Google Analytics shipped two changes this month that matter to anyone running their own marketing. Dashboards went live on 9 September 2026, and Include filters for hostnames followed on 21 September 2026, according to Google Analytics' own release notes (accessed 26 September 2026). Together they fix the two most common complaints small business owners have about GA4: it takes too long to find the numbers that matter, and you cannot trust the numbers once you find them.
This is not a feature tour. It covers what a UK owner should build, what to switch on, and where the new tools can quietly go wrong.
What changed in Google Analytics in September 2026?
Google's release notes describe Dashboards as bringing drag-and-drop building and simpler chart creation to GA4. Search Engine Roundtable (11 September 2026) lists the six chart types available: score cards, tables, line charts, bar charts, donut charts and funnel charts. Dashboards publish straight to the left-hand navigation rather than sitting in the report library.
The second change is less visible but arguably more useful. GA4 has had exclude-style hostname filters since 11 June 2026, which meant blocking spam domains one at a time. The new Include mode flips that, as Search Engine Journal (22 September 2026) explains. You list your own domains and GA4 drops everything else, including events with an empty hostname.
Why does this matter more for a small business than a big brand?
A national retailer with a million sessions a month barely notices 300 junk visits. A Derby electrician with 1,500 sessions a month does. The same is true of time. Large companies have an analyst who builds reports. An owner checking GA4 on a Sunday evening has perhaps ten minutes, and the old GA4 interface spent most of those ten minutes on navigation.
The practical effect is that small businesses have been making budget calls on figures that were both hard to reach and partly wrong. These two releases address both problems directly, at no cost.
What should a UK owner put on a 15-card dashboard?
PPC Land reports a limit of 15 cards per dashboard on a standard property and 30 on Analytics 360. It also notes that segments, card-level comparisons and API access are not supported, and that every published dashboard is shared with everyone who has access to the property. There are no private dashboards, so do not put anything on there you would not want a freelancer with access to see.
A workable owner scorecard for a service business:
- Score card: total enquiries (form submissions plus call clicks) this month
- Score card: enquiries from Google Ads
- Score card: enquiries from organic search
- Score card: enquiries from Google Business Profile traffic, if you tag your profile link
- Line chart: enquiries per week, last 12 weeks
- Bar chart: enquiries by channel
- Funnel chart: landing page view, contact page view, form start, form submit
- Table: top 10 landing pages by enquiries
- Table: top 10 landing pages by sessions with zero enquiries
- Score card: sessions, for context only
- Donut chart: mobile vs desktop enquiries
- Table: enquiries by town or city
For ecommerce, swap enquiries for purchases and revenue, and add average order value and a checkout funnel. Leave three cards spare. You will think of one you need within a month.
If you want help turning card 9 into a list of changes, the thinking behind conversion rate optimisation for small businesses starts from exactly that report.
What is the hostname Include filter and who should switch it on?
Every event GA4 receives carries a hostname, the domain the tracking code fired on. Spam traffic often sends fake events straight to your measurement ID with a made-up hostname or none at all. Search Engine Journal notes that a missing hostname usually points to spam or abnormal traffic, and Google confirms the Include filter blocks empty hostnames automatically.
Use this decision framework:
- If you run one website on one domain, set an Include filter with both the www and non-www versions. This is the simplest and safest case.
- If you use subdomains (shop.yoursite.co.uk, book.yoursite.co.uk), list each one explicitly. Search Engine Journal notes that Google's documentation does not explain how subdomains are matched, so do not assume they are covered.
- If a booking, quote or checkout tool runs on someone else's domain but carries your GA4 tag, add that domain too, or those enquiries will disappear from your data.
- If one GA4 property collects data from two sites, for example a main site and a microsite for a local campaign, list both.
- If you send server-side events through the Measurement Protocol, know that the filter will not touch them. That is by design, and spam sent this way will still get through.
Which UK small businesses should be careful with the allowlist?
Search Engine Journal reports that once a filter is active, filtered events "will never be available in Google Analytics or BigQuery". There is no undo. That matters most for:
- Salons, clinics and gyms that take bookings through an external booking page
- Trades that use a hosted quote form
- Shops with a checkout or payment step on a separate domain
- Anyone who moved domain in the last year, for example from a .com to a .co.uk, and still gets traffic on both
The protection is the Testing state. During testing, matching events are labelled with a "Test data filter name" dimension rather than deleted. Run the test for at least seven days so it covers a full weekly cycle, then build a quick table of hostnames the filter would have removed. If you recognise any of them, add them before you activate.
What does this look like with real numbers?
Take an illustrative Nottingham kitchen fitter. The figures below are a worked example, not sourced data.
- GA4 shows 2,400 sessions and 40 enquiries a month. Conversion rate: 1.67%.
- Google Ads spend is £900 a month, and GA4 credits Ads with 15 enquiries. Cost per enquiry: £60.
After a week in Testing, the hostname filter flags 400 sessions from unknown or empty hostnames, carrying 12 fake form events, 4 of them attributed to paid traffic.
- Clean data: 2,000 sessions, 28 enquiries. Conversion rate: 1.4%.
- Clean Ads data: 11 real enquiries for £900. Cost per enquiry: about £82.
If the fitter closes one enquiry in four at an average job of £7,500, 11 enquiries is roughly 2.75 jobs, or about £20,600 in revenue from £900 of spend. Ads are still clearly profitable. What changes is the ceiling. At £60 per enquiry, the owner might have raised the budget on the assumption that each extra £300 buys five enquiries. On clean data it buys fewer than four. That is still worth doing, but the forecast is different, and so is the cash plan. If paid search is part of the mix, a sober view of paid media for small businesses starts with exactly this sort of cleaned figure.
How does this connect to the August conversion window change?
Google's release notes say click-through conversion windows can now be set anywhere from 1 to 90 days, and engaged-view windows from 1 to 30 days, where the engaged-view window was previously fixed at 3 days. For a business where customers research for weeks, such as kitchens, extensions, solicitors or wedding venues, a longer window credits the channels that started the journey. For a takeaway or a same-day plumber, a short window is more honest.
Set the window first, clean the data second, and build the dashboard third. In that order, the numbers on your dashboard mean something.
What are the trade-offs and limits?
The honest trade-offs:
- No segments or comparisons on cards, per PPC Land. You cannot, for example, put "this month vs last month" on one card. Use a line chart instead.
- Everything is shared. Every published dashboard is visible to every user on the property.
- Filters are one-way. Anything filtered out after activation is gone.
- Garbage in, garbage out. If your form submission event fires on page load, or call clicks are not tracked, a dashboard just presents the error more neatly.
- Looker Studio still has a place if you need to blend GA4 with Google Ads cost data or a CRM export on one page.
What should you do this week?
A five-step plan that fits around running a business:
- Monday: Open Admin, then Events, and confirm your enquiry, call and purchase events are marked as key events and firing correctly. Submit your own form once and watch it appear in Realtime.
- Tuesday: Run a quick report of sessions by hostname for the last 90 days. Write down every hostname you recognise as yours.
- Wednesday: Create a hostname Include filter with that list and leave it in Testing. Allow 24 to 36 hours for it to start applying.
- Thursday: Write your 12 cards on paper. Cut any that would not change a decision.
- Following week: Review what the test would have removed, add any missing domains, activate the filter, and build the dashboard.
If you want to save the weekly checking time once the dashboard exists, the approach in cutting admin hours with AI covers turning a weekly report into a short summary you actually read.
Is this a reason to rethink how you report on marketing?
Most small businesses do not have a data problem. They have a habit problem, because the numbers were too slow to reach and too messy to believe. With one page and cleaner inputs, a Monday review becomes a short routine: check enquiries, check the channel mix, look at card 9, pick one page to improve. Owners who keep that routine for a quarter are far better placed to make budget calls than those who look at analytics only when something feels wrong.
GA4 Dashboards and Spam Filters — FAQ
What are Google Analytics Dashboards and when did they launch?
Dashboards arrived in Google Analytics on 9 September 2026, according to Google's own release notes. A dashboard is a single page where you place the numbers you care about as cards, using a drag-and-drop grid. You can pick from score cards, tables, line charts, bar charts, donut charts and funnel charts. Once you publish a dashboard, it shows up in the left-hand navigation of your property. For a small business, that means you can open GA4 and see enquiries, sales and traffic sources on one screen, with no need to click through several standard reports every Monday morning.
How many cards can I put on a GA4 dashboard?
PPC Land reports a limit of 15 cards per dashboard on a standard (free) Google Analytics property and 30 on an Analytics 360 property. Each chart counts as one card, even a single score card. For most UK small businesses, 15 is enough if you are strict about it. A useful test is to ask of every card whether it could change a decision you make this month. If it could not, leave it off. You can build a second dashboard for detail if you really need one.
What does the new hostname Include filter do?
Google Analytics added Include data filters for hostnames on 21 September 2026. You list the domains that belong to you, and GA4 drops events that come from any hostname not on that list. Google's release notes say events with an empty hostname are blocked automatically, and Search Engine Journal reports that Measurement Protocol events are exempt. Before this, you had to keep an exclude list of spam domains up to date by hand. The allowlist reverses that. You maintain a short list of your own domains instead of chasing spammers.
Can the hostname filter delete real data by mistake?
Yes, and that is the main risk. Search Engine Journal reports that once a filter is active, matching events will never be available in Google Analytics or BigQuery. If you forget a subdomain, a booking system domain that carries your tag, or a second website that shares the property, those visits are lost for good. That is why the Testing state exists. Run the filter in Testing for at least a week, check which hostnames it would have removed, and only then activate it. A filter can take 24 to 36 hours to apply.
Do I need to pay anyone to set this up?
Not usually. If you have Editor or Administrator access to your GA4 property, you can build a dashboard and set up a hostname filter yourself in an afternoon. The hard part is not the clicking. It is deciding which 10 to 15 numbers actually drive your revenue, and making sure your enquiry and sale events are tracked properly in the first place. If your key events are broken or missing, a dashboard will just show you wrong numbers more neatly. Fix the tracking first, then build the dashboard.
