Attribution, cost and ROI
Website traffic up but leads down: where the drop usually hides
Written by the Top SEO Experts Team
Reviewed by Akshay VRLast updated 18 min read
Quick answer
When website traffic rises and leads fall, the extra visitors are usually not the people who buy. The usual causes are new traffic from informational queries, buyers shifting to the map pack, and calls your tracking misses. Others are bot visits, a broken form, and fewer clicks on high-intent searches.
Example In the hypothetical we build below, a plumber’s organic traffic climbs 50 percent in six months while leads fall about 5 percent.
The site never broke. The mix of visitors changed, and the dashboard only counted visits.
This problem is a trap because every report you open can look healthy. The fix is to confirm the drop is real, then work through six places it hides, in an order that saves money.
Why is my website traffic up but leads are down?
Leads are the product of three numbers. They are visits, the share of visitors with buying intent, and the share of those who can reach you and be counted. Traffic is only the first, so it can rise while the other two fall.
For a service business that makes “more traffic” a weak goal. A plumber needs the few dozen people searching “water heater repair near me” more than thousands reading about why water heaters make noise.
Three things can break the chain:
- The visitor mix tilts toward researchers.
- The path changes, so buyers call from Google without ever visiting the site.
- Measurement fails: real leads happen and nobody records them.
A lot of advice on this topic starts with conversion rate optimization: button colors, form length, page speed. Those matter, and our guide to CRO for service businesses covers them.
Our view The first fault usually sits upstream of the page.
How do you confirm the drop is real before changing anything?
Compare the same weeks last year, define “lead” in writing, and list what changed on your side. If the drop disappears under those three checks, it was never a drop, and you have saved yourself an agency call.
- Compare the same weeks last yearUse the same window a year earlier, not last month.
- Define "lead" in writingSay which calls, forms and bookings count.
- List what changed on your sideNote every site, plugin, vendor or tag change in the past six months.
Which window should you compare?
Start with the window. Google’s guide to debugging traffic drops says to view 16 months in the Performance report to see whether a decline repeats every year. It also says to check Google Trends for wider shifts in demand.
A March to April comparison for an HVAC company says almost nothing.
What counts as a lead?
Next, write down what counts as a lead: a call over 30 seconds, a form submission, a booking request. In GA4, Google defines a key event as “an event that measures an action that’s particularly important to the success of your business.”
If your key events list is empty, you can’t say leads fell. Our GA4 for SEO guide shows the setup.
What changed on your side?
Last, list every change on your side in the past six months. Examples are a new site, a form plugin, a phone vendor, a cookie banner, and an analytics tag. Any one of them can halve recorded leads overnight with no change in real demand.
Where does the drop usually hide?
It hides in six places. The table is the short version, listed in the order we read the reports. The sections after it explain how to run each check, and a later section gives the order we fix them in.
| What you see | Likely cause | Where to check | Typical fix |
|---|---|---|---|
| Clicks up, mostly on “how to” queries | Wrong-intent traffic | Search Console, Queries | Strengthen service pages |
| Site visits steady, calls from Google moving | Map pack shift | Business Profile performance | Work the profile and reviews |
| Fewer calls logged, phones still ring | Tracking gap | Call tracking, GA4 key events | Fix numbers, tags, goals |
| Sessions jump, engagement near zero | Bots or spam | GA4 vs Search Console clicks | Report on real clicks |
| One page loses calls or forms | Broken conversion path | Test on a phone | Fix form, phone link |
| Impressions steady, top-query clicks fall | Fewer clicks per search | Search Console filters | Rework snippets and intents |
Is the extra traffic coming from the wrong queries?
Often, yes. A blog post or glossary page takes off for a broad informational query, total clicks rise, and those readers were never going to hire you.
How do you split service pages from blog posts?
Open the Performance report in Search Console, set the date filter to the last 90 days, then to the same 90 days a year earlier. Use the Pages tab to separate service and location pages from blog posts.
Google’s report documentation defines a click as someone clicking a link from Google to your site, so these are real searches, not tag noise.
How do you split brand from non-brand?
Then split brand from non-brand. Google announced a branded queries filter on November 20, 2025, according to Search Engine Land.
Google says the filter is unavailable for sites with few impressions, so you may not see it. Our guide to branded vs non-branded search shows a manual split.
How do you read the shape?
Read the shape. If service-page clicks held and blog clicks tripled, flat leads are normal. If service-page clicks fell while the blog grew, you traded buyers for readers.
Our view The cure is rarely fewer posts. It is stronger service pages matched to what people want, as our search intent guide explains.
Did buyers move to the map pack?
Searches for a service near the searcher often end on the map. A caller can read your reviews and tap call without a single website visit. So site traffic looks fine while leads move to a place you aren’t watching.
Google’s Business Profile performance help defines the numbers. Calls are “the number of times a customer clicked on the call button on your Business Profile.”
Directions are adjusted for repeat taps and spam. Google also notes the data includes both organic results and Google Ads.
What should you pull for the profile?
Pull calls, directions and website clicks for six months and compare with last year. If profile calls fell while site visits held, the problem is local visibility, not the website. Google says local results rest mainly on relevance, distance and prominence, and that you can’t pay for a better local ranking.
You can’t change distance. Relevance and prominence respond to categories, services, reviews and links. Our post on proximity and the map pack explains why a competitor two miles closer can win calls you used to get.
Is call tracking missing real leads?
Calls are the leads most often lost to measurement. A tracking number replaces the number on your site, a vendor changes or a plugin updates, and recorded calls drop while real calls don’t.
Check three things:
- Does the number on the site match what the vendor expects, on mobile too?
- Is the call passed into GA4 as a key event?
- Does your Business Profile show the number you intend?
The third has a rule behind it. Google’s Business Profile guidelines require a number “under the direct control of the business” and warn against numbers that redirect or refer users. Its phone number help allows up to two additional numbers beyond the primary.
Our view: keep your real local number as the profile’s primary, and use tracking numbers on the website and in ads. Then ask whoever answers the phone for last month’s count.
Say they log 40 calls and your report shows 28. The report is missing 30 percent of calls (28 / 40 = 0.70), which points to tracking, not demand. Our call tracking guide covers setup.
Is some of the traffic bots or spam?
Automated visits can inflate sessions and dilute conversion rate without adding a lead. Imperva’s 2025 Bad Bot Report, covering 2024, found automated traffic was about half of internet traffic for the first time. It found bad bots at 37%, as Malwarebytes reported.
Watch out That figure is internet-wide, not your site. But it explains why sessions alone can’t be trusted.
Compare GA4 organic sessions with Search Console clicks for the same dates. They count different things and won’t match, but they should move the same way.
If sessions jump 40 percent and clicks barely move, check engagement. Warning signs are one landing page taking all the growth, one country or browser dominating, and sessions lasting zero seconds. Check form entries for spam too, since junk fills can bury real enquiries.
Did the conversion path break?
A page that used to bring calls can quietly stop working. Common causes are:
- A form that submits nowhere
- A phone number that is plain text on mobile
- A pop-up over the button
- A slow script
Traffic holds and leads fall.
Test it on a phone, on mobile data, the way a customer would. Open your top five landing pages from Search Console, tap the phone number, submit the form with a test entry, and confirm it arrives. Then repeat in a second browser.
Quick check Can you tap the phone number and submit the form with a test entry, and see it arrive, in two different browsers? If not, fix that first.
It takes twenty minutes and finds faults no report shows. A form that fails in one browser reads in analytics as lower interest, because the thank-you page is never reached.
Are fewer people clicking on the searches that matter?
Sometimes impressions hold and clicks fall on valuable queries. Google’s debugging guide says that when impressions stay the same but clicks drop, your titles and snippets may not be compelling enough. Competing sites may also be showing more attractive rich results.
Key takeaway If both fell, you have a real visibility problem.
How do AI features change the picture?
AI features add to the picture, and the evidence is mixed. Google says sites shown in AI Overviews and AI Mode are reported in the Performance report under the Web search type, with no separate report.
Ahrefs compared December 2023 with December 2025 across 300,000 keywords. It found AI Overviews correlate with a 58% lower average click-through rate for the top result. Ahrefs calls it a correlation, based on a forecast, desktop only.
For service businesses the effect depends on the query. Informational searches are more exposed than “emergency plumber near me.” Our posts on zero-click searches and falling CTR as visibility grows show how to read the shift without panicking.
Want to apply this to your site?
Not sure where to start? Get a plan for your own business.
What does this look like with real numbers?
This is a hypothetical, not a client result. A plumber in Phoenix gets 3,000 organic visits a month.
Of those, 2,600 land on service and location pages and convert to a call or form at 2.2 percent. The other 400 read blog posts and convert at 0.2 percent.
Before: 2,600 x 0.022 = 57.2 leads, plus 400 x 0.002 = 0.8. Total: 58 leads from 3,000 visits, a blended rate of 1.9 percent.
Definition A blended conversion rate is total leads divided by total visits, across every page type.
Six months later a post titled “why is my water heater making noise” ranks well, and blog visits grow to 2,200. A competitor with a stronger profile takes some map and service-page traffic, so service visits slip to 2,300.
After: 2,300 x 0.022 = 50.6 leads, plus 2,200 x 0.002 = 4.4. Total: 55 leads from 4,500 visits, a blended rate of 1.2 percent.
| Page type (hypothetical) | Visits before | Leads before | Visits after | Leads after |
|---|---|---|---|---|
| Service and location pages | 2,600 | 57.2 | 2,300 | 50.6 |
| Blog posts | 400 | 0.8 | 2,200 | 4.4 |
| Total | 3,000 | 58 | 4,500 | 55 |
Traffic is up 50 percent and leads are down about 5 percent. A dashboard showing only sessions calls this a win, while the person answering the phone feels the slowdown.
Read the example this way. The service pages kept converting at the same 2.2 percent. They simply got fewer visits.
The two charts below show the same story. The first shows visits by page type, and the second shows leads. All numbers are hypothetical.
Source: Hypothetical example
View the data as a table
| Item | Value |
|---|---|
| Service pages, before | 2,600 |
| Service pages, after | 2,300 |
| Blog posts, before | 400 |
| Blog posts, after | 2,200 |
Source: Hypothetical example
View the data as a table
| Item | Value |
|---|---|
| Before | 58 |
| After | 55 |
| After, if service visits held | 61.6 |
Split the change into its two causes. The 300 lost service visits cost 300 x 0.022 = 6.6 leads. The 1,800 extra blog visits added 1,800 x 0.002 = 3.6.
Net: 3.6 - 6.6 = -3.0 leads. Blog growth masked more than half the loss.
Had service visits held at 2,600, the same blog growth would have lifted leads to 61.6, about 6 percent above the start. So the blog was not the villain. The service-page slide was, and it only shows up once you segment.
Key takeaway Report leads per page type and query intent, not one blended number.
The SEO ROI calculator helps price a lead against traffic, and our post on SEO attribution explains why blended numbers mislead.
In what order should you fix things?
Measurement first, then the conversion path, then visibility, because improving visibility on bad data wastes money. This is our view, drawn from how we work through audits. It is not a Google rule.
These are the decision rules we apply, in this order:
- If recorded calls run 20 percent or more below what the phone owner reports, fix tracking before anything else.
- If a landing page fails the phone test, fix it the same day.
- If service-page clicks fell year over year while blog clicks rose, strengthen the service and location pages and keep the blog.
- If Business Profile calls fell but site visits did not, work the profile: categories, services, reviews, photos.
- If GA4 sessions and Search Console clicks diverge sharply, report on Search Console clicks until you find the cause.
- If everything checks out and Google Trends shows demand down market-wide, wait.
The website SEO checker flags basic page problems quickly. It doesn’t replace the manual checks above.
What mistakes make a traffic-up, leads-down report misleading?
These are patterns we flag in diagnostics. They are not statistics.
- Reporting on traffic alone. For a service business, traffic split by intent is informative and traffic in total is not. Our post on SEO for lead generation covers what to measure instead.
- Blaming the agency or the site when a phone vendor changed. Check your own side first.
- Cutting content that brings readers. A good post earns links, brand searches and trust. It isn’t meant to produce a call on the first visit.
- Treating all calls as equal. If new calls are wrong-service or out-of-area, volume is up and booked jobs are down.
- Comparing against last month instead of the same weeks last year.
- Mixing paid and organic. If you paused ads, calls may fall while organic traffic looks fine. Our post on leads from Google Ads, not SEO covers the mix-up.
What does the data not show?
Search Console can’t tell you which visitors became leads, and GA4 can’t tell you which calls came from the map pack. No single report ties the whole path together, which is why we triangulate.
- Search Console shows queries and clicks, not people.
- Business Profile data covers verified profiles only, and its search terms update monthly.
- Our worked example is hypothetical. The conversion rates are illustrative, and yours will differ.
- We can’t say how much AI Overviews affect your queries without your data. The Ahrefs figure is a correlation across a keyword sample.
- The bot share figures are internet-wide. They don’t tell you your site’s share.
What should you do this week?
Spend two hours:
- Pull Search Console for the last 90 days against the same period a year ago, split by page type.
- Pull Business Profile calls and directions for the same periods.
- Ask whoever answers the phone for last month’s count.
Then run the phone test on your top five landing pages and check your GA4 key events. Write down which of the six causes fits, and if several do, fix tracking first.
What does this mean for your business?
Do not judge your site by visits alone. Judge it by leads per page type and query intent, because a rise in blog traffic can hide a slide in service-page traffic.
Start with the checks that cost nothing. Compare the same weeks last year, define a lead in writing, and ask whoever answers the phone for last month’s count. Then fix measurement first, the conversion path second and visibility third.
If you want a second pair of eyes, our SEO service starts with this same diagnostic. The local SEO checker gives a quick read on your profile and site basics.
FAQ
Can website traffic go up while leads go down?
Yes, and it is common. Leads are visits times the share of visitors who want to buy times the share who can reach you and be counted. Visits can grow from readers or bots while buyers shrink, move to the map pack, or call a number your tracking never sees.
Is it a problem if my traffic grew but conversions stayed flat?
Not necessarily. If the new visits came from informational searches, flat conversions are expected. Those readers may still build brand searches later.
It becomes a problem when pages that used to bring calls lose visits. It is also a problem when blended conversion rate falls with no change in what people are searching.
How do I tell if the extra traffic is bots?
Look for near-zero engagement concentrated in one country, one browser or one landing page. Then compare GA4 sessions with Search Console clicks for the same dates.
Search Console counts clicks from Google results to your site. Sessions jumping while clicks barely move is a reason to investigate before celebrating.
Do calls from my Google listing show up in website analytics?
Not on their own. Taps on the call button in your Business Profile are counted in the profile’s performance data, not in your website reports.
Calls from the site need separate tracking. Add the two together before judging how many leads organic search is producing.
How long should I wait before deciding leads are really down?
We suggest comparing 8 to 12 weeks against the same weeks last year, not last month, because home-service demand swings with the seasons. Google also suggests allowing a few weeks after changes before judging them.
A single bad week is noise. Three months of decline in the same segment is a signal.
Should I stop writing blog posts if they bring traffic that does not convert?
No, but stop grading them on leads alone. Posts can earn links, build branded searches and answer questions buyers ask before they call. Grade them on assisted conversions and branded growth, and hold service and location pages, where buying intent sits, to the direct lead goal.
Sources
- Google Search Central, Debugging drops in Google Search traffic (accessed 22 Oct 2026)
- Google Search Console Help, Performance report (Search results) (accessed 22 Oct 2026)
- Google Search Console Help, Performance report metrics (accessed 22 Oct 2026)
- Search Engine Land, Google Search Console adds branded queries filter, November 2025 (accessed 22 Oct 2026)
- Google Analytics Help, About key events (accessed 22 Oct 2026)
- Google Business Profile Help, Understand your Business Profile performance and insights (accessed 22 Oct 2026)
- Google Business Profile Help, Improve your local ranking on Google (accessed 22 Oct 2026)
- Google Business Profile Help, Guidelines for representing your business on Google (accessed 22 Oct 2026)
- Google Business Profile Help, Add or edit phone numbers (accessed 22 Oct 2026)
- Google Search Central, AI features and your website (accessed 22 Oct 2026)
- Ahrefs, AI Overviews reduce clicks (update) (accessed 22 Oct 2026)
- Malwarebytes, Hi, robot: Half of all internet traffic now automated, April 2025, reporting Imperva’s 2025 Bad Bot Report (accessed 22 Oct 2026)
Find where your leads went with the website SEO checker, then Get Your Organic Growth Plan.
Want to apply this to your site?
Not sure where to start? Get a plan for your own business.
