Your Conversions Went Up And Your Diary Did Not

The dashboard says forty conversions. Best month on record.

Your diary says twelve genuine enquiries, of which maybe six were worth quoting for, and two of those were people who wanted something you do not sell.

Nobody has done anything wrong. The account is doing exactly what it was told.

This is the single most common complaint I hear about Google Ads, and it is almost never solved by the thing people reach for first, which is more filtering.

Lead quality is not a spam problem. It is a signal problem, with a spam problem sitting next to it.

Get those two confused, and you spend a month blocking bots while the real leak carries on.

Why more leads made things worse

Here is the part that feels backwards until you see it.

If your campaigns run on Maximise Conversions or Target CPA, they optimise towards whatever you defined as a conversion.

Suppose every form submission counts as one conversion, all equal. You have told the system that a form fill is the goal, and that all form fills are worth the same.

So it goes and finds you more form fills. Efficiently. At the lowest cost it can manage.

The cheapest ones are the worst ones

The cheapest form fills come from the least serious people. Somebody price-shopping with no budget. Somebody who wanted a different service. Somebody who fills in every form on page one of Google.

A serious buyer with a real job is more expensive to reach, because everybody else is bidding for them too.

So an account optimising for volume drifts, month by month, towards the cheap end. Your conversion count rises. Your cost per conversion falls. Everyone congratulates everyone.

And the diary stays empty.

The uncomfortable implication

Chasing more conversions, on an automated strategy, actively degrades lead quality over time.

Not through malice or a bad algorithm. Because you set the target and it hit it.

Which means the fix is not to filter harder at the end. It is to change what the system is aiming at.

The three kinds of bad lead

Before fixing anything, work out which problem you have. There are three, and they look identical in the dashboard while needing completely different fixes.

Applying the wrong fix is the reason people work on this for a month and change nothing.

One, bots and automated submissions

Nonsense names. Gibberish in the message field. Foreign phone numbers. Twelve submissions in ninety seconds.

These are a website problem, not an advertising one. No amount of keyword work will stop them.

Two real people who could never buy

They filled the form themselves. They are not spam. They are just wrong.

Somebody outside your service area. Somebody wanting a domestic job when you only do commercial. A student doing research. Somebody after a job rather than a quote.

This is a targeting and qualification problem.

Three: real buyers with the wrong intent

The hardest group, and the most valuable to fix.

These are genuine potential customers who arrived at the wrong moment or with the wrong expectation. Price shoppers when you are premium. People wanting an emergency callout when you book three weeks out. People expecting something your advert implied and your business does not do.

This is a message and expectation problem, and it usually starts in the ad copy.

How to tell which you have

Read the last twenty enquiries and put each into one of the three buckets.

That is the whole diagnostic, and it takes twenty minutes. The bucket with the most in it tells you where to spend your effort.

Most businesses assume bucket one and discover they are mostly in bucket two.

The feedback loop nobody builds

This is the real fix, and it is the one most small businesses never attempt because it sounds technical.

The idea is simple. Tell the account which enquiries were actually good, and let the bidding optimise towards those instead of towards raw volume.

The mechanism is offline conversion import, or its close relative, enhanced conversions for leads. You send the outcome back to Google, matched to the click that produced it.

What changes when you do

Instead of learning that a form fill is worth one point, the system learns that this kind of person, from this search, on this device, in this area, produced something you were pleased with.

Over time, it shifts spend towards those and away from the ones that produced nothing.

The traffic cleans itself, because the algorithm is finally being scored on the right outcome.

You cannot filter your way to good leads. You can only teach the system what a good lead looks like.

Why so few businesses do it

Because the guides describe it as a CRM integration, and most small businesses do not have a CRM.

That framing puts people off something they could genuinely do with a spreadsheet.

Building the loop without a CRM

You do not need enterprise software. You need a habit and a spreadsheet.

Here is the version that works for a business doing twenty to fifty enquiries a month.

• Make sure your form or call tracking captures the identifier the import needs, so an enquiry can be matched back to the click that produced it. This is the one bit worth asking a developer about, and it is a small job.
• Export your enquiries monthly into a simple sheet. One row per enquiry, with the date and the identifier.
• Add one column. Was this worth having, yes or no. Not whether it became a sale. Whether it was a real prospect you were glad to hear from.
• Optionally add a second column with a rough value, even a banded one. Small, medium, large. Value beats a yes or no if you can manage it.
• Upload the qualified ones back into the account as offline conversions on a monthly rhythm.
• Then set the campaign to optimise towards that qualified conversion action rather than towards raw form fills.

That last step is the one that matters, and it is the one people forget. Importing the data changes nothing if the bidding is still aiming at the old signal.

The honest caveats

It takes time. The system needs a reasonable volume of qualified conversions before it can learn anything, so a business with three enquiries a month will struggle.

And your reported conversion numbers will fall dramatically when you switch the target, because you have gone from counting every form fill to counting only the good ones.

That is the number becoming honest. It is also the point at which some businesses lose their nerve and switch back.

Write down your before figures so you can hold your ground.

The spam fix, which is a website job

If your twenty-enquiry review put most of them in bucket one, this is your afternoon.

Automated submissions are handled on your website, not in your advertising account.

The standard defences are unglamorous and effective. A honeypot, which is a hidden field a human never sees, and a bot fills in. Server-side validation, so the checks cannot be bypassed. Rate limiting, so one source cannot submit repeatedly. And a privacy-friendly challenge that only appears when behaviour looks suspicious.

Most decent form plugins support some of this already, and a developer can add the rest in an hour or two.

Why it matters beyond your inbox

Every spam submission you counted as a conversion also taught your bidding something false.

So clearing them out improves the signal as well as your morning. Those two benefits compound.

What it will not fix

Blocking bots does nothing for buckets two and three, which is where most businesses actually live.

If you install spam protection and your lead quality does not improve, that is not a failure. It means you had a different problem, and now you can see it clearly.

Fixing wrong-fit people

Bucket two. Real humans who could never buy from you.

Three levers, in order of how much they usually move.

The geographic one

If people outside your area are enquiring, that is a location settings problem before it is anything else.

The default location option serves people who merely searched about your area from elsewhere. Our separate post on location targeting covers the fix, and it is often a single dropdown.

This is frequently the largest single source of wrong-fit enquiries for a local business.

The search terms one

Open your search terms report and read what people actually typed.

Domestic searches when you only serve business. Repairs when you only sell new. Job searches, if your trade name doubles as a job title.

Each of those becomes a negative keyword. This is slow, unglamorous work and it is reliably worth doing.

The qualification one

Say who you are for, on the page and in the advert.

“Commercial contracts only, minimum twelve months” filters more effectively than any spam tool. So does a minimum project size, or a clear service area named in the copy.

Businesses resist this because it feels like turning work away. You are turning away work you were never going to win, before it costs you an hour on the phone.

The form length trade-off, honestly

Every guide tells you to shorten your form. That advice is right for some businesses and wrong for others, and almost nobody says so.

Here is the actual trade.

A short form gets you more enquiries and a lower proportion worth having. A longer form with a qualifying question gets you fewer enquiries and a higher proportion worth having.

Neither is better in the abstract. They solve opposite problems.

Which one you need

If you are drowning in unsuitable enquiries and your team is spending its day on the phone to people who will never buy, add a qualifying field. Budget range, project size, timescale, or simply what the job is.

If your phone barely rings, cut the form back to a name, a contact and a sentence. You cannot afford friction.

The mistake is applying the advice that suits a busy business to a quiet one, or the reverse.

The field that earns its place

For most service businesses the single most useful qualifying question is a plain description of the job.

It filters gently, because a serious enquirer will happily type a sentence and a tyre-kicker often will not. And it saves you a phone call either way, because you can judge fit before you ring back.

Fixing wrong intent, which starts in the advert

Bucket three. Genuine buyers who turned up expecting something else.

This one is the hardest to spot, because the enquiry looks perfectly reasonable until the conversation starts.

Where the mismatch is created

Almost always in the advert copy, and occasionally on the page.

An advert promising “affordable” attracts people shopping on price. If you are the premium option, you have just bought a conversation you cannot win.

An advert saying “fast response” attracts emergencies. If you book three weeks out, those enquiries waste everybody’s time.

Nothing in either advert was untrue. They simply set an expectation the business does not meet.

How to find it

Read your own adverts as a stranger, then read the last ten enquiries that went nowhere.

Ask what the enquirer seemed to be expecting. Nine times out of ten you can trace it to a specific phrase in the copy.

The fix is uncomfortable

Say the qualifying thing out loud in the advert.

“Contracts from twelve months”. “Projects from five thousand”. “Booking three weeks ahead”. “Commercial only”.

Your click-through rate will fall. That is the point. You are paying for fewer clicks from people who were going to waste the call.

Businesses resist this harder than anything else in this article, because a lower click-through rate feels like failure. It is the cheapest filter available, and it works before you have paid for the click rather than after.

The number that replaces cost per lead

Cost per lead is the metric that got you here.

It rewards volume, because every lead counts the same. Drive the number down by finding cheaper enquiries, and it looks like progress right up until you notice the diary.

Cost per qualified lead is the honest version. Spend divided by the enquiries genuinely worth having.

How to work it out

You already have both numbers if you did the twenty-enquiry review.

Take last month’s spend. Count the enquiries you were glad to receive. Divide.

That is the figure to report on, to compare month against month, and to judge every change by.

What it does to decisions

It changes what a good month looks like, and it changes what you do next.

A month with fewer enquiries at a lower cost per qualified lead is a better month, and cost per lead would have told you the opposite.

It also settles arguments about budget. If your cost per qualified lead holds when you increase spend, growth is working. If it rises sharply, you have found the ceiling of your market at that price.

The inventory question worth asking

One more source of poor leads, and it is easy to check once you know to look.

Not all of your traffic comes from Google’s own search results. Some arrives from search partner sites, which are other sites showing Google search results.

That inventory can behave quite differently from Google Search proper.

How to check it

Segment your performance by network and compare the qualified lead rate on Google Search against search partners.

You need your own qualified judgement for this, not the platform’s conversion count, so it works best once you have started the monthly review described earlier.

What to do with the answer

If the two are broadly similar, leave it alone and stop worrying about it.

If partners produce noticeably worse leads at meaningful volume, you have two options. Remove that inventory from your core campaigns, or isolate it into a separate test where it cannot pollute what your main campaigns are learning from.

That second point matters more than people realise. If a chunk of your conversions come from inventory that produces nothing, your bidding is learning from that too, and it will keep looking for more of it.

Why isolate rather than just switch off

Occasionally partners perform perfectly well, and switching them off costs you volume for no reason.

Isolating gives you the answer without the guesswork, and it keeps the learning in your main campaigns clean while you find out.

A worked example, start to finish

A hypothetical to pull it together. A commercial cleaning company on eleven hundred pounds a month, running Maximise Conversions.

The account reports sixty-one conversions last month at about eighteen pounds each. On paper, excellent.

The owner reads the last twenty enquiries and sorts them into the three buckets.

Four are obvious bots. Nonsense names, gibberish messages.

Eleven are real people who could never buy. Domestic cleaning requests, which they do not do, and several enquiries from towns two hours away.

Three are genuine but wrong-intent. People wanting one-off cleans when the business only takes contracts.

Two were good.

What they change, in order

Bucket two is the biggest, so that is where the effort goes first, not the bots.

The location option moves from presence or interest to presence, and two distant regions get excluded.

The search terms report gives them thirty negatives covering domestic, one-off, and cleaning jobs as employment.

The advert copy and the landing page both gain a line saying commercial contracts, twelve months minimum. The form gains one field asking what the job is.

Then the bots. The developer adds a honeypot and server-side validation, which takes an hour.

Then the loop

They start a monthly spreadsheet. Every enquiry, marked as worth having or not, is uploaded back as an offline conversion.

After two months, there is enough qualified data to switch the bidding target from all form fills to qualified leads only.

Where it lands

Reported conversions fall from sixty-one to about fourteen. Cost per conversion rises from eighteen pounds to nearly eighty.

On the dashboard, that is a catastrophe.

In reality, the fourteen are nearly all worth having, whereas the sixty-one contained two. The cost per qualified lead went from roughly five hundred and fifty pounds to about eighty.

The business is unrecognisably better, and the report looks far worse. That gap is the whole reason this is hard to do.

What to expect while it is working

The awkward middle of this process is worth describing, because it is where most attempts get abandoned.

Everything gets worse on paper before it gets better in reality.

Month one

You add negatives, tighten locations, add a qualifying field and block the bots.

Conversions fall immediately. Cost per conversion rises. Nothing yet proves you were right, because you have removed volume without having built the loop.

This is the month people panic.

Month two

You are uploading qualified leads back, but the bidding is still aiming at the old signal, because there is not enough qualified data yet.

Performance looks flat. The dashboard has stopped falling and has not started improving.

This is the month people conclude it did not work.

Month three onwards

You switch the target to the qualified conversion action. Reported conversions drop again, sharply, because you are now counting only the good ones.

And then the useful thing starts. Spend drifts towards the searches and audiences that produced real opportunities, because that is finally what the system is scored on.

How to hold your nerve

Record the before numbers, in writing, at the start. Spend, total enquiries, enquiries worth having, and cost per qualified lead.

Then judge every month against that last figure and nothing else.

If cost per qualified lead is falling, it is working, whatever the conversion count is doing. If it is rising after month three, something in the setup needs revisiting rather than abandoning.

The mistakes that waste the most money

• Treating every poor lead as spam, when most are real people who were never a fit.
• Optimising bidding towards all form fills, which reliably finds you the cheapest and worst ones.
• Installing spam protection first when the twenty-enquiry review points at targeting.
• Importing qualified leads back into the account but never switching the bidding target to them.
• Shortening the form because an article said to, while drowning in unsuitable enquiries.
• Judging the change on cost per lead, which will always say you made things worse.
• Losing your nerve when reported conversions collapse, and reverting to the flattering signal.
• Refusing to say who you are not for, and paying for the phone calls that follow.

Frequently asked questions

Where to start this week

Read your last twenty enquiries and sort them into the three buckets. Bots, wrong fit, wrong intent.

Twenty minutes, no tools, and it will tell you which of the three problems you actually have. Most businesses guess wrong before they do this.

Then work the biggest bucket first, rather than the one that feels most urgent.

Then work out one number. Last month’s spend divided by the enquiries you were genuinely glad to receive. That is your cost per qualified lead, and from now on it is the figure that decides whether anything you change is working.

One thing to sort before any of it. If your conversion tracking is not honest, none of this is measurable, and the feedback loop will teach the system something false with great efficiency.

The businesses that get out of this trap are not the ones with the best filters. They are the ones who stopped asking for more leads and started telling the system which ones counted.

Written by Uzair Kharawala

Partner at SF Digital Studios, a Google Premier Partner agency. Uzair has been helping businesses grow with Google Ads since 2002 and teaches it every week on YouTube.