The twelve checks I run before touching anything

Cost per lead audit card showing the number twelve, the checks run before changing anything
Contents 9 sections

Thirty days of writing about one metric comes down to a fairly short list. This is the cost per lead audit I run before changing anything in an account, in the order I run it, and the order is the important part.

Measurement first, because a counting problem wearing the costume of a performance problem will waste your entire quarter. Bidding last, because it is the smallest lever and everyone reaches for it first.

This is for you if you want to work out what is actually wrong before you start fixing.

Why a cost per lead audit starts with measurement

Almost everybody starts in the wrong place, and it is an expensive habit.

The instinct is to open the campaign view, find the worst performer, and start adjusting. That assumes the numbers describing it are true. In a large share of the accounts I open, they are not.

A tag that fires twice halves your reported figure. A tag that broke doubles it. Spam lowers it. Untracked phone calls raise it. Any one of those turns a healthy account into a problem on paper, or hides a real problem behind a flattering number.

So the first three checks are not about performance at all. They establish whether there is anything to fix, and they take about an hour.

I have twice finished a cost per lead audit at check three, told the client the advertising was fine, and watched the relief in the room. Both times the money had been going out for months on a rebuild nobody needed.

Checks 1 to 3: is the number even real

Nothing else means anything until these pass.

One. Count real enquiries by hand for a week and compare against what the platform reports. A large or growing gap means you have a measurement problem, not a performance problem.

Two. Check for double-counted conversions. A tag firing on both the form and the thank-you page halves your reported cost per lead and makes everything downstream look better than it is.

Three. Check for junk. Pull the last 200 submissions and mark the spam. Every fake one lowers your reported figure and teaches the platform to find more of whatever produced it. Lead form spam covers the defences.

Checks 4 to 6: what the platform is aiming at

This is where I find the most damage, and it is almost always invisible in reports.

Four. Read the optimisation event on every ad set and campaign, in the settings rather than from memory. Page views and link clicks are common and wrong for lead generation. The wrong event is the commonest leak I see.

Five. Check whether outcome data flows back. If your CRM does not send qualified or closed-won data to the ad platform, the system is optimising towards form fillers by default. Offline conversion tracking is the fix and it outranks everything below it.

Six. Compare platform-reported conversions against your own enquiry count as a ratio. Write it down monthly. A stable ratio is workable. A ratio that jumps means something broke, and attribution explains why the two never match exactly.

Checks 7 to 9: who is arriving and what they see

Seven. Read the search terms report by hand, thirty days, sorted by cost. On Google this is the fastest waste-finder there is, and reading it properly takes half an hour.

Eight. Check the landing page. Does the headline repeat the search that brought them there, how many form fields are there, and how fast does it load on a phone on mobile data? Conversion rate is the only lever with no auction attached to it.

Nine. Look at creative age on Meta. If the top ad has run unchanged for months, that is your likeliest cause, and a weekly testing rhythm is the fix.

Checks 10 to 12: structure, then bidding

Ten. Check conversion volume per campaign. Anything under roughly 30 to 50 conversions a month is starving its bidding, and consolidating usually beats spreading the same budget wider.

Eleven. Check exclusions. Existing customers, brand where it should be excluded, geography matching where you actually operate. Cheap to fix and frequently wrong.

Twelve. Only now, look at the bidding strategy. It is the last check for a reason. Bidding decides what you pay, not whether the search was worth having or whether the page converts.

What every cost per lead audit is measured against

All twelve checks are pointless without a ceiling to judge against.

Profit per sale multiplied by close rate gives you what a lead is worth. Your working target sits below that, usually around half. Without it you are comparing your figure to a benchmark built from other people’s economics.

For scale rather than as a target, WordStream’s 2026 study of 13,474 US search campaigns put the all-industry average at $66.69, with cost per lead falling for the first time in five years as conversion rate rose for 87% of industries.

That last detail is the whole argument of this list. The market lowered its costs by converting better, not by bidding harder.

Fix one thing at a time

The discipline matters more than the speed.

Change one item, leave two weeks where a learning phase is involved, and write down what you expect to happen before you look. Fixing six things in a week guarantees you will never learn which one mattered.

Expect some fixes to make the headline figure worse. Correcting the optimisation event, filtering spam and feeding back closed-won data all remove cheap junk from the count, so the number rises while the business improves. Tell whoever reads the report before you start.

And do not rebuild the account because performance dipped for four days.

When the cost per lead audit says nothing is wrong

It happens, and it is a useful answer rather than a wasted afternoon.

If measurement is clean, the event is right, the page converts and the search terms are relevant, then your constraint is upstream of the ad account. Usually the offer, the price, the close rate, or a market that cannot support what you are paying.

Advertising cannot fix a business that does not make enough money per sale. Knowing that after a day of checking is far cheaper than discovering it after two more quarters of optimisation.

What to change this week

Four steps.

Work through checks one to three, since they take an hour and decide whether anything else on this list is even relevant.

Write down your ceiling from profit and close rate.

Pick the single highest item that failed and fix only that.

Then put a recurring half hour in the calendar for search terms and negatives, because drift is what makes good accounts slowly expensive.

The wider piece on judging lead cost is the foundation for all of it, the order I use to reduce lead cost covers the fixing rather than the finding, and lead quality covers the measurement that should replace cost per lead in most of your reporting. If you are weighing whether to bring this in-house or hire out, the freelance and agency trade-off is honest about both sides. If you would rather I ran these twelve checks on your account, that is exactly what a free teardown is.

Frequently asked questions

What does a cost per lead audit actually check?

Measurement first, then the offer, then creative and pages, then structure, with bidding last. The order matters more than the list, because a measurement problem disguised as a performance problem will send you optimising something that was never broken.

How long does a cost per lead audit take?

A useful first pass is a few hours across the account, the landing pages and the enquiry records. The slow part is never the ad platform, it is getting hold of which enquiries became customers, because that data usually lives somewhere nobody has connected.

What is the most common problem you find?

Campaigns optimising towards the wrong event. Usually a page view or a form fill when what the business needs is qualified enquiries. It is almost always because somebody set it once, years ago, and nothing in the interface flags it as wrong.

Can I run a cost per lead audit myself?

Yes, and the twelve checks below are the order to do it in. The parts that need experience are judging whether a search term is genuinely irrelevant and deciding what to do with what you find. Collecting the facts is mostly mechanical.

Should I fix everything at once?

No. Change one thing at a time, in the order below, and leave two weeks between anything that resets a learning phase. Fixing six things in one week means you will never know which one worked, and you will repeat the useless five forever.

What if the audit says my account is fine?

Then your problem is upstream: the offer, the price, the close rate, or the market. That is a genuinely useful finding, because it stops you spending another quarter optimising an account that was never the constraint.

How often should this be done?

A full pass twice a year, plus a monthly half hour on search terms and negatives. Accounts do not fail suddenly. They drift, and by the time anyone notices the cause is a hundred small things rather than one obvious break.

What is the single highest-value check?

Counting your real enquiries by hand for a week and comparing that against the dashboard. It costs nothing, it takes an hour, and it decides whether you have a performance problem at all or just a counting problem.

Want a second pair of eyes on your account?

Book a free teardown

Book a free teardown

Send me your worst-performing campaign.

I will tell you what is wrong on the first call. No obligation, no hard sell, and I will say honestly if I am not the right fit.

Book a free 20-minute account teardown

Replies within one business day, Lahore time.