Rule out measurement before you rule out performance
Start here because it is fast and because it is the explanation that most often turns out to be right. A rise in cost per lead is a ratio, and ratios move when either half moves. If the count of leads changed for a reason that has nothing to do with demand, the ratio moves and nothing real happened.
What to check, in order
- Did anyone change the website — a new form, a new thank-you page, a new plugin, a consent banner?
- Did the definition of a conversion change, so that something previously counted no longer is?
- Did a duplicate stop firing, or start firing?
- Did phone enquiries stop being recorded, or start being recorded twice?
- Did a spam filter start working, quietly removing junk that used to inflate the count?
A measurement change and a performance change look identical in a report. One of them takes an hour to rule out.
The last one on that list catches people out constantly. Stopping the junk is an improvement, and it makes every cost-per-lead figure look worse overnight. If the count fell on a single day and stayed down, that is a settings change somewhere, not a market. Confirm what your site actually records as a conversion and when the definition last moved before you go any further.
Tag configuration is where most of these live, and one misfiring tag is enough to invert a whole quarter's reporting — the mechanics are set out in the briefing on the single tag that breaks your data.
Check the window before you check the account
Second cheapest check. Most reported increases are comparisons between two windows that were never comparable. A fortnight against a month. A period with a public holiday against one without. A run of days that happened to contain one large, cheap enquiry against one that did not.
Match the window length, match the days of the week, and make the window long enough that a couple of enquiries either way cannot move the answer. If the increase disappears when you do that, there was no increase.
If it survives a fair comparison, you now have a genuine change to explain and the remaining three causes are worth your time.
Did the market get dearer, or did you get less competitive?
These feel the same from inside the account and they have different fixes. Costs in the auction move with what other businesses are prepared to pay and with how relevant your ad and landing page are to what was searched. Google's own documentation is explicit that there is no rate card.
A market that got dearer looks like higher costs across everything you run, arriving over weeks rather than overnight, with your own conversion rate roughly unchanged. Somebody with money has entered your category, or an existing competitor has decided to grow.
Getting less competitive looks different: costs rise unevenly, concentrated where your offer is weakest, and often alongside a falling share of the impressions available to you. That is a message and landing-page problem, and it is the half you can actually do something about.
You cannot lower what a competitor bids. You can be worth more per click by converting a higher share of the same visits, which is the only lever that works in both scenarios.
A measurement change and a performance change look identical in a report. One of them takes an hour to rule out.
Is the demand itself different, or just your share of it?
Some categories swing hard with the seasons and with weather. Heating, cooling, roof repairs, storm damage, moving, end-of-lease cleaning — all of them have periods where the people searching are fewer and less urgent. When urgency drops, the same spend buys enquiries that take longer to close and cost more each.
Google Trends is free and settles this in five minutes. Look at interest in your category in your state across the last few years. If the curve dips at the same point every year and you are in the dip, the account is behaving normally and the correct response is to expect it and plan for it, not to rebuild anything.
The trap is treating a predictable seasonal dip as a failure and making structural changes during it. You then get the recovery you were going to get anyway and credit it to the changes.
Your site may be converting fewer of the same people
Cost per lead is cost per click divided by the share of clicks that enquire. If the second number falls, the first does not need to move at all for your costs to rise.
What usually moved
| Change | Effect |
|---|---|
| The page got slower | People leave before it renders, and mobile suffers first |
| A form field was added | Every extra field costs completions, especially on a phone |
| The phone number moved or stopped being tappable | Urgent enquiries quietly vanish |
| A consent banner now covers the call button | The most valuable action becomes the hardest one |
| A price or a policy changed on the page | Same traffic, different decision |
Open your own site on your own phone, on mobile data rather than office wi-fi, and try to enquire. Most of the causes on that list are obvious within thirty seconds and invisible from a desktop. If the page is the constraint, that is a website problem wearing an advertising costume.
Cost per lead can rise while the business gets healthier. That happens every time the counting gets stricter.
What to change first, and what to leave alone
Change one thing at a time, and pick the one with the shortest feedback loop. In practice that means fixing measurement first because it is fast and it makes every later decision trustworthy, then fixing the landing experience because it improves both channels at once, and only then touching what the account is doing.
Leave the budget alone while you investigate. Cutting it during a genuine seasonal dip guarantees you miss the recovery, and raising it to chase a number that is rising for measurement reasons pours money into a report artefact.
Leave the campaign running unless the spend itself is the emergency. Pausing destroys the before-and-after comparison you need to prove what fixed it, which means the next time this happens you will start from nothing again.
Finally, write down what you found. The five causes recur, and a business that has diagnosed this once should be able to do it in an afternoon the second time. If tracking turns out to be the culprit more than once, the problem is not the tags — it is that nobody owns the measurement setup.
Common questions
Is a rising cost per lead always bad?
No. If the counting became stricter — spam submissions stopped being recorded, or a duplicate stopped firing — cost per lead rises while the business gets healthier. Ask what changed in the definition before treating the rise as a performance problem.
How long a window should I compare?
Long enough that a couple of enquiries either way cannot swing it, matched to the same length and the same days of the week. Comparing a fortnight against a month, or a period containing a public holiday against one that does not, produces a difference that means nothing.
My competitors increased their budgets. Can I do anything?
You cannot lower what they bid. You can be worth more per click by converting a higher share of the same visits, and you can concentrate spend where you are strongest rather than defending every part of the market at once.
Should I pause everything while I investigate?
Rarely. Pausing destroys the comparison you need and restarts any learning the account has done. Investigate with the campaign running unless the spend itself is the emergency.
Cost per lead is up but revenue is up too. Which do I believe?
Revenue. Cost per lead is a proxy for the thing you care about, and proxies drift. If booked work and margin are both improving, a rise in the proxy usually means the mix shifted toward larger jobs that cost more to win and are worth more when won.