If your landing page converts at 2% and a competitor's converts at 6%, they can pay three times your click price and still win. No amount of bid tuning closes that gap.

This is why we almost always start off-platform. Doubling landing page conversion halves cost-per-lead instantly, and unlike bidding changes it doesn't decay.

The highest-leverage fixes are boring: message match between ad and page, one clear action, proof near the claim, and a form short enough to fill on a phone at a bus stop.

Pull your search terms report and read it properly. Not the keywords you chose — the actual queries that triggered your ads. In most accounts we audit, somewhere between a fifth and a third of spend has gone to searches that were never going to convert.

'Free', 'jobs', 'salary', 'DIY', 'courses', competitor brand names, and the endless long tail of people researching rather than buying. Each one is cheap on its own, which is exactly why it goes unnoticed.

Build the negative keyword list as a habit, not a one-off. Fifteen minutes a week beats a quarterly cleanup.

Someone searching 'emergency plumber near me' and someone searching 'how much does a boiler cost' are at completely different points. Sending both to the same page wastes the first and confuses the second.

Intent should shape the whole chain: the keyword, the ad copy, the page, and the action you ask for. High intent gets a direct ask. Research intent gets something lower-commitment — a guide, a calculator, a price range.

Trying to force research traffic into a 'book now' button is a reliable way to burn budget while concluding that the channel doesn't work.

Modern ad platforms optimise toward whatever you tell them is a conversion. Tell them a form fill is a conversion and they will find you people who fill in forms, whether or not those people ever become clients.

This is how accounts end up with a beautiful cost-per-lead and a sales team complaining about junk. The metric improved; the business didn't.

Pass qualified leads back into the platform rather than raw submissions. Once it optimises toward leads that actually became opportunities, cost-per-lead often rises slightly and cost-per-client falls — which is the number that pays wages.

With the page converting, waste cut, intent matched and clean signals going back in, bidding changes finally do something predictable.

Shift budget toward what converts, not toward what gets clicks. Give campaigns enough volume to actually learn instead of splitting spend across a dozen thin ad groups. Change one thing at a time and wait for enough data before judging it.

Done in this order, bid tuning is a refinement. Done first, it's guesswork.

Every lever above compounds into the ones after it. Optimising bids against a broken landing page just finds a cheaper way to lose. Feeding the algorithm signals from unqualified leads teaches it to bring you more of them.

The accounts that get genuinely cheap leads are rarely the ones with clever bidding. They're the ones where the offer, the page and the tracking were sorted out first — and then the bidding was left alone long enough to work.