A logo is one file. A brand is the rules around it.
Most small businesses buy a logo and then improvise everything else. The result drifts within a year. What stops the drift is a small set of written rules, and here is what goes in them.
When a Google Ads account underperforms, the instinct is to change the bids, the budgets or the keywords. We understand why. Those are the controls on the dashboard. But in an account running on automated bidding, the most important control is not on the dashboard at all. It is the definition of a conversion.
Target CPA, Target ROAS and Maximise Conversions all do the same basic thing: they learn which auctions tend to produce the outcome you are counting, and they bid more in those auctions. If the outcome you are counting is wrong, the system learns the wrong lesson, and it learns it efficiently.
So before we change a bid, we check what the account believes a conversion is.
The four ways the signal usually goes wrong
Counting things that are not business outcomes. Page views of the contact page. Clicks on a phone number that may or may not have been dialled. Scroll depth. These are sometimes set as primary conversions because they fire often, which makes the account look healthy. The bidding then chases people who look at contact pages, which is a different group from people who get in touch.
Counting the same thing twice. A form fill tracked by the Google Ads tag and also imported from Google Analytics 4. A purchase recorded on the confirmation page and again on a refresh. Duplicates inflate performance and tell the system that some clicks are worth twice what they are.
Missing conversions that did happen. A thank-you page that stopped loading the tag after a site update. Forms that submit through a script and never reach a confirmation URL. Phone calls from the website with no call tracking at all. When real outcomes go uncounted, campaigns that work look like they do not, and the budget moves away from them.
Losing signal to consent. In the UK, visitors can refuse analytics and advertising cookies, and many do. Without consent mode set up properly, those visitors simply vanish from measurement. With it, Google can model some of the missing conversions. Either way, the setup should be deliberate, documented and tested, not whatever the cookie banner plugin did by default.
How we audit it
This is the order we work in, and it usually takes the first two weeks of an engagement.
1. List every conversion action and ask what it means. For each action in the account: what triggers it, where the data comes from (the Ads tag, a GA4 import, an offline upload), whether it is primary or secondary, what value it carries, and how it is counted, once per click or every time. Most accounts have at least one action nobody can explain.
2. Decide what should drive bidding. Normally that is one or two primary actions that represent real enquiries or real revenue. Everything else is useful to observe and should be set as secondary, so it is reported but does not steer the bidding.
3. Test every primary action by hand. We submit the form, make the purchase in a test mode, place the call, and watch the conversion arrive with Tag Assistant and in the account's diagnostics. Anything that does not arrive, or arrives twice, gets fixed before anything else happens.
4. Check consent and enhanced conversions. We confirm consent mode is passing the right states, and that enhanced conversions are enabled where they are appropriate, so a hashed email address from the form helps match conversions that cookies alone would lose. That hashing happens on your site before anything is sent, and it should be covered in your privacy policy.
5. Put values on things. If one enquiry type is worth far more than another, the account should know. Even rough relative values, agreed with whoever handles the sales, are better than treating every lead as identical.
6. Close the loop with offline data where it exists. For businesses that qualify leads by phone or in a CRM, uploading which leads became customers lets bidding learn from outcomes, not form fills. This is the single most useful change for many lead-generation accounts, and also the one most often skipped.
What changes afterwards
Once measurement is trustworthy, two things usually happen.
First, the reported numbers get worse. Removing duplicates and soft conversions makes the account look less productive than it did. This is uncomfortable and completely expected. The old numbers were never real.
Second, the account gets easier to improve, because every later decision is made on information you can rely on. Search term reviews, landing page tests and budget shifts all depend on knowing which clicks actually led somewhere.
Automated bidding usually needs time to adjust after the conversion setup changes, so we plan for a settling period and agree in advance how we will judge it. We would rather tell you that up front than explain an unexpected dip afterwards.
The short version
If you take one thing from this: before you change a bid, open the Conversions page, look at every primary action, and ask whether each one is something you would happily pay for. If the answer is no for any of them, that is where to start.
If you would like a second opinion, give us read-only access to the account. We will come back with the tracking and structure problems we can see, written plainly, and we will tell you if the account is in good shape.
