The first 90 days: read the numbers and make them better

Building a campaign is a project with an end. Running one is a habit, and the habit that most new advertisers fall into — opening the account every morning to see what happened — is the one that does the most damage. Not because attention is bad, but because the numbers you are looking at in the morning are not finished yet.

Google reports a conversion against the date of the click, not the day the conversion happened. Its own example: "if your ad was clicked on last week and that traffic converted this week, both the click and the conversion are reported back to last week." On top of that, reporting lags by hours. So the last few days of any report you open are understated by design, and they will quietly improve over the following week without you doing anything.

That cadence is not laziness. It is the only way to get a readable result out of a small account, because every change you make restarts the clock on the thing you were trying to measure — a mechanism we come back to when we get to bidding.

If you only do one thing each week, do this one. The search terms report shows the actual phrases people typed before your ad appeared, which is different from the keywords you chose — your keywords are what you bid on, the search terms are what you bought. It is the only report where a single decision has an immediate and permanent effect on spend.

1. 1

   Set the date range to the last 7 days

   You are looking for new waste, not re-reading terms you already judged. A rolling week keeps the list short enough to actually read.
2. 2

   Read down the search term column first, before any metric

   The question is not which term cost the most. It is: would I want to pay for this person? A term can look fine on cost and still be someone who will never hire you.
3. 3

   Select everything that is not a potential customer

   The classic categories for a local trade: job seekers ("plumber salary", "hvac apprenticeship"), people looking to do it themselves ("how to fix", "diy"), price researchers with no location intent, students and courses, wholesale and supply searches, and other towns you do not serve.
4. 4

   Add them as negative keywords at campaign level

   Google offers a choice of where the exclusion lands — the ad group, the campaign, or a shared negative keyword list. With one campaign, campaign level or a shared list is right. Ad-group-level negatives are a precision tool for later, and using them early creates exclusions you will forget you made.
5. 5

   Add the singular and the plural, every time

   Negative keywords match literally. Excluding "job" does not exclude "jobs". Misspellings are the one exception and are handled for you, so do not waste an evening building a misspelling list.

A share of your search terms will never appear in that report. Google filters it by a privacy threshold — the report shows terms "that a significant number of people have used" — which means terms searched by too few people are withheld no matter how much of your budget they consumed. This is not a setting, not a bug, and not something an agency can unlock for you.

The practical consequence is that the report is a sample, not a ledger. You cannot add up the clicks in it and reconcile them against your total spend, and you will drive yourself mad trying. Estimates of how much is hidden vary enormously between accounts, so treat any specific percentage you read on the subject with suspicion.

There is a partial answer, and most small advertisers never find it. Separately from the report, Google generates search categories on the Insights page, and those categories are built from every search term — including the ones withheld from the report. Google’s wording is that they "take all search terms into account, including those not exposed in the search terms report due to privacy reasons." Terms with low spend or no identifiable category get grouped as "other search terms".

This is the section that separates this lesson from every "20 Google Ads optimization tips" article you will find alongside it. Those articles are written for accounts with thousands of conversions a month, and applied to an account with eight, their advice is not merely useless — it is actively misleading, because it invites you to act on numbers that are moving for reasons that have nothing to do with your decisions.

A local business running a few hundred clicks and somewhere between five and fifteen conversions a month is in a specific statistical situation: one extra lead can swing a keyword’s cost per conversion by half. That is not a signal. That is a coin landing heads twice.

The right column is not junk. Those are real reports that will matter later. But acting on them now means pausing a keyword that would have produced three leads next month, or setting a device bid adjustment on the strength of four clicks. The discipline is to leave them alone until the volume arrives.

And the number that matters most is not in Google at all. Google can tell you a call lasted ninety seconds. It cannot tell you the caller was inside your service area, wanted the job you actually do, or ever booked. Cost per lead comes from the dashboard; cost per customer comes from your phone and whatever you use to track jobs. If you only ever build one habit from this series, make it writing down where each new customer came from.

Sometime in your first month, Google will show you an optimization score with a number under 100% and a list of things you could do about it. The pull is real: it looks like a grade, and nobody enjoys a bad grade on their own account.

Google defines it precisely: "Optimization score is an estimate of how well your Google Ads account is set to perform. Scores run from 0-100%, with 100% meaning that your account can perform at its full potential." Read the words carefully — it is an estimate of how the account is set, which is to say configured, measured against Google’s preferences.

And then there is the sentence that settles the whole argument, which is also Google’s own: "Your account can reach an optimization score of 100% by applying or dismissing all recommendations." Dismissing counts the same as applying. A perfect score is available to anyone willing to clear the queue, which tells you exactly what the number is measuring — and it is not results.

One more thing worth knowing: Google states that optimization score "is not used by your Quality Score." And when you look at what Google does list as the components of Ad Rank — bid, ad and landing page quality, the Ad Rank thresholds, auction competitiveness, context, and expected asset impact — the optimization score is not among them. Nothing about this number reaches the auction.

Lesson 3 launched the campaign on Maximize clicks with a bid cap, because a campaign published that afternoon has no conversion history for anything smarter to run on. The question this lesson answers is when that stops being true.

The trigger is volume, not the calendar. Google’s own evaluation standard for conversion bidding is the cleanest threshold available: "For evaluation, we recommend you measure performance for the last 30 days, including at least 30 conversions." Google also says plainly that you may start with no history at all. Both are true, and together they define the gap: you can switch whenever you like, but you cannot judge the switch until you are near that bar.

Google publishes a status for the bid strategy itself, and it is the diagnostic almost nobody opens. "Active" means "The bid strategy is active and setting bids to optimize performance." "Limited" means something is constraining it — often that keywords are Limited by budget. And there are two misconfigured states, one of which is the single most important status in this lesson.

When you do change the strategy, the campaign enters Learning, which Google describes as showing "minor performance fluctuations as Google Ads optimizes your bids." Hovering over it tells you which of three things caused it: a new strategy, a setting change, or a composition change — campaigns, ad groups or keywords added or removed.

That third trigger — composition change — is why the cadence at the top of this lesson matters mechanically and not just temperamentally. Adding keywords on Tuesday and reorganising an ad group on Thursday restarts the recalibration both times. Batch your changes into one session, then leave the campaign alone long enough to read what they did.

At some point in month one you will open the account and find a day that spent nearly double the daily budget, and conclude something is broken. Nothing is broken. Google documents two limits: "Your daily spending limit (two times your average daily budget for most campaigns) on any particular day" and "Your monthly spending limit (30.4 times your average daily budget for most campaigns) in any particular month." The average is the contract; the individual day is not.

The opposite problem is more common and more serious: a campaign spending nothing. That is almost never a budget issue. Work down this list before touching the budget.

1. Are the ads approved? A disapproved ad shows no impressions and spends nothing.
2. Is the bid cap so low the campaign cannot enter auctions? Google notes that if CPC bid limits are too low, the strategy may not be able to achieve your goal.
3. Is the targeting too narrow — a radius too small, or a location list of one small town?
4. Are the keywords too restrictive? An exact-match-only list on low-volume local terms can genuinely run out of searches.
5. Is the conversion action still recording? A tag that stopped firing will not stop spend, but it will make everything above it unreadable.

And when the campaign is spending its budget and producing leads at a price you are happy with, "Limited by budget" stops being a warning and becomes a business decision: the market has more demand than you are buying. Raise the budget when the leads are worth more than they cost you — not because Google labelled the campaign limited.

A course that ends with "and now you can run this forever" would be selling you something. Plenty of local businesses should run their own Google Ads indefinitely — a single campaign, a good negative list, twenty minutes a week — and get better results than they would from a cheap agency that puts them on autopilot.

The signals that it is time to hand it over are not signals of failure. They are these:

• The weekly review stops happening. Three missed weeks in a row is a leak with no one watching it, which costs more than management would.
• The account has reached readable volume — around Google’s thirty conversions in thirty days — and the next gains need someone who does this daily.
• Your time is worth more than the spend. At $600 a month in ads, an hour a week of an owner’s time is a large share of the total cost of the channel.
• The campaign is working and you want to add a second service, a second city, or a second channel — that is a structural job, not a weekly one.

If you take nothing else from these four lessons, take the part nobody can outsource: know what a customer is worth to you, and know where your customers actually came from. Every decision in this lesson gets easy once those two numbers exist, and no amount of dashboard-reading substitutes for them.