The Google Ads Search Account Restructure Playbook

Most accounts that need restructuring did not start broken. They grew. A campaign was added for a launch, another for a promotion, a third because someone wanted to test a budget. Two years later the same keyword competes with itself across three campaigns, ad groups hold forty unrelated keywords, and nobody can say which campaign is responsible for the leads. A restructure fixes that by making the account legible: every campaign has one job, one budget, and one definition of success.

It is also a disruptive project. In our experience, changing structure can reset what the system has learned about a campaign, and results often wobble for a few weeks. That is a reason to be selective. If the account converts at an acceptable cost, the conversion tracking is sound, and the main complaint is that it looks messy, tidy the labels and the naming and leave the structure alone. Restructure when you cannot attribute performance to a decision, when budget is trapped in campaigns that cannot spend it, or when one bidding strategy is being asked to serve goals that need different targets.

Run the restructure as a sequence, not a rebuild-and-switch weekend. Build the new campaigns alongside the old ones, move traffic over in stages, and pause the legacy campaigns only after the replacements have been running long enough to compare. The method below is a recommendation drawn from how Google Ads is documented to work, not a guarantee of a particular outcome. Your account, margins, and search volume will set the pace.

Start with a list of the things a buyer is trying to do when they search, written in plain language. For a services business that is usually something like: find a provider now (urgent, high intent), compare providers (research), learn how something works (early), and find us by name (brand). Each of those deserves its own campaign because they need different budgets, different bid targets, and different landing pages. A person typing a brand name and a person typing a general question will not convert at the same rate, and blending them into one campaign hides that.

Inside each campaign, build ad groups around a tight theme so the ad can echo the search and the landing page can answer it. The test is whether one headline and one page genuinely fit every keyword in the group. If you find yourself writing a vague ad to cover everything, split the group. Google Ads documents match types (broad, phrase, exact) and how each one widens or narrows the queries a keyword can trigger, so choose them deliberately per campaign rather than defaulting everything to one setting. A common and defensible pattern is to give high-intent, high-cost themes tighter control and let research themes run wider with stricter negatives.

Keep the number of campaigns proportional to the data you have. Every extra campaign divides budget and conversions, and automated bidding needs conversion volume per campaign to learn from. A small account with thirty conversions a month across eight campaigns is starving every one of them. Fewer campaigns with clear intent boundaries almost always beat many thin ones.

Conversion tracking is the foundation, and it is the step teams skip because it is dull. Smart Bidding strategies set bids using the conversion data in the account, so if a page-view is counted as a conversion, or the same lead is counted twice, the system will chase the wrong behaviour efficiently. Open the conversion actions list and, for each action, record what it counts, how it fires, whether it is set as a primary action used for bidding, and what it is worth.

The questions that find real problems are mundane. Does a form submission fire once or on every page refresh? Are phone calls from ads counted, and above what call length? Is a "thank you" page reachable by direct URL, which would let non-leads trigger it? Are soft actions such as scrolling or a button click set as primary actions alongside real leads, which blends low-value and high-value events into one number? Google Ads Help describes enhanced conversions, which supplements tag data with hashed first-party customer data to improve measurement accuracy, and offline conversion imports, which let you send back the outcome of a lead after it closes. For lead generation, the import is often the more important of the two, because a form fill and a sale are different things.

Decide which actions are primary, meaning they feed bidding, and which are secondary, meaning they are reported but not optimized toward. Give each primary action a value that reflects the business, even if it is a rough, clearly labelled estimate to start with. Then check the numbers: compare the lead count in Google Ads with the lead count in your CRM for the same period. They will not match exactly, but a large and unexplained gap means something is mis-firing, and it should be fixed before Smart Bidding is given more control.

Negative keywords stop your ads from showing on searches you have decided are not worth paying for. They are the cheapest control in the account, and they only work if they are maintained. The source of truth is the search terms report, which lists the queries that actually triggered your ads. Reviewing it on a fixed schedule, weekly while the account is being rebuilt and monthly after, is how irrelevant traffic gets caught. Reviewing it once is not enough, because new queries appear continuously, particularly under broader match types.

Organize negatives so they can be managed. Account-level or shared lists hold terms that are never wanted anywhere, such as job-seeker words ("jobs", "salary", "careers"), free-resource words if you sell a paid service, and your own competitors if you have chosen not to bid on them. Campaign-level negatives do the structural work of keeping intent buckets apart: your learning campaign should carry the buying-intent words you want routed to the ready-to-buy campaign, so the two do not bid against each other for the same search. Google Ads Help documents that negative keyword match types behave differently from positive ones, and notably that negatives do not match close variants, so a negative must be added in the forms you actually want blocked.

Be cautious about adding a negative from a single data point. A query that cost money and did not convert in a small sample may simply be unlucky. Record the reason for each negative addition in the change log, use phrase or exact negatives where a broad one would block good queries, and revisit the list when you change the offer. A sloppy negative can quietly cut off the best traffic in the account.

Smart Bidding refers to the automated bidding strategies in Google Ads that use machine learning to set bids per auction, aimed at a conversion or conversion value goal. They can work well, and they are only as good as their inputs. Google Ads Help describes them as optimizing for conversions or conversion value, which means they depend on your conversion tracking, so the readiness question is about your data rather than about whether the technology is good. Before moving a campaign onto a target-based strategy, check four things: the primary conversion action is clean and reconciles to the CRM; there is a steady flow of conversions into that campaign rather than a trickle; the campaign is not budget-limited in a way that stops it from learning; and the target you set is derived from your economics, not copied from a benchmark.

Google does not publish a single universal minimum that applies to every account, and you should be wary of anyone who quotes one as a fact. Treat conversion volume as a judgement call: a campaign with only a handful of conversions a month is a poor candidate for a strict target and is better served by a less constrained strategy until volume builds. When you do switch, expect a learning period during which performance may be unstable, avoid editing the campaign during it, and set a review date in advance so you judge the change on enough data rather than on day three.

Guardrails limit the damage if something goes wrong. Set daily budgets that you can afford to lose at the highest daily rate the account could spend, remembering that, in our experience, spend on a given day can run above the daily budget; check Google Ads Help for how budgets are applied and the monthly limit. Add a spend alert in your reporting rather than relying on memory, define a stop-loss rule in writing (for example, pause and review if cost per lead exceeds an agreed multiple for a stated number of days), and keep a human owner for every budget change. Automation moves quickly, and the guardrail is what keeps a bad week from becoming an expensive month.

The sequence below treats each step as a gate. Do not start a step until the gate on the previous one is met. Every change goes into a log with the date, what changed, why, who did it, and what you expect to see. Google Ads keeps its own change history, but it records what changed and not why, so the reason has to live in your log. Without it, a dip in leads three weeks later is impossible to diagnose.

Use the 30-60-90 frame to keep the project from sprawling. In the first 30 days: audit, fix tracking, and build the new structure alongside the old. In days 31 to 60: shift budget to the new campaigns in stages, tighten negatives from real search term data, and begin Smart Bidding on the campaigns that pass the readiness checks. In days 61 to 90: retire the legacy campaigns, set the ongoing review rhythm, and write down what you learned. The attached worksheet carries the log template.