PPC audit and restructure

Why the audit starts with tracking

We check conversion tracking first because every other judgement depends on it. If the numbers are wrong, then the campaign that looks best may be the worst, and every optimisation made on that data has pushed the account further in the wrong direction. Auditing structure before verifying tracking is auditing fiction.

The check is practical rather than theoretical. We complete a conversion ourselves and confirm it records once, with the right value, in the right account. We compare platform conversion counts against the CRM or order system for the same period and explain the gap. We look at what is set as a primary conversion, because bidding follows that setting and it is frequently set to something that is not a sale.

The faults we find most often are consistent. Duplicate tracking firing the same conversion twice. Thank-you page conversions counting on refresh. Micro-conversions like page views and button clicks set as primary, so bidding is optimising towards people clicking rather than buying. Tags missing from the pages that matter. Consent settings quietly blocking measurement.

• Conversions counted more than once, inflating reported return
• Micro-actions set as primary, so bidding chases the wrong outcome
• Platform numbers that never reconciled with the back-end
• Tags missing, misfiring or blocked by consent configuration

Structure and settings review

Next we look at whether the account is built in a way that can be managed. That means checking whether brand and non-brand are separated, whether campaigns are split by the things you need to control separately, and whether ad groups have enough volume for automated bidding to learn from.

Settings carry more waste than most people expect, because they are set once at launch and never revisited. We check network settings, since search partners and display expansion left on inside search campaigns is one of the most reliable sources of hidden spend. We check location targeting, where the presence or interest option quietly buys clicks from people in other countries. We check ad rotation, device modifiers, ad schedules and language settings.

We also look for structural contradictions. Campaigns competing for the same keywords. Audiences overlapping so ad sets bid against each other. Budget caps limiting the one campaign that actually performs while an underperformer runs uncapped. These are cheap to fix and often make the largest immediate difference.

Where the money is actually going

The waste analysis is the part clients find most uncomfortable and most useful. We pull the search term report for the full period and sort by spend with zero conversions. In most accounts a meaningful share of budget is sitting in terms that have never produced anything and were never negated.

We look at the same question from several angles. Which keywords have spent more than several times the target cost per acquisition without converting. Which placements on display and video have consumed budget, since automatically placed campaigns often accumulate spend on mobile apps and low-quality sites. Which products in a shopping feed absorb spend without selling. Which geographic areas cost more than they return.

Then we quantify it. A finding is only useful with a number attached, so each item comes with the spend involved and what recovering it would mean. That turns the audit from a list of opinions into a ranked set of decisions with money against each one.

• Search terms with spend and no conversions, by budget share
• Keywords running far past target cost per acquisition
• App and site placements absorbing display and video budget
• Products, regions and devices returning less than they cost

Bidding, budget allocation and pacing

We review whether the bidding strategy suits the data available. Automated bidding on an account with a handful of conversions a month has nothing to learn from and behaves erratically. Manual bidding on a mature account with strong conversion volume is leaving performance on the table. Both are common, and both are usually the result of a setting nobody revisited.

Targets get checked against reality. A target cost per acquisition set well below what the account has ever achieved will simply throttle delivery, and the campaign sits there spending nothing while everyone wonders why volume dropped. A target set too loosely buys volume at a loss. We show what the account has actually delivered and where the target should sit.

Budget allocation is the other half. We look at how spend is distributed against where the return is, and how often campaigns hit their daily caps. Accounts frequently have their best performer capped and their weakest running free, which is the single easiest thing to correct on the day the audit lands.

Ads, assets and the landing page

On the creative side we check coverage before quality. Are there enough active ads per ad group to test against each other. Are asset ratings showing gaps. Are extensions and assets in place, because sitelinks, callouts and structured snippets increase click-through rate and take an hour to add. Missing assets are the most common quick win in a search account.

Then relevance. Does the ad match what the person searched, and does the landing page match the ad. A generic homepage receiving traffic from specific service queries costs you twice, once in a lower Quality Score and again in a lower conversion rate.

Landing pages get assessed on the things that decide outcomes rather than on design taste. Load speed on a mobile connection. Whether the offer is visible without scrolling. Form length and how many fields are genuinely required. Trust signals near the point of action. Whether the page works properly on a phone, which is where most of the clicks arrive.

• Ad and asset coverage, including missing extensions
• Query to ad to landing page message match
• Mobile load speed and above-the-fold clarity
• Form length and friction at the point of conversion

What you receive and how findings are ranked

The deliverable is a written report in plain language, ordered by impact rather than by the order we found things. Each finding states what is wrong, what it is costing, what to do about it, and roughly how long the fix takes. Screenshots and data extracts back up the significant items so nothing has to be taken on trust.

Ranking is by expected value, which usually puts the boring items at the top. Tracking corrections come first because they change every subsequent decision. Then settings and waste, because those recover money immediately at little cost. Then structural work, which takes longer and pays back over weeks. Then creative and landing page improvements, which compound but need time and testing.

We also include what we would not change. Audits that recommend rebuilding everything are usually selling the rebuild. If a campaign is performing, it stays, and we say so. That distinction is what makes the rest of the report credible.

When a rebuild beats a repair

Most accounts should be fixed rather than rebuilt, because a rebuild discards the conversion history that automated bidding depends on and resets the learning period. That cost is real and it is usually underestimated by whoever is proposing the rebuild.

A rebuild becomes the right call when the existing structure prevents the fixes. Campaigns mixing unrelated goals into one budget. Years of accumulated settings that conflict with each other. Ad groups so fragmented that no campaign has enough volume to bid intelligently. Tracking that has never been correct, which means the historical data being preserved is not worth preserving.

When we do rebuild, we do it alongside rather than instead. The new structure runs in parallel while the old one continues, budget shifts across as the new campaigns prove themselves, and the old campaigns pause once the numbers justify it. That takes longer than switching everything over on a Monday morning, and it avoids the fortnight of lost revenue that the faster approach usually produces.

• Default to repair, a rebuild costs conversion history and learning time
• Rebuild when the structure itself blocks the fixes
• Run new and old in parallel and shift budget on evidence
• Keep the old campaigns available until the new ones have proved out