Define waste before finding it
Wasted spend is not every click that fails to convert. Advertising buys uncertain opportunities. A useful definition is cost that had no reasonable path to the agreed business outcome, or cost whose expected value was below the company’s threshold.
That includes irrelevant queries, unavailable products, duplicate conversion signals, low-quality leads presented as successes, accidental geographies and spend that continued after decisive negative evidence. It does not automatically include an expensive but plausible test.
Build a spend ledger
Export cost and business outcomes at campaign, query theme, product, placement, location, device and landing-page level. Join lead campaigns to CRM stages and ecommerce to margin/refund data. Label rows as profitable, uncertain, structurally invalid or commercially unqualified.
Suppose a hypothetical theme spends £1,500 and records 30 leads. Google Ads CPA is £50. If only three qualify and none close, the relevant metric is £500 per qualified lead, not £50 per form. The spend is not proven waste until enough lag has passed, but it should not be scaled on the raw-lead number.
Audit in dependency order
- Measurement: duplicates, missing values, wrong primary actions and attribution changes.
- Eligibility: stock, locations, schedules, policies and broken final URLs.
- Demand: search terms, placements, product groups and customer type.
- Experience: ad promise, page, form, checkout and mobile errors.
- Economics: margin, lead quality, close rate, returns and repeat value.
Search demand and negatives
Use Google’s search terms report and theme-level insights. Privacy thresholds mean the report is incomplete. Classify intent rather than adding negatives from a simplistic “spent without conversion” filter.
Add a negative when the meaning is clearly incompatible with the offer. Keep a rollback log with scope and match type. A campaign-level negative can be correct while the same term remains valuable elsewhere.
Waste hidden by good averages
- Brand demand masking weak acquisition.
- Returning customers counted as new-customer growth.
- High-revenue products with low contribution.
- Cheap leads rejected by sales.
- Performance Max values inflated by duplicate purchases.
- High-performing campaigns capped while weak campaigns keep budget.
Segment before cutting. The objective is not a dramatic “waste percentage”; it is reallocating the next pound toward a better expected outcome.
Use a decision rule that includes uncertainty
For each segment record spend, mature outcomes, expected CPA/value range, conversion delay and sample size. Pause immediately when intent or fulfilment is impossible. For plausible but weak segments, set a pre-agreed evidence limit based on allowable CPA. Continue when results remain uncertain and the learning value justifies the cost.
A useful recommendation states evidence and mechanism: “Pause Product Group X because 42 advertised items were out of stock for 18 days and received £730 cost” is stronger than “Shopping wastes budget.”
Waste patterns by campaign type
Search
Separate irrelevant meaning from relevant demand that failed to convert. Review query themes, match route, network, location and final URL. Add negatives with documented scope. Search Partners should be evaluated as a segment, not included or excluded by folklore.
Shopping and Performance Max
Join item spend to availability, margin, price competitiveness, returns and new-customer contribution. Check products accidentally included through labels or listing groups, final URL expansion and exclusions. The feed optimisation guide covers the product-data branch.
Display, YouTube and Demand Gen
Inspect placements, exclusions, geography, frequency, audience expansion and attribution. A placement without direct conversions is not automatically waste in an awareness test, but its role and measurement must be declared before spend.
Measurement errors create allocation waste
Duplicate purchases, placeholder lead values and micro-conversions marked primary make bad spend look efficient. Sample IDs and compare actions with the system of record. Suppose duplicate events inflate revenue from £40,000 to £68,000 on £10,000 spend. Reported ROAS is 680%; reconciled ROAS is 400%. Not all spend is waste, but allocation used a materially false objective.
Budget allocation waste
Compare marginal opportunity. A profitable campaign may be capped while an unproven campaign consumes a fixed allocation. Conversely, moving everything to the best historic ROAS can saturate it and starve acquisition learning.
Create columns for budget, mature CPA or ROAS, marginal estimate, available demand, confidence and strategic role. Reallocate in increments and track the next mature cohort.
Creative and destination waste
A misleading ad can earn high CTR and low-value sessions. A correct ad can land on an error, unavailable item or generic homepage. Follow highest-cost paths from promise to final URL, mobile QA and conversion step. Assign ownership accurately: targeting, feed, ecommerce, development, pricing or sales.
Maintain a waste register
| Field | Purpose |
|---|---|
| Finding and segment | Defines where loss occurs |
| Spend exposure | Quantifies materiality |
| Evidence confidence | Separates fact from hypothesis |
| Action and rollback | Makes the fix controlled |
| Owner and date | Prevents orphan findings |
| Validation metric | Shows whether loss fell |
Delete resolved rows only after validation; otherwise exclusions, feed errors or tracking failures tend to return.
Quantify without false precision
Report confirmed invalid, mature under-threshold and unresolved spend separately. Confirmed invalid includes wrong geography, broken destinations or unavailable stock. Mature under-threshold has enough lag and outcomes to fail economics. Unresolved remains plausible but lacks evidence.
If £2,000 is confirmed invalid, £5,000 is mature under-threshold and £8,000 unresolved, do not announce £15,000 “wasted”. The first is defensible, the second depends on the agreed rule, and the third is still a research budget.
Negative-keyword implementation
- Group unwanted queries by meaning.
- Decide account, list, campaign or ad-group scope.
- Choose match type and check valuable variants.
- Record examples, spend and reason.
- Apply a controlled batch.
- Review blocked volume and themes.
A single-word broad negative can block more demand than intended. Keep a versioned list and rollback date.
Worked product example
A hypothetical Shopping campaign reports 500% ROAS: £18,000 revenue on £3,600 spend. One discounted product contributes £9,000 revenue but only £450 before ads and absorbs £1,200 spend. The campaign headline is healthy while that product is contribution-negative. Use a margin label, listing-group control or corrected value; do not pause the whole campaign.
Governance prevents recurring waste
Assign owners for search demand, feed and stock, conversion diagnostics, pacing and CRM quality. Use change history and an issue register. Review auto-applied changes under a documented policy rather than accepting or rejecting them wholesale. Connect this routine to the full audit after material change.
Validate after cutting spend
Track excluded or paused segments, total qualified outcomes, contribution, brand/non-brand mix and impression availability. Apparent efficiency can rise simply because volume fell. The desired result is lower invalid cost without losing more valuable demand than expected.
Review sales and search-term feedback for blocked good enquiries. If the action was reversible, restore or narrow it when evidence contradicts the hypothesis. Waste reduction is a portfolio decision, not a competition to produce the longest negative list.
Lead-generation waste
Join each lead to campaign, query theme, timestamp, contactability, qualification, opportunity and outcome. Separate spam, duplicates, existing customers, job seekers, wrong services and no-budget prospects. Each category suggests a different fix: targeting, form validation, customer exclusion, ad copy, service-area control or sales follow-up.
Hypothetically, 100 forms at £40 appear efficient. Twenty are spam, thirty are outside area, thirty are unreachable, fifteen qualify and three close. Raw CPA is £40, qualified CPA £266.67 and customer acquisition cost £1,333. The correct “waste” estimate depends on customer contribution and whether unreachable leads were followed up properly.
Preventive controls
- Feed and stock alerts for advertised products.
- Conversion-count and value anomaly alerts.
- CRM quality report by campaign and cohort.
- URL and checkout monitoring.
- Budget and location change approvals.
- Versioned negatives and final URL exclusions.
- Named owners and incident response.
Controls are valuable because they catch waste before a retrospective audit turns it into a dramatic percentage.
A 90-minute first pass
- Minutes 0–15: verify primary conversion actions and a sample value.
- 15–35: sort campaigns and product/query themes by cost and downstream value.
- 35–50: inspect invalid geography, network, placement, stock and URLs.
- 50–65: review query meaning and negative scope.
- 65–75: compare budget allocation and marginal opportunity.
- 75–90: classify findings as confirmed, mature under-threshold or unresolved.
Do not implement every finding during the audit. Preserve exports, quantify exposure and sequence dependencies. Measurement and fulfilment faults come before bid tuning.
Reporting example
“We removed 30% waste” is not auditable. A stronger statement is: “£1,840 in the prior 30 days came from countries the business cannot serve; location settings were corrected on 11 July. No qualified leads were recorded from that segment. We will validate geographic lead volume and total qualified CPA after the normal 21-day lag.”
This wording identifies evidence, action, limitation and validation. It also avoids pretending that every non-converting click was preventable.
Repeat the review after normal conversion lag. Some “savings” will reduce volume; others will reveal better allocation. Report both so waste reduction does not become a vanity metric of its own.