Problem. Action. Result.

01 Problem

Media spend had roughly doubled over four quarters while the number of sales-accepted opportunities stayed flat. The dashboards told a happier story than the CRM: the ad platforms claimed thousands of conversions, but sales could not find them.

• Every form submission counted as a conversion, including support requests and job applications.
• No connection existed between ad platforms and the CRM, so nothing was optimised toward revenue.
• Twenty-three campaigns competed for the same queries, splitting signal and inflating cost per click.
• A single generic demo request served nine distinct buyer situations.

02 Action

Measurement first, media second. Nothing in the accounts changed until the numbers could be trusted.

1. Rebuilt the conversion definition. A lead only counted once sales accepted it. Offline conversion imports pushed accepted and closed-won stages back into the ad platforms daily.
2. Cut the account down. Twenty-three campaigns became six, organised by buying stage rather than by internal team structure.
3. Rebuilt the offer. The generic demo request was replaced by three situation-specific entry points, each with its own landing page and follow-up path.
4. Set a floor, not just a target. Every campaign got a cost-per-accepted-opportunity ceiling agreed with finance, and anything above it was paused rather than defended.
5. Ran a geo-holdout. Two comparable markets were held back for six weeks to measure how much of the branded demand was genuinely incremental.

03 Result

[+X%]

Sales-accepted pipeline

[-X%]

Cost per opportunity

[X.X×]

Return on media spend

Total media spend ended the period [-X%] lower than when we started. The gain came from removing waste, not from adding budget — which is why the improvement held after the initial rebuild.

The geo-holdout also produced an uncomfortable finding we reported anyway: a share of branded search spend was buying demand that would have arrived regardless. That budget was moved to non-brand acquisition.

[CLIENT QUOTE]

01 Problem

Revenue was growing and the business was losing money on every incremental order. Platform-reported return looked acceptable in each channel; blended return across all spend had fallen below break-even and nobody was watching that number weekly.

• Channels double-counted the same purchases, so the sum of reported revenue exceeded actual revenue.
• Budget decisions used gross revenue, ignoring discounting, shipping, returns and payment fees.
• Around two fifths of the product catalogue was disapproved or missing required feed attributes.
• Three ad creatives had been live for eleven months, with frequency climbing and click-through falling.
• Retargeting absorbed a large share of budget while claiming credit for repeat buyers.

02 Action

1. Changed the steering metric. Budget moved to contribution margin after discounts, shipping, returns and fees, tracked as a single blended figure reviewed every Monday.
2. Fixed the feed. Attribute coverage, variant structure and disapprovals cleared, then monitored automatically so it stayed fixed.
3. Built a creative system. Six named angles derived from support tickets and product reviews, produced as a monthly batch across static, motion and short-form video.
4. Rebalanced toward new customers. Retargeting budget was capped and the remainder moved to prospecting, with new-customer revenue reported separately from total revenue.
5. Deduplicated measurement. Server-side tracking with a single source of truth for orders, reconciled monthly against the store's own financial reports.

03 Result

[+X%]

Blended ROAS

[+X%]

New-customer revenue

[-X%]

Cost per acquisition

Reported revenue fell in month one, because it stopped being counted twice. Actual contribution margin rose in the same period. That conversation is why we insist on agreeing the steering metric before the first campaign changes.

Creative volume went from three live assets to [X] per month, and the two best-performing angles came from customer support transcripts rather than from a brand brief.

[CLIENT QUOTE]

01 Problem

Organic sessions had grown steadily for two years while organic revenue had not moved. The site ranked well for a large volume of informational queries and almost nothing with commercial intent. Growth depended entirely on paid, and paid costs were rising.

• The blog attracted readers who were years away from buying anything.
• Commercial category and comparison pages were thin, unlinked, and in several cases not indexed.
• Faceted navigation generated a large volume of near-duplicate URLs, wasting crawl budget.
• Templates rendered core content client-side, so a meaningful share of pages were indexed nearly empty.
• Success was reported as sessions and keyword count, so nobody noticed revenue standing still.

02 Action

1. Re-mapped demand to intent. Every target query was grouped by buying stage and assigned an expected revenue value, so priorities followed money rather than search volume.
2. Fixed rendering and crawl waste. Core content moved server-side, facet parameters were controlled, and the crawl budget was redirected to pages that could actually earn.
3. Rebuilt the commercial layer. Category, use-case and comparison templates were rewritten with real substance and linked properly from the existing high-traffic articles.
4. Templated the long tail. A programmatic set of location and service-combination pages was built from structured data already in the product database, with thin variations deliberately excluded.
5. Changed the reporting. Organic performance was reported by intent group and by revenue, and the old sessions-and-keywords dashboard was retired.

03 Result

[+X%]

Non-brand organic revenue

[+X%]

Commercial-intent sessions

[-X%]

Blended acquisition cost

Total organic sessions grew more slowly than in the preceding two years, because several thousand low-intent pages were consolidated or removed. Revenue per organic session rose [+X%] over the same period.

With organic carrying a larger share of acquisition, paid budget was reduced by [-X%] without a fall in total new business — which is what "compounding" is supposed to mean in practice.

[CLIENT QUOTE]