Maddie Lightening has spent the last twelve years turning advertising budgets into measurable growth for travel, retail and B2B brands. As head of paid media at UK agency Hallam, she now oversees search, social, programmatic, digital-out-of-home and account-based marketing campaigns that run across five time-zones. When we caught up with her for the WP in EU podcast, she walked us through three turning-points that reshaped how she thinks about performance data, account structure and AI adoption.
The GBP vs AUD mix-up that doubled ROAS overnight
Early in her career, Lightening inherited an Australian AdWords account whose invoices were billed in Australian dollars. The client, however, was British and wanted weekly reports in pounds. Instead of asking whether conversion values were already converted, she simply pulled the raw revenue numbers and divided by spend—creating the illusion of mediocre returns.
“We were telling the client we were hitting a 2.8 ROAS when the real figure was 5.6,” she recalls. “The fix took ten minutes once we compared the platform with CRM data, but for six weeks we had been pausing keywords that looked unprofitable and leaving money on the table.”
The lesson: always check the currency setting in both the billing profile and the conversion value column. Even if your dashboard auto-converts spend, revenue figures may stay in the account’s original currency, silently warping every efficiency metric you share.
Why 5,000 micro-campaigns broke Smart Bidding
In 2023, Hallam won a European travel brand whose paid search program had grown organically since 2016. Every destination, device and match-type combination had spawned its own campaign, ballooning the total to just under 5,000. The setup had worked when exact-match keywords and manual CPC were king, but it collided head-on with Google’s AI-driven Smart Bidding.
“The algorithms need volume to learn,” Lightening explains. “When you split 1,200 monthly conversions into 5,000 buckets, each campaign gets 0.24 conversions. Smart Bidding simply couldn’t exit learning mode.” Performance dipped 18% year-on-year and the client threatened to cut spend.
Her team proposed a radical consolidation: collapse the account into 80 geo-based campaigns, move keywords into broad match and let broad-match audiences feed the bid engine. Yet the travel vertical has a short high season—July to September—and any structural change risked volatility during peak bookings.
Timing beats perfection: why they moved in January
Lightening originally planned the migration for October, after the revenue spike. But when January numbers came in 24% below forecast, the client demanded immediate action. “We had two weeks to restructure, test and stabilise before Valentine’s Day short breaks went on sale,” she says.
The team used a traffic-light system:
- Green campaigns—high volume, low seasonality—were merged first.
- Amber campaigns—medium ROAS but peak-season sensitive—were moved in tranches with bid caps.
- Red campaigns—brand terms and flagship routes—were left untouched until the new structure proved itself.
By mid-February, consolidated campaigns were delivering 21% more revenue at the same cost, and Smart Bidding had enough data to raise bids on high-value routes automatically. “Waiting for the perfect window would have cost us another quarter of decline,” she notes. “Sometimes you launch and iterate.”
AI mistakes that taught her to keep humans in the loop
Lightening is bullish on AI, but only when guardrails are coded in. Last autumn, Hallam tested Performance Max for a luxury luggage retailer. Within a week, Google’s algorithm identified “cheap suitcases” as a high-volume query and started serving generic luggage-set creatives, eroding margin.
“The CPA looked amazing—down 35%—but average order value dropped 42%. Net profit fell even though revenue rose,” she says. The team fixed the issue by:
- Feeding only high-margin SKUs into the merchant centre feed.
- Adding “cheap,” “discount,” and “clearance” as negative keywords at account level.
- Creating a profit-based custom column so that Smart Bidding optimised toward gross margin, not revenue.
Two weeks later, profit was up 19% and CPA stayed lower than before. “AI will chase the signal you give it. If that signal is flawed, scale just makes the mistake bigger,” she warns.
Key takeaways for European advertisers
Lightening left us with three principles that now guide every account she audits:
- Audit currency and attribution first—before you touch bids, make sure revenue, spend and conversion windows are aligned.
- Consolidate early, but stage the risk—use a traffic-light rollout so peak season isn’t disrupted.
Leave a Comment