Finding why TakenPlace Weddings' sales fell off a cliff — in 4 days

The boring problem

Sales fell off a cliff. TakenPlace Weddings — a QR photo collection service for weddings — went from twelve orders in a normal three-week period to one order in the next 21 days. A 92% drop, with no obvious reason.

The owner suspected something technical. But diagnosing it meant digging through Google Analytics, WooCommerce, Google Search Console, and mobile performance data — across multiple tools, none of which talk to each other clearly. The kind of investigation that takes days of focused work, or a consultant billing by the hour.

What I built

An AI agent tasked with a specific question: why did sales collapse, and what needs to change?

Over four days, the agent worked through:

• Google Analytics 4 — traffic patterns, user behaviour, where people dropped off
• WooCommerce — sales data, conversion funnels, cart abandonment
• Google Search Console — rankings, impressions, click-through rates
• Mobile performance — page speed, load times, checkout experience on phone

Three problems came back, each one quietly killing conversions:

1. 40% of traffic was bots. Not customers — automated crawlers inflating the visitor numbers and making the analytics unreadable. Real traffic was a fraction of what it looked like.

2. Mobile pages took 16.3 seconds to load. Three out of four people trying to check out on a phone were abandoning before the page finished loading.

3. The best SEO page wasn’t converting. One page was appearing in search results 6,445 times a month but getting only 10 clicks. The title and description weren’t doing the job.

What changed

Metric	Before	After
Bot traffic	40% of all visits	Down 88%
Quality traffic (real potential customers)	14% of visits	38% of visits
Mobile load time	16.3 seconds	Significantly reduced
Key search ranking	Position 9.8	Position 6.3
Hosting costs saved	—	~€150/month

The analysis took four days. A human consultant working through the same data sources manually would have taken weeks — and would have billed accordingly.

The agent didn’t just produce a report. It produced a prioritised action list: fix the mobile speed first (highest revenue impact), then the bot filtering, then the SEO page. The owner had a clear sequence to work through, not a pile of findings with no direction.

The same agent now runs continuously, monitoring the same metrics and alerting the team if costs start rising or traffic quality drops again. The investigation is over; the watching never stops.

---

This kind of diagnostic work — mapping what’s broken before recommending what to build — is at the heart of the AI business assessment service. The barbershop scheduling and review automation referenced in that page is another example of workflow automation applied to a simple, repeating problem.