480 demo bookings for MyPulse through Meta Ads

The first campaign used one ad set and produced one lead at $107. We split it into three ad sets based on product capabilities, each with its own budget. The revised test produced 21 leads at $33.17 each. A lookalike audience that delivered two leads at $49 each in a week was switched off.

In February we introduced new static ads and copy. The strongest audience focused on e-commerce specialists and produced 38 leads at $37.82 each. Lookalike tests were mixed, so we kept only segments with enough volume to justify further investment.

In March we launched AI videos and new static creative. The videos generated 21 demo bookings during the test. We also tested RAYA, a chatbot lead-magnet funnel. After several unsuccessful setups, a route through a SendPulse intermediary page began to work. The bot generated low-cost entries, but conversion from that entry to a demo needed separate evaluation.

In April, Facebook lead forms produced 125 leads at $8.97 each, while the RAYA funnel brought 294 entries at $4.72 each. The monthly report lists 419 leads and 14 demo bookings. We stopped treating CPL as the sole measure of success: a cheap entry is not a booked product demo. Website attribution was still being repaired during this period.

In May and June, we separated lead-form campaigns, website demo-booking campaigns and retargeting. AI videos and lookalike video audiences became the main working formats. A campaign optimized directly for Demo session booked generated 17 bookings at $48.94 each. We connected CRM data to Meta to provide a stronger quality signal. Advertising was stopped on 22 June, which limits direct comparison between adjacent periods.

In July we tested optimization for Closed Won and Completed / Decision Making stages. The Closed Won-optimized campaign generated 68 leads at $11.16 each that month. Following a budget increase, one working campaign grew from 21 to 51 leads per week while CPL fell from $14.58 to $11.18. Retargeting with new creative did not sustain results and was paused.

The July working report records 369 leads, 104 bookings and 52 attended demos. Half of the booked sessions took place. More rescheduled calls showed why optimizing for bookings alone does not settle the quality question. Bookings continued in early August, and tests using Quotation issued and Invoice issued stages began on 10-11 August.

The account moved from an expensive first test to a structure built around funnel stages. Direct optimization for bookings and CRM event feedback were the most useful changes for the sales objective. AI video and lookalike audiences helped generate volume, but we judged each combination by its progress toward demos as well as its lead cost. Ad interruptions, limited Calendly slots and UTM/CRM tracking issues constrained stability and attribution.

Stabilize event tracking through the thank-you page and CRM, and report bookings, attended demos and won deals separately. Continue video and deeper-event tests only where the data volume and quality support them. Monitor rescheduling and demo attendance as core funnel measures.