Most B2B teams that say they're 'testing' are actually just trying things sequentially and comparing the results informally afterwards (which isn't testing, it's guessing with extra steps. Working with an A/B testing agency should mean something more specific) properly randomised experiments, run for long enough to reach statistical confidence, on changes big enough to plausibly matter.
We design and run experimentation programmes for tech, fintech and professional services companies where traffic and lead volume are often lower than the ecommerce world that most testing methodology was originally built for. That constraint changes almost everything about how a B2B testing programme should be designed, and it's the first thing we address before running a single test.