We took over a B2B SaaS account in early 2024. Good product, decent market, reasonable budget — around $40K/month across Google and Meta. The previous agency had been running it for about eight months and the results were... fine. Not bad. Not great. Cost per demo was hovering around $285, and the client was frustrated because they'd been promised "$150 demos within 90 days" during the sales pitch. (Side note: if an agency gives you a specific CPA number before they've even seen your account, run. But that's a different post.)
By month 6 with us, we had cost per demo down to $210. Good improvement, but nothing earth-shattering. The client was happier but still a little antsy. Then something happened in year two.
By month 14, cost per demo hit $138. By month 18, it was $112. Same budget. Same platforms. Same product. The campaigns didn't suddenly get smarter because we had some revelation. They got smarter because of everything we'd learned over those 18 months — every test, every failed creative, every audience that didn't convert, every search term we added as a negative. All of that accumulated knowledge was baked into the account, and it started compounding.
That's the story I want to tell today. Not a story about genius. A story about patience and accumulated intelligence.
Month 1 Is Educated Guessing
I don't care how experienced your strategist is or how good your agency's onboarding process is. Month 1 of any new paid media engagement is, at best, educated guessing. You can study the business, interview the founder, read every page of the website, analyze the competitive set. You should do all of those things. But you still don't know anything yet.
You don't know which of your three value propositions resonates most with cold traffic. You don't know whether long-form landing pages outperform short-form for this particular audience. You don't know if "book a demo" converts better than "start free trial" on Google versus Meta. You don't know if the Tuesday-through-Thursday scheduling that works for most B2B accounts also works here, or if this particular audience converts on weekends because they're researching solutions on their own time.
You have hypotheses. Good ones, hopefully, based on experience with similar accounts. But hypotheses aren't data.
Here's the thing most businesses don't understand: the first three months of paid media aren't primarily about driving results. They're about generating intelligence. Every dollar you spend in those early months is buying you information — what works, what doesn't, and what to do next. The agencies that perform best long-term are the ones that treat those early months as a structured learning phase, not a "hit my KPIs or I'm fired" panic sprint.
What You Actually Learn Over 12 Months
Let me walk you through what the learning curve actually looks like on a typical account, because I think people drastically underestimate how much there is to discover.
Months 1-3: The Basics. You figure out which keywords actually drive quality traffic versus which ones just drive clicks. You test 3-4 audience segments on Meta and learn that two of them are duds. You discover that the client's "main" landing page actually has a worse conversion rate than a product page they barely promote. You build your first negative keyword list based on search term data. You start to understand the sales cycle — is it 7 days from click to close, or 45?
Months 4-6: Pattern Recognition. Now you've got enough data to see trends. You notice that leads from certain keyword themes close at 2x the rate of others, even though the CPA is similar. You find that video creative on Meta outperforms static for prospecting but static wins for retargeting. You discover that the client's industry has a seasonal pattern nobody mentioned — maybe a trade show every Q3 that spikes branded search, or a summer slowdown where CPCs drop and you should actually increase budget, not decrease it.
Months 7-12: Refinement. This is where things get interesting. You've run enough A/B tests on ad copy to know what language converts. Not "we think professional tone works better" but "we tested 14 headline variations and the ones that mention [specific benefit] outperform by 32%." You've identified the micro-conversions that predict macro-conversions — maybe people who watch the product video convert to demo at 3x the rate, so you start optimizing toward video views as a mid-funnel signal. You've learned the client's business well enough to anticipate things: you know a pricing change is coming in Q4, so you've already prepped updated ad copy and landing page variants.
All of this compounds. Each insight feeds the next decision. The negative keyword list you built in month 2 saved you money that let you test more audiences in month 5, which revealed a high-performing segment that became your top campaign by month 9.
Year 2 Is Where the Magic Happens
Year 2 is when you stop guessing entirely. You know this account. Your strategist knows the business, the audience, the competitive dynamics, the seasonality, the internal politics (which stakeholder cares about which metric), and the conversion patterns down to the day of week.
I pulled numbers across 11 accounts that we've managed continuously for 2+ years. On average, Year 2 ROAS was 41% higher than Year 1. Forty-one percent. Same spend, same platforms, dramatically better returns. And that's an average — some accounts saw 60%+ improvement.
One ecommerce client in the home goods space went from a 3.2x blended ROAS in Year 1 to a 5.1x in Year 2. We didn't change the strategy dramatically. We refined it. We knew which products to push in which seasons. We knew that their audience responded to lifestyle imagery in spring/summer and product-focused shots in fall/winter. We knew that Meta prospecting worked best when we refreshed creative every 3 weeks (not the 4-week cycle we'd started with). We knew that their Google Shopping feed performed best when we used specific title structures we'd tested over months.
None of this knowledge came from a flash of brilliance. It came from methodically documenting what worked, what didn't, and why.
Why Switching Agencies Resets the Clock
This is the part that should make you a little uncomfortable if you're thinking about switching agencies or bringing things in-house.
When you switch agencies, you reset the learning curve to month 1. Full stop. The new agency might be smarter, more experienced, and better at their jobs. They might genuinely be an upgrade. But they still don't know your account. They don't have the 14 months of search term data in their heads. They don't know that your Meta audiences fatigue faster than average because your market is niche. They don't know that your highest-LTV customers tend to come from a specific geographic region that the previous agency discovered by accident in month 8.
Some of this knowledge lives in the platform — historical data, conversion history, audience learnings. But a shocking amount of it lives in people's heads. Or in scattered notes. Or in Slack threads that nobody saved. When the person who managed your account leaves (or when you leave them), that knowledge walks out the door.
I've seen this play out dozens of times. A company switches to us from another agency. Months 1-3 with us, performance dips slightly — not because we're worse, but because we're learning. By month 6, we're back to where the previous agency left off. By month 9, we're surpassing it. But those first 6 months? That was expensive re-learning. Knowledge that already existed, bought and paid for by the client, was lost because it wasn't captured anywhere permanent.
That bothers me. It should bother you too.
The Knowledge Loss Problem Is an Industry Problem
Look, I'll be honest about something: this isn't just a switching problem. It's a retention problem, and it affects every agency in the industry.
Average tenure at a digital marketing agency is about 18 months. Think about that. Your account manager finally hits their stride at month 12, starts delivering compounding returns, and then at month 18 they leave for a new job. Now someone new takes over — someone at the same agency, using the same tools — and they still have to relearn a bunch of context because institutional knowledge transfer is hard.
I've watched this happen at the big agencies. Accounts would cycle through managers every year or two, and each transition created a performance dip. The knowledge was supposedly in the "platform" or the "account documentation," but in reality, the documentation was usually a stale onboarding deck from two years ago and some disorganized notes in a shared drive.
The agencies that win long-term are the ones that solve this problem. The ones that build systems to capture account intelligence in a structured way — not just "what campaigns are running" but "what we've learned, what we've tested, what works for this specific business and why." The ones that treat account knowledge as an asset that belongs to the client and persists regardless of which human is in the chair.
Most agencies don't do this. Most agencies rely on individual people to carry the knowledge in their heads. And when those people leave — which they always do — the knowledge evaporates.
What This Means for You
If you're running paid media right now, here's what I want you to think about:
How much of your account's intelligence is written down somewhere? Not campaign settings — anyone can see those in the platform. I mean the why behind every decision. Why did you choose those audiences? What did you test and reject? Which creative themes outperformed? What seasonal patterns have you identified? What does your sales team say about lead quality from different campaigns?
If the person managing your account disappeared tomorrow, how much of that knowledge would survive? If the answer is "not much," you have a serious vulnerability. You've been paying for intelligence — through ad spend, through management fees, through time — and none of it is being stored in a way that persists.
Now think about this: what if all of that intelligence was documented? Structured. Updated monthly. Available to anyone who takes over the account. What would that mean for your Year 2 performance? For your Year 3?
The compounding advantage in paid media isn't about having a bigger budget or fancier tools. It's about accumulated, organized intelligence applied consistently over time. The accounts that compound are the accounts where nothing is forgotten — where every test, every insight, every failure gets captured and feeds the next decision.
I've built my entire agency around this idea. Not because I'm smarter than other agency founders, but because I've been burned by the alternative enough times to know better. The unfair advantage isn't talent. It's institutional memory.
And institutional memory doesn't build itself. Someone has to build the system. Someone has to commit to capturing what you learn, month after month, in a format that outlasts any individual person.
If your current setup doesn't do that — if your agency or your in-house team is relying on tribal knowledge and good intentions — you're leaving compound growth on the table. And compound growth, in my experience, is the difference between accounts that plateau and accounts that get better every single year.
Year 2 always beats Year 1. But only if you keep what you learned in Year 1.