I've written the SOW and I've signed the check. I've been the agency rep walking into a QBR and I've been the client sitting across the table wondering if any of this is actually moving my business. Running Ad-Apt while also operating a construction company, a vineyard, and a fintech platform means I'm perpetually in both seats. That dual vantage point is the most useful thing I own.
And what it's shown me is that most agencies — including versions of Ad-Apt I'm not proud of — make the same four mistakes on repeat. They're not incompetence mistakes. They're incentive mistakes. And incentive mistakes are harder to fix because they're structural.
The Real Mismatch: Retention vs. Outcomes
Agencies are quietly optimized for one thing: keeping the contract. That's not cynical — it's just math. Churn kills agency economics. A lost client at $15K/month is a six-figure hole in the P&L. So everything gets oriented, often unconsciously, around not losing the account.
The client is optimized for something else entirely: outcomes. Revenue. Pipeline. ROAS that actually maps to margin. And for the first year or two of a good engagement, those two things point in the same direction. Then they diverge.
The divergence looks like this: the agency starts protecting what's working instead of pushing what's next. They hit a comfortable ROAS and stop testing. The QBR deck gets shinier as the growth curve flattens. They win the presentation and lose the trust — because the client can feel that nobody in the room is treating their money like it's real.
I've been that client. I've also, honestly, been that agency. The difference now is I can name the mechanism. When the agency's survival brain kicks in, the client's results are the first casualty. That's the core truth I keep coming back to: override the survival brain, or it makes all your decisions for you.
Three Things Agencies Consistently Get Wrong
Every agency I've hired — and this includes firms with strong reputations — has gotten at least two of these three wrong. Every time.
1. Reporting cadence that serves the agency, not the client
Most agencies default to monthly reporting. Monthly is comfortable for the agency. Monthly means you have time to smooth out the bad weeks, contextualize the bad months, and present a narrative rather than raw data. For a client running a live business, monthly is nearly useless. By the time you see the trend, you've already missed the window to act on it.
What I want as a client — and what we build at Ad-Apt — is a live dashboard with weekly check-ins that are explicitly structured around decisions, not updates. There's a difference. An update is: here's what happened. A decision cadence is: here's what we're seeing, here are the two options, here's our recommendation, do you approve or redirect? The second one respects that the client's time is the only currency that matters.
2. Murky decision rights
Nobody ever writes down who gets to call what. The agency assumes they have autonomy on creative. The client assumes nothing goes live without a sign-off. That gap costs weeks and kills momentum. I've watched campaigns sit in review limbo for 11 days because nobody established upfront whether a $5K budget test needed director approval or if the agency could just run it.
Now I push for a one-page RACI on day one of every engagement, whether I'm the client or the agency. Who approves budget changes above $X? Who owns creative sign-off? Who can kill a campaign mid-flight? Write it down before you need it, or you'll be negotiating it in the middle of a crisis.
3. Escalating the wrong things (or nothing at all)
Agencies under-escalate bad news and over-escalate wins. I get it — nobody wants to be the person who delivers a down month. But selectively surfacing information is a form of dishonesty, even when it's well-intentioned. If my ROAS dropped 30% week-over-week and the agency knew on Tuesday but waited until the Friday report to tell me, I've lost four days I could have been reallocating budget or pausing spend.
At Teton Gravity Research, we were driving 7.2x ROAS on their paid media. That's a number worth celebrating. But the only reason it held is because we were paranoid about the signal degradation that precedes a performance cliff — and we escalated early when something looked off. The escalation cadence is what preserved the result, not the launch strategy.
What I Demand as a Client Now That I Didn't Before
Before I'd run an agency, I hired agencies the way most founders do: I evaluated their pitch deck, checked their case studies, and hoped the team in the room was the team I'd actually get. I was optimizing for the wrong inputs.
Here's what I demand now:
- Raw data access, always. I want the GA4 property access, the Meta Ads account access, the Search Console login — everything. I'm not logging in every day, but I want to be able to. Any agency that resists this is telling you something about how they intend to control the relationship.
- A named senior person with real accountability. Not a CSM who relays messages to the strategist. I want the strategist in the room, and I want to know they have actual authority to make calls. If my account escalates to a VP only when I threaten to leave, the structure is wrong.
- A clear theory of the case. Why is this channel the right bet? What's the falsifiable hypothesis? If something doesn't work in 60 days, what does that tell us? Agencies that can't answer that upfront are running on vibes, not strategy.
- Explicit thresholds for what triggers a conversation. If CPA goes above $X, I want a call that day — not a footnote in next month's report. This has to be agreed on in writing before the campaign launches.
None of this is unreasonable. But most clients don't know to ask for it until they've been burned once. I've been burned enough times — on both sides — to know exactly where the gaps are.
The One Move That Breaks All Trust (It's Not Losing Performance)
Most clients assume the trust-breaker is a bad month. It's not. Bad months happen. Markets shift, creative gets stale, a competitor floods the auction. I can work with a bad month if we're honest about it.
The trust-breaker is when I find out the agency knew something I didn't, and chose not to tell me.
It might be a platform policy change that's about to affect delivery. A creative that's been flagged but they're hoping it gets approved anyway. A budget pacing issue they're planning to "make up" in the last week of the month. These aren't catastrophic problems in isolation. The cover-up is the catastrophe.
The moment I realize the agency has been managing my perception instead of managing my account, I'm done. Not angry. Done. Because perception management is a tell: it means they've decided their relationship with me matters more than my results. And that's the opposite of what I'm paying for.
Data over gut means nothing if you're selectively surfacing the data. The gut always wins when the data is curated to confirm it.
How We Structure Engagements at Ad-Apt to Avoid All of This
We run on the agency-as-extension-of-team model, which is a phrase a lot of agencies use and almost none of them operationalize. Here's what it actually means in practice.
Every client gets a shared Slack channel with our team, not because it's a nice perk but because it forces the kind of real-time communication that prevents the information gaps I described above. The account lead at Ad-Apt is the person doing the work, not a layer of management translating what the person doing the work is thinking.
We build live dashboards — custom, not templated — that the client owns and can access independently of us. When True Blue Car Wash was scaling aggressively before their acquisition by Couche-Tard, the measurement infrastructure we built meant they could see performance in real time, not at the end of a reporting cycle. That visibility was part of what made the acquisition story clean. You can't tell a compelling growth story with monthly screenshots.
Decision rights get documented in the first week. Budget thresholds, escalation triggers, creative approval chains — all of it. It takes two hours and saves a dozen arguments.
And when something goes wrong — because it will — we say so that day. Not in a panicked way, but clearly: here's what we're seeing, here's why it matters, here's what we're doing about it. That discipline is harder than it sounds because it requires overriding the instinct to fix it quietly and present a clean narrative. We override that instinct on purpose. Every time.
I've been the client who fired an agency for the exact mistakes I've described. I've also been the agency that made them. The only way I know how to avoid repeating them is to stay in both seats at once — which is uncomfortable, and also the most honest form of quality control I've found.