Switch when the diagnosis is wrong, not when the numbers are down. Numbers fall for reasons outside anyone's control. What is inside their control is whether they identified the cause early, told you, and proposed something specific. Two quarters of vague explanations is the signal.
The short version
- A bad quarter is not a reason. A bad diagnosis is.
- Two quarters of the same unshipped initiatives means capacity, not strategy.
- No kill recommendation across a whole catalog means the fee is funding optimism.
- Fix first if the problem is scope or caseload. Both are negotiable.
- Month-to-month terms make this decision cheap. A long contract makes you rationalize staying.
Switch triggers versus fix triggers
I run Flapen with 50 operators managing about 70 brands, and I have been the client terminating agencies at BRANDED and Moonshot Brands. Not every problem justifies a switch, and switching carries real cost.
| Situation | Fix it | Switch |
|---|---|---|
| Caseload too high, work is slow | Ask for a reassignment first | If refused |
| Scope narrower than you assumed | Renegotiate scope | If they cannot cover it |
| One bad quarter, clear diagnosis | Stay | No |
| Two quarters, vague explanations | No | Yes |
| No kill recommendation, ever | Ask directly once | If nothing changes |
| Cost of customer acquisition drifting up, unnoticed | No | Yes |
| Account manager churn each quarter | Ask for stability | If it continues |
| Reporting without profitability metrics | Ask for it in writing | If refused |
The pattern: problems of capacity or scope are usually fixable, because they are commercial. Problems of judgment are not, because you cannot negotiate someone into diagnosing better.
The three genuine switch signals
They cannot tell you why. Revenue is down and the explanation is general market conditions, increased competition, or algorithm changes. Those are sometimes true. What matters is whether the explanation arrived before you asked and came with a specific proposal.
Nothing ships. Put your last three monthly plans side by side. If the same initiatives appear in all three, the constraint is their capacity, not your account. Ask how many brands your named manager carries. We run about 1.4 brands per operator, and above eight nothing moves quickly.
Nobody has ever recommended stopping. Any agency on a recurring fee has a structural incentive to keep every product alive. An agency that has never proposed killing a product across a catalog is not exercising judgment. Ask what they killed last quarter, once, and listen to the answer.
What switching actually costs
Be honest about it before you act.
Expect four to eight weeks of reduced momentum: transition, audit, and a new team learning the account. Campaign history transfers imperfectly even in good conditions, and some ad learning restarts. Creative may need rebuilding if you only received flattened exports.
Against that, the cost of staying is compounding. If cost of customer acquisition is drifting up unaddressed, every month adds to the gap. The transition cost is one-time and known. The staying cost is not.
Fix it first, then decide
Before giving notice, have one direct conversation and put three things on the table:
- The numbers you are worried about, specifically cost of customer acquisition trend and organic share of revenue.
- The initiatives that have not shipped, with dates.
- What you need in the next 60 days to stay.
A good agency will respond with a reassignment, a scope change, or a concrete plan. A weak one will respond with reassurance and a longer report. That conversation is worth having, because you learn more from the response than from another quarter of watching.
What most agencies will not tell you
Contract length shapes this decision more than performance does. On month-to-month terms with 30 days' notice, acting on your judgment costs a month, so you act when you notice. On a twelve-month term with a 90-day notice window, the same judgment costs most of a year, and sellers talk themselves into staying rather than absorb it.
That is not an accident. A long term exists to make this exact moment expensive.
The other thing: agencies know which clients are unhappy long before those clients say anything, because the signals show in meeting attendance and question tone. If nobody has proactively asked whether you are getting what you expected, that silence is information too.
Related answers
- Red flags your Amazon agency is underperforming
- Checklist for changing Amazon agency smoothly
- Amazon agency transition timeline and risks
- How to audit current Amazon agency performance
- Hiring an Amazon agency: the complete guide
Our terms are month-to-month so the decision stays cheap, at Flapen.
