The agency is not performing: what to check before changing anything

A quarter closes below target. The agency's report explains why, with charts that mostly point up. Head office, several time zones away, cannot tell whether the explanation is sound, and the question on the next regional call becomes: should we change the agency?

Sometimes the answer is yes. But changing a marketing agency in Vietnam is slow and expensive, and the most common way it goes wrong is not choosing the wrong replacement. It is changing agencies without having first established what went wrong, so the next agency inherits the same accounts in the wrong name, the same reporting head office cannot check, and the same targets nobody could have hit.

This piece sets out what to check before changing anything, in the order that makes each step useful to the next. Whether the Vietnam budget as a whole should be cut is a separate question, covered elsewhere in this cluster; here the budget stays and the question is the agency.

"Not performing" against which number, set by whom

Most disputes with an agency start with a number that was never properly agreed. Before judging the agency, write down three things about the metric that disappointed:

• Who set the target, and on what basis. A target copied from another market, or from a regional plan built before anyone looked at Vietnam, can be missed by a good agency.
• Whether the agency controls it. Cost per lead is largely the agency's; sales from those leads depend on the distributor's or the sales team's follow-up; revenue depends on price, stock and competitors as well.
• Where the number comes from. If the only source is the agency's own report, head office is being asked to judge the agency by the agency's account of itself.

If the number was set without local evidence, sits largely outside the agency's control, or comes only from the agency, the conclusion "the agency is not performing" is not yet supportable, whatever the result.

Separate the agency's share from everyone else's

In Vietnam, a marketing result passes through more hands than most head offices expect: the agency, often a distributor, a marketplace, a sales or customer-service team answering messages, and the platforms themselves. When a result falls, each of them can be the cause.

Only the last row points clearly at the agency. It is also the least common first finding. That does not excuse the agency when it is responsible; it means the evidence has to be assembled before the verdict, not after.

Losing control usually comes before losing performance

Head offices often describe the moment they decided to review their agency as "we lost control of marketing in Vietnam". Looking back, the loss of control almost always predates the loss of performance by months. It shows up as small things that seemed reasonable at the time:

• the advertising accounts were opened under the agency's business manager, because it was faster;
• the monthly report uses the agency's definitions of a lead or a sale, and nobody at head office has seen the underlying export;
• decisions to shift budget between channels are made locally and reported afterwards;
• only one person at the agency understands how tracking is set up;
• head office cannot answer a simple question, such as how many first-time customers came in last month, without asking the agency.

None of these is misconduct. Together, they mean head office cannot judge the agency at all, so every discussion about performance becomes an argument about whose numbers are right. Changing agency does not fix this. It resets the same arrangement with a new party.

Take back visibility before judging anything

The first step is therefore not a performance review. It is a control review, and it can be done without accusing anyone:

1. List every asset — advertising accounts on each platform, analytics and tag management, pages and channels, the Zalo official account, marketplace stores — and who owns each one. 2. Move ownership to the company, with the agency given the access it needs to work. This is ordinary practice, not a sign of distrust. 3. Get a sales figure the agency does not produce: your own order records, or your distributor's sell-out reporting, broken down at least into new and returning customers. 4. Agree definitions in writing: what counts as a lead, a qualified lead and a sale, and which system is the source for each.

A well-run agency will usually welcome this, because it turns vague dissatisfaction into specific questions it can answer. How a finished oversight report looks when these pieces are in place — for a brand selling through a distributor — is shown in the sample audit report.

One written chance: a sixty-day correction plan

Once head office can see, it can ask. The request that works is short, specific and dated. A useful correction plan has three to five items, each with an owner, a date and a way of judging it:

Sixty days is usually enough for changes inside existing channels to show. A plan that involves a new channel or new creative production may need a full quarter. What matters more than the period is that the plan exists in writing before it starts, so that at the end the question "did it improve" has an answer both sides accept.

A worked example: the same quarter, read two ways

The figures below are illustrative, invented for this example.

A foreign personal-care brand sells in Vietnam through a distributor and a Shopee store, with a local agency running paid social and marketplace advertising. The agency's quarterly report shows cost per lead down 20% and lead volume up 30%. Head office sees flat sell-out and is ready to change agency.

Assembled from sources the agency does not produce, the quarter looks different:

The agency did what it was asked: more leads, more cheaply. Fewer of them were contacted, and the brand raised its effective price on the marketplace while competitors did not. Changing agency would have removed the one part of the chain that was working. The corrections belong to the distributor's follow-up and to the pricing decision — and the agency's brief should now be rewritten around leads that are actually contacted and converted, not leads generated.

Testing the agency's explanation instead of arguing with it

When results disappoint, an agency will usually have an explanation: the auction got more expensive, a competitor launched, the platform changed its algorithm, the marketplace campaign pulled demand forward. Many of these are true. The problem for head office is that each one sounds equally plausible from a distance, and arguing about them in a review meeting settles nothing.

Each common explanation leaves a trace that can be checked without taking the agency's word or disputing it:

Asking for the evidence behind an explanation is not an accusation. A good agency will produce it quickly, and the conversation moves from whether the explanation is believable to what to do about it. An agency that cannot produce it, repeatedly, has told head office something important — usually about its own visibility into the account rather than its honesty.

When changing is the right call

Changing is the right decision when, with visibility in place and a written plan given a fair period, one of these holds:

• written requests were not carried out, twice;
• the people working on the account keep changing, or are not named;
• raw data is still withheld after a clear request;
• the brand's needs have outgrown the agency — more channels, marketplaces, provinces, or a regional reporting line the agency cannot serve.

The last reason is nobody's failure, and it is the easiest to discuss openly. Which model suits the next stage — a local agency, a regional one, or someone of your own — is covered in a separate piece in this cluster.

If you change, change in sequence

A change handled in the wrong order loses the history that makes the next agency effective. The order that preserves it:

1. Secure ownership first, while the relationship is still working: accounts, pixels and tracking, audiences, pages, stores, creative source files. 2. Export twelve months of history by campaign and by month, and ask the outgoing team for a one-page summary of what they tested and learned. 3. Brief the pitch on your numbers and definitions, not on the outgoing agency's. 4. Plan an overlap on the channels that cannot pause, and expect results to dip for several weeks while the new team learns the accounts. 5. Write the new contract with exit terms — ownership, notice, handover — so the next change, if it comes, is cheaper.

Performance comparisons in the first two months of a new agency say little. Judge it on the written plan it proposed, and on whether the numbers head office can now see move in the direction that plan predicted. How the numbers themselves should be built and reconciled is the subject of the measurement cluster.

Where this work stops

MWY does not run advertising, operate stores or take commission from any agency or platform. When a client needs a new agency, MWY shortlists three vetted options with the criteria to judge them, and takes no commission from the one chosen.

Taking back control of accounts and data, separating the agency's share of a result from everyone else's, writing the correction plan and judging it at the end of the period is the core of Vietnam Marketing Advisory & Oversight. The first month is an audit of exactly these questions, before any decision about the agency is taken.