Before Apopo I ran sales and marketing for New Zealand Trails, a guided walking operator. Over that time I grew direct bookings 236% year on year. I also hired and fired three marketing agencies, which taught me considerably more.
None of them were cowboys. All three had good people, decent decks and reasonable invoices. They failed in three different ways, and the failures looked structural rather than personal – which is why I think they are worth writing down. Three agencies is an anecdote, not a study, and you should weigh it that way. It is also three more than most people who write about this have paid for.
Agency one: generic is a choice, and you pay for it.
The first agency was competent and completely interchangeable. They ran the same playbook on us that they ran on a SaaS company and a homewares brand, because that playbook was the product.
The problem is that what moves someone to book a two-week guided walk has almost nothing in common with what moves someone to start a software trial. The consideration window is months. The purchase is expensive, occasionally once-in-a-lifetime, and frequently made by one person on behalf of two. The objections are about fitness, group size, weather and whether they will be the oldest person there – not about price and features. An agency that has not lived that writes ads which are technically correct and put people off.
You cannot brief that in. I tried. Understanding a market is not a document you hand over at kick-off, it is the thing that lets someone write a headline that lands – and if they have to learn it on your budget, you are paying tuition fees.
Agency two: defaults are not a strategy.
The second agency put nearly everything into Performance Max and reported enthusiastically on the volume.
Performance Max is not a bad product. It is a bad default, and it is the easy option, which is why lazy agencies reach for it. It will find you a great deal of cheap traffic and it will hand back very little control over where that traffic came from or what it was looking for. For a considered, high-value purchase, control is the whole game – I would rather have a tenth of the clicks and know exactly which question each one was asking.
When we moved that account off Performance Max and onto specific search campaigns, Google Ads conversion rate went up 27%. Fewer people, better matched. And I should be straight about that number, because conversion rate is a ratio and taking cheap traffic out of the denominator flatters it almost by construction. The reason I trust it is not the percentage – it is that enquiry quality went up at the same time and the sales team stopped complaining.
“The volume chart looked worse and the business did better. That is an awkward slide to present, and it was still the right call.”
The failure here was not the tactic, it was the incentive. Automated campaign types are easy to run, easy to report and hard to argue with. If nobody is accountable for the quality of what arrives, the easy option quietly becomes the permanent one.
Agency three: my biggest lever was not in the ads.
The third agency was good at what they did. What they did was one channel, and the retainer went up if you wanted a second.
Meanwhile the largest single improvement we made that year came from one A/B test on the homepage. New-user conversion rate up 15%. No extra spend, no extra traffic, no new channel – just making more of what we already had. It was completely outside the agency’s scope, so it was nobody’s job, so for a long time it did not happen.
That is the structural failure in the channel-shop model. Your biggest lever is wherever it happens to be – sometimes the ads, often the site, occasionally a fact on a page that is answering a question wrong. An agency scoped to one surface can only ever find opportunities on that surface, and will honestly report that it found none elsewhere, because it was not looking. It is also why conversion rate optimisation keeps getting sold as an upsell rather than done as a matter of course.
What I would ask before signing anything.
These are the questions I wish I had asked. They are deliberately concrete – each one has an answer that is either specific or evasive, and the evasion is the information.
- Who runs my account, and how many others do they hold?
A name and a number. If the person answering is not the person doing the work, ask why they are the one in the room.
- What in my business would change your recommendation?
Tests whether they understand the market or the channel. An agency that cannot name a sector-specific constraint has one playbook.
- What are you not going to do, and why?
A real strategy excludes things. If everything is in scope and nothing is ruled out, nobody has made a decision yet.
- Which number are you accountable for, and by when?
Named before the contract starts, not chosen retrospectively from whatever moved.
- What would make you tell me to spend less?
The most revealing question on the list. Every agency has a point at which more budget stops helping. Ask them where yours is.
One more thing worth knowing before you sign a two-year plan with anyone. Roughly a quarter of UK agency staff left their agency in 2025, and most of those were resignations rather than redundancies. The specific person who spent six months learning your business is, statistically, quite likely to be somewhere else by the time that knowledge is worth anything. Ask what happens to your account when they go – and whether anything they learned is written down.
Two things to check tomorrow.
If you are already in a relationship and quietly unsure about it, there are two checks that do not require a difficult conversation.
The first is ownership. Open Google Ads, Meta Business Manager, Analytics and Tag Manager and confirm the accounts are in your name, with you as an admin, and that the agency has access rather than possession. This is not paranoia – it is the difference between changing supplier and starting again, and you find out which one you have on the worst possible day. An agency that hesitates has told you something.
The second is a decision log. Ask for the last ten changes anyone made to your account and the reason for each. A good agency will send it slightly annoyed that you asked. A struggling one will send you last month’s report again, which is the answer. Reports describe what happened; a decision log shows whether anybody was steering.
What “more” actually means.
Not more channels, and not more reporting. More of the two things that were missing in all three cases: someone senior who actually knows your market, and a remit wide enough that they can go where the opportunity is instead of where the contract is.
That is the whole reason Apopo is built the way it is. One senior expert runs the account end to end across every surface – paid, organic, the answer layer, creative and the site itself – and holds a deliberately capped number of accounts, so the person who learns your business is the person who runs it. It is not a clever model. It is just the arrangement I kept wishing I could buy.
If you are currently unhappy with an agency, my honest advice is to ask the five questions above before you start a search for a replacement. Quite often the answers explain the results you are getting, and occasionally they fix them. And if you want the definitional version of all this first, we wrote what a performance marketing agency should actually be separately.