Cost per opportunity down 71%,on a smaller budgetthan the season before.
Samiad runs week-long international summer language camps for children and teenagers at boarding schools in the United Kingdom. The camps happen in July and August only, so a full year of budget has to land inside two months, and until this work started Google Ads was the single channel producing leads. Below is what the 2024 season looked like against the one before it.
- +45.76%
- more camp weeks sold
- more than 3×
- as many leads as the year before
- -71.27%
- lower cost per sales opportunity
- -34.3%
- lower cost per week sold
Half the data, and decisions made on it anyway.
Half the data, and decisions made on it anyway
Partial information from the CRM gave a picture, but not one solid enough to decide on. Analytics were rebuilt from the ground up rather than patched, because a number you cannot trace is a number you will eventually act on by mistake, and you will not know which one it was.
Two months to sell the whole year
The camps ran in July and August, nothing else. A selling window that short leaves no room to learn in-market: the budget has to be right before the season opens, because there is no second half of the year in which to correct it.
Targeting spread across too many countries at once
Campaigns ran in many markets in parallel, which diluted the message and made the results hard to read. Spread a limited budget that wide and every market gets too little to prove anything, so the one that would have scaled stays buried in the average next to the one that never would.
Read before writing. Then make the map smaller.
Strategy written after reading what was already running
Before anything changed, the existing activity was read: conversations with the internal team, a review of the tools in place, and a pass over the marketing material already live. The marketing and creative strategy came out of that reading rather than out of a template.
Creative built to be replaced, not to survive the season
Instead of a few fixed banners running from the start of the season to the end, Google Ads and Meta Ads got a cycle of testing and iteration, from video through to creatives built on social proof. With a budget this tight, the cheapest thing you can do is stop an ad going stale before your window closes.
Budget pulled back to the markets that were selling
Campaign structure and analytics were rebuilt around the markets generating the most sales, instead of broad global activity, and Performance Max was introduced alongside. Google Ads had been capturing existing demand well. The problem was that it was capturing all of it alone, which caps how far you can scale before the next lead stops being worth its price.
A second lead source, wired to move from test to scale
Meta Ads launched into markets where the brand already had a footprint and into new directions worth testing, with the structure set up so a market could move from test to scale on its own results. Event tracking and dedicated dashboards went in at the same time, so the reporting could answer which market had earned the next increment of budget.
Two seasons, and the cheaper one sold more.
The study compares the 2024 season with the year before it, when spend was higher. The two cost figures from that comparison sit below. The two volume figures appear in the row of numbers above instead. Both are rises, and a chart of two falling costs has no honest place for them.
Both bars point the same way. They are costs, both of them fell, so a longer bar means a bigger reduction and the comparison is honest. The two growth figures are deliberately not on this chart. Weeks sold went up 45.76%, which is the opposite direction on a different unit, and drawing a rise and a fall against one scale invites you to read 71.27 as beating 45.76 when the two measure nothing in common. Leads are the harder case: the study gives that figure only as a multiplier, “more than threefold”, with no percentage and no base number behind it, so there is no bar length that would be anything other than invented. It appears above exactly as the source phrased it. Two of the figures do close against each other: 45.76% more weeks sold at a 34.3% lower cost per week lands just under the previous season for total spend, which is exactly what the study reports happened. The CPO drop cannot be checked the same way against the leads figure, because a sales opportunity and a lead are two different counts in this text and nowhere does the source say they are the same one.
The decision that paid switched something off.
A limited budget spread across many countries buys you an average, and an average hides the market that would have scaled just as thoroughly as it hides the one that was never going to. Narrowing the map was the decision that paid here. How far you can narrow it is your own question: a business that sells its whole year inside two months behaves differently from one that sells every month.
The uncomfortable half of that decision is the switching off. Narrowing means cutting spend in places that still produce some leads, and you cannot do that on partial CRM data without guessing. That is why the analytics work came first here rather than last. If your reporting cannot tell you which market earned the last increment of budget, cutting markets is a bet rather than a decision, and in a business with one season a year that bet does not get a second run.
One more thing the study is explicit about, because it changes how you should read the numbers above. SEO went in alongside the paid work, as a third lead source built for the following season. It is not in these figures. Everything on this page came from Google Ads, Meta Ads and the analytics underneath them, and the part built for the season after has not reported yet.
This work was delivered by MTA Group, which Zero Fluff Digital is part of, by the same specialists who would work on your account. Every figure here comes from the case study MTA published, republished here with their consent. The reviews behind it are public and verified on Clutch, where clients rate the work rather than the agency describing itself.
An average across markets hides the one that would have scaled.
30 minutes, no deck. You will see where your spend is spread thinner than it can prove, and how long pulling it back takes to show up in your numbers.
Book a callNext case studyMultikino: cost per purchase down 54.70% →