Illustrative market work, described honestly
What follows are anonymised, composite write-ups built from the kinds of engagements we actually take on — no client names, no logos to show off, and no promise that your market will see identical numbers. We would rather walk through method and trade-offs honestly than hand over a polished highlight reel.
A creative fatigue check for a BC consumer brand
A British Columbia consumer brand came to us with a paid account that had quietly stopped growing. The same three ad variants had been running for months, cost per result had drifted upward for six straight weeks, and the internal team suspected a targeting problem. Our first move was not a new campaign — it was a creative fatigue check against the existing audience cohort map, using AI analytics to separate genuine audience saturation from a simple bidding issue. The check confirmed fatigue: cohort signal on the top variant had fallen by more than half since launch. We rebuilt a variant test cell with five new creative directions, kept the original cohort definitions largely intact since the targeting itself was sound, and pushed a fresh performance wave in staged budget blocks rather than a full reset. Within the first learning board cycle, cohort signal on the new variants was measurably stronger than the fatigued set had been at any point in its final month — a result we reported plainly, including the two variants that underperformed our own expectations.
Media orchestration for a retailer entering two cities
A regional retailer opening physical storefronts in two cities in the same quarter needed one coordinated market brief that still respected real differences between the two audiences. Treating both cities as a single generic launch would have wasted budget on channels that performed unevenly by market; running two entirely separate plans would have duplicated cost and lost the benefit of shared creative testing. We built a single audience cohort map with city-specific variants, drawing on AI audience insights to compare search and social behaviour between the two populations before a dollar moved, ran media orchestration in parallel across both expansion city briefs using AI-assisted media notes on local pacing, and kept one shared learning board so a signal from one city could inform a budget shift in the other within days rather than at the end of the quarter. Paid social and local search performed differently enough between the two cities that we rebalanced budget roughly three weeks in — a change we would have missed running the plans in isolation. Store-visit and local engagement signal, not follower counts, were the primary read we reported back.
Cost-per-acquisition versus cohort quality for a scale-up
A software scale-up arrived with a cost-per-acquisition target set by a previous agency and a growing suspicion that the number, while technically improving, was masking a decline in the quality of the cohort actually converting. We ran a signal-to-noise review across their existing channels, using AI analytics and a set of AI research briefs to separate raw acquisition volume from downstream retention and revenue-per-account by cohort. The review confirmed the suspicion: the channel with the lowest reported CPA was also producing the weakest twelve-week retention of any paid source. Rather than optimizing purely to lower acquisition cost, we rebuilt the media plan around cohort quality as the primary measurement, accepting a modestly higher blended CPA in exchange for cohorts that retained meaningfully better. This is a case where honest attribution meant recommending a metric most vendors are not eager to report against, and where a human-owned call — not an automated bidding rule — decided which channels stayed in the plan.
Why client logos and exact numbers stay out of this page
Confidentiality terms cover most of what we do before a single detail becomes public, and plenty of clients would rather their campaign numbers stayed private even once a push has wrapped. So instead of a client logo wall, we would rather talk through the actual decisions we made and be upfront about outcomes — including the ones that landed softer than we had hoped going in. Wherever an AI-assisted draft fed into putting these summaries together, a strategist checked it before publication, and none of the figures mentioned on this page should be read as a forecast for what a new engagement would produce.
Have a market that looks something like this?
Give us the category, who you are trying to reach, and whatever already exists. We will point you toward whichever pattern above looks closest to your situation.