Method

    Running a social channel for a Moroccan fashion label.

    How we build organic distribution when paid has been doing all the work

    Client

    Confidential — available on request

    Social Media

    Fashion · E-commerce

    Availability

    On request

    Sector

    Fashion · Morocco

    Method note

    This page documents a method, not a client result. It is written in full because the approach is the reusable part of the work. The client's name and the figures from their engagement are not published yet, because they have not been cleared for publication — we would rather show the method than invent a result.

    Stand-in frame for this method note, with the page title set in cream on a charcoal field and an amber asterisk mark
    Stand-in frame standing in for the monthly reporting dashboard used to run a social channel

    What Usually Goes Wrong

    The first thing to establish is that there is no editorial position. Content is being produced to fill a calendar rather than to answer a question a buyer is already asking, so nothing accumulates: every post starts from zero and nothing the brand publishes makes the next one easier.

    The second is format collapse. The same asset is posted everywhere, so reach, proof and conversion are all measured against the same thing — and a post that underperforms cannot be diagnosed, only replaced.

    The third is attribution. Decisions get made on the numbers the browser reports, which have undercounted since iOS 17. This matters more than any creative choice, because optimising a signal that is wrong is worse than not optimising at all.

    The Approach

    The work splits the channel into three jobs — reach, proof and conversion — and gives each one a different format and a different language rather than posting the same asset everywhere.

    Editorial position first

    Before any calendar, we established what the label stood for and who it was for, because content without a position produces noise rather than recognition. The position was written down in one sentence and used as a filter for every subsequent post.

    Three formats, three jobs

    Reels carried reach, carousels carried product proof, and Stories carried the in-stock and delivery information buyers ask about daily. Each format was measured against the one job it was doing, so a low-performing post could be diagnosed rather than just replaced.

    Language per audience

    Darija for reach and warmth, French for consideration, and the product copy kept in both so nothing was lost in translation. Running one message in three languages reads as a translation; writing three messages reads as a strategy.

    Server-side tracking before optimisation

    Conversions were recorded outside the browser and reconciled weekly against real orders. Any creative decision made before that reconciliation was reversible; made after it, it was not.

    What We Measure

    This page documents the method and the measurement frame rather than a result set. The client's name and the figures from their engagement are published once publication clearance is in place. What follows is exactly what we measure and how each signal is read.

    Reach

    Monthly, per platform

    Proof

    Saves and shares per post

    Revenue

    Orders reconciled weekly

    Reach

    • Monthly reach is read per platform, never in aggregate — a total that hides the one channel actually carrying the account is not reach.
    • The share of views coming from non-followers is the figure that matters. Reach that only reaches people who already follow is a broadcast, not distribution.
    • Median and best-performer saves are read together, because the median is what the system actually produces on a normal week.

    Attributed revenue

    • Orders are reconciled weekly against the back office, not taken from the platform's reported conversions, which count a fraction of what happened.
    • Blended cost per acquisition is read before and after, since the entire point of the work is what paid has to do less of.
    • Assisted conversions are reported separately from direct ones, so that social is neither credited nor blamed for demand it did not create.

    Efficiency

    • Production hours per published asset is the number that tells you whether a system is working, because it is the one thing a team actually controls.
    • Cost per asset at the start of an engagement versus the end separates the value of the system from the value of the people running it.

    What Makes It Work

    • A written position that filtered content instead of decorating it
    • Formats matched to jobs, so performance data was diagnostic rather than ambient
    • Tracking fixed before optimisation, so budget followed a number that was real
    • One system producing all three channels, so output stopped depending on whoever was free

    Conclusion

    The method above holds regardless of who the client is. The figures from this engagement are deliberately not published: the client has not cleared their name or their numbers, and we would rather show the approach than invent a result. When clearance lands, this page is updated in place with the measured numbers and the period they cover.

    More method notes

    All method notes

    Want similar results for your brand?

    Book a Free Strategy Call