These are real engagements described honestly: what the client needed, what we actually changed in the workflow, and what we measured. Where a client has not cleared their numbers for publication we say so rather than inventing a percentage. Most agencies publish the opposite trade.
Legal / personal injury
Injury Map
Official social media, video editing and creative partner
The brief
Injury Map needed a social presence that educated injured claimants without straying into legal advice, in a sector where most competitors post stock-footage quote cards. The requirement was steady educational short-form output, strict compliance language, and creative that looked like a media brand rather than a law directory.
What we did
- — Built a compliance-first script template: every claim framed as general information, no outcome promises, disclaimers burned into the caption rather than buried in the bio.
- — Standardised a talking-head plus B-roll format at 2-4 second shot length so a single filming session yields a month of shorts.
- — Caption styling tuned for sound-off viewing, since a large share of legal-sector viewing happens muted in feed.
- — Monthly reporting focused on watch-through rate and saves — the two signals that correlate with enquiry intent in this niche far better than raw view count.
What the engagement looks like
- Cadence
- Consistent monthly short-form batch
- Format
- Talking head + B-roll, 9:16, burned-in captions
- Primary KPI
- Watch-through rate and saves
- Compliance
- Zero outcome claims, disclaimer on every asset
What we took from it
In regulated sectors, the constraint is the creative advantage. Once the compliance rails were fixed, the team could move faster because nobody was relitigating what could be said in each video.
Media Strategy Lab is the official partner of Injury Map. Published with the client's knowledge; performance figures are not disclosed publicly.
B2B software
B2B SaaS founder channel (anonymised)
Personal brand, done for you — one filming day per month
The brief
A technical founder with a strong point of view and no time. Footage existed only as podcast guest appearances and webinar recordings. The goal was a founder-led LinkedIn and YouTube Shorts presence that produced inbound demo requests rather than applause from other founders.
What we did
- — Repurposing pass first: mined existing webinars and podcast appearances for the strongest 40 seconds of argument before shooting anything new.
- — Moved to a single monthly filming day, batching 20-30 shorts from a scripted shot list built off the previous month's best performers.
- — Hook rewrite discipline: the first line of every video states the problem in the customer's words, never the founder's product language.
- — Long-form YouTube edits run alongside as the depth layer that shorts link back to.
What the engagement looks like
- Output
- 20-30 shorts/month from one filming day
- Turnaround
- 48h per short-form edit after brief approval
- Primary KPI
- Demo requests attributed to social
- Secondary KPI
- Average watch time, follower growth
What we took from it
Founder channels stall on filming logistics, not on editing capacity. Fixing the batch day fixed the pipeline; everything downstream was already solvable.
Anonymised at the client's request. Scope and workflow described accurately; commercial results are covered by NDA.
Media / creator
Podcast show (anonymised)
Podcast-to-shorts clipping and distribution
The brief
A weekly interview show with a solid back catalogue and almost no short-form footprint. Episodes were being clipped inconsistently by different freelancers, so the clips had no visual identity and no reliable hook structure.
What we did
- — Built a clip selection rubric: a clip ships only if it contains a self-contained argument, a stated tension, or a number the viewer would repeat.
- — One caption and lower-third system across every clip so the show became recognisable in feed regardless of guest.
- — Cut density tightened to roughly two cuts per ten seconds, with the dead air and verbal filler stripped rather than kept for authenticity.
- — Back catalogue mined in parallel with new episodes, so the posting calendar was never dependent on the recording schedule.
What the engagement looks like
- Source
- Weekly episodes + back catalogue
- Cut density
- ~2.1 cuts per 10 seconds
- Primary KPI
- 3-second retention and completion rate
- Distribution
- Reels, Shorts and TikTok from one master
What we took from it
Clip volume is not the bottleneck — clip selection is. Halving the number of clips shipped and raising the selection bar produced a better feed than posting everything the transcript allowed.
Anonymised at the client's request. Workflow and production standards described accurately; audience figures are not published.
How we document an engagement
Every case study on this page follows the same internal discipline, and it is worth explaining because it is the reason there are three of them rather than thirty. A study only gets written when the account has been running long enough to separate signal from launch noise — in practice at least three posting cycles — and when the client has cleared what we are allowed to say. If a client will not clear numbers, we publish the workflow and label the numbers withheld. We do not substitute a plausible percentage, and we do not recycle a stock chart with a rising line on it.
The internal version of each study carries more than what is public: the retention curve for every published asset, the hook variant that was tested against it, the reason a format was retired, and the specific edit decisions that changed between month one and month four. That archive is what new editors read during onboarding, and it is why a second account in the same vertical starts faster than the first. The public version strips the client-identifying material and keeps the reasoning.
Production standards that apply across every account
The creative differs wildly between a personal injury firm and a B2B podcast, but the technical floor does not. Verticals are cut and framed at 1080×1920 with safe zones respected for each platform's UI, so a caption never sits under a username overlay on one app because it looked fine in the editor on another. Captions are burned in and legible at arm's length on a phone in daylight, because sound-off viewing is the default case, not the edge case. Dialogue is mixed to a consistent perceived loudness so a viewer does not reach for the volume between your video and the next one in the feed, and music sits far enough under speech that clarity never depends on good headphones.
Every delivery is a pack, not a file: platform masters, the vertical and any square or landscape variant in scope, a caption file where the platform accepts one, thumbnail or cover frame options for long-form, and a short posting brief with the suggested caption text and the reason the hook was chosen. That last line matters more than it sounds — it is the difference between a client posting what they were sent and a client understanding why it was cut that way, which is what makes their next filming day better.
What we measure, and what we deliberately ignore
Four numbers drive the monthly review on every account. The three-second hold tells us whether the opener earned the watch. Average watch time as a share of duration tells us whether the middle of the video kept the promise the hook made. Saves and shares tell us whether the content had utility beyond entertainment, which in professional-services verticals is the closest available proxy for buying intent. And where the client can attribute it — a form field, a booking source, a demo note — we track the enquiries that actually cite social.
What we ignore is just as deliberate. Raw view count is a lagging vanity figure that a single algorithmic push can distort for a fortnight. Follower count is reported but never optimised for directly, because the fastest way to grow a follower number is to post content that attracts people who will never buy anything. Comment volume is read qualitatively, not as a target. And we do not benchmark a client against an unrelated account with a different offer, budget and filming cadence; the only fair comparison is the same account three months ago.
Engagements that did not work, and why
Three patterns account for almost every account that has ended badly. The first is the footage drought: a client signs enthusiastically, films once, and then never schedules the second day. No editing standard survives an empty folder, and after two months of re-cutting the same material the output starts to look like exactly what it is. We now refuse retainers where nobody on the client side owns the filming calendar.
The second is approval by committee. When four people with different opinions must sign off on a thirty-second video, the video ships late and arrives sanded down to something nobody objects to and nobody watches. Single-approver accounts consistently outperform committee accounts in our own data, regardless of sector. The third is the expectation of a spike: a client who needs pipeline this quarter and treats organic short-form as an emergency valve. Organic compounds over months. When someone needs demand next week, the honest answer is paid, and we say so rather than sell a retainer that will disappoint.
