— 09 media & performance channel-agnostic, thesis-led

we don't run ads. we run a system.

a thesis for what each channel is for, a scoreboard for what good looks like, and a kill-switch when it isn't. creative built for the feed, budget allocated by job, and a weekly read that says one of three things: scale, hold, or kill.

audit your media → see how it works ↗ replies in < 24h
— minimum spend
$25k/mo media
— review cadence
weekly read, monthly board
— creative output
20–40 ads / month
— attribution
mer + incrementality
01. why this matters now

most paid programs are spending, not buying.

the platforms got better at extracting budget. the buyers got better at ignoring ads. and most agencies still send the same monthly deck full of vanity dashboards and "we've optimised the campaign."

a paid program that works in 2026 isn't a media-buying engagement — it's three jobs at once: a creative shop that ships ad-native work, a media team that knows what each channel is actually for, and a measurement layer that tells you which dollars made the next dollar. most agencies do one of those well, two badly, and bill for all three.

we run paid the way a category challenger runs paid: with a thesis on every channel, a creative volume that makes platform algorithms stop guessing, and an attribution stance that survives ios updates and consent windows. not "we'll get you ROAS." roas of what, against what baseline, in which window.

and we make the kill calls. if a channel doesn't pay back inside the agreed window, we move the budget. no agency that earns a percentage of media will ever say "spend less here." we will. often.

~38%
— of paid budget spent on "branded search" that customers were going to buy anyway
3:1
— typical creative-to-buyer ratio. ads die because the creative is tired, not the channel
+24%
— mer lift from cutting the bottom-quintile spend, not adding more
0
— agencies incentivised to tell you to spend less when they bill on % of media
02. how we think about media

four operating rules every account runs on.

we charge a flat fee, not a percentage of media — so when we say "spend less here," we mean it. our incentive is your marginal efficiency, not your invoice.

01
— rule 01

every channel needs a job. write it on the wall.

tiktok is for discovery. google brand-search is for capture. meta is split. youtube is for weight. amazon ads are for shelf. if you can't say in one sentence what a channel is for, you're spending without buying. we write the thesis up front, and pace against it.

— what we audit first
channel mix vs. role — how much of your spend is doing capture work vs. demand-creation. usually the ratio is wrong by half.
02
— rule 02

creative is the algorithm's favourite variable.

the platforms have solved targeting. they haven't solved why anyone should care. we ship 20–40 ads a month built for the feed, not the brand deck — with one or two new angles tested every week. creative is the lever; everything else is the floor.

— what we make
hooks, ugc, native talking-head, static testers, motion variants. shot for the platform, edited for the platform, written for the platform.
03
— rule 03

measure the marginal, not the average.

average ROAS is a story. marginal ROAS is the truth. we track the incremental return on the next thousand dollars, not the blended report card. and we run quarterly incrementality holdouts — not modelled, real — to know what the channel is actually adding vs. what was going to happen anyway.

— what's in the read
marginal CAC by channel, MER vs. forecast, holdout-validated incrementality, brand-vs-non-brand split, cohort payback curves.
04
— rule 04

pull the kill-switch. early. often.

every channel and creative has a pre-agreed kill threshold: spend window, learning window, payback window. if it doesn't clear, it's dead — the budget moves. most paid programs die from optimism, not bad ideas. we kill quickly so we can spend on what's working.

— what gets killed
ads that don't clear at 7-day spend window, channels that miss MER target two weeks running, audiences below 0.6× index. no "give it more time."
03. the media-mix scoreboard

every channel, given a job and a number.

this is the operating doc we run for every account — one row per channel, the role it plays, the kpi we hold it to, the benchmark, and the budget share. no channel sits on the page without a job and a number it has to clear.

— channel
role
primary kpi
benchmark
budget share
status
tiktok adscreator + spark
discovery
3-second VTR+ thumb-stop rate
≥ 32%
22%
scale
meta — prospectingadvantage+ + manual
consider
marginal CAC7-day click + 1-day view
≤ $42
28%
hold
meta — retargetingsite + engagers
convert
ROAScap'd, decay-adjusted
≥ 4.2×
14%
scale
google — non-brand searchcommercial intent
consider
CAC + LTV cohort90-day payback
≤ $58
12%
hold
google — brand searchdefensive only
capture
incremental liftvs. organic baseline
≥ 18%
6%
hold
youtube adsin-stream + shorts
brand weight
brand-search lift+ assisted conv
≥ +12%
10%
scale
amazon adssponsored + DSP
shelf
TACoStotal advertising cost of sale
≤ 11%
8%
scale
7
channels with a stated job
100%
budget covered, no orphans
1.8×
avg MER vs. category benchmark
3:1
creative-to-buyer ratio held
04. how we wire it up

six weeks from audit to first weekly read.

we don't take over a media account on day one. we audit, we model, we get the creative engine started, then we go live with a scaled-down learning budget — and ramp once the read is honest.

01
— week 01–02audit + thesis

what's working, what's burning, what's missing.

we pull the last 12 months of every channel, strip out branded search, run a marginal-efficiency curve on each line, and write the channel thesis — what each platform is for, what the kpi is, what the kill threshold is. delivered as a one-page board doc, not a 60-slide deck.

spend audit channel thesis scoreboard v1
02
— week 02–03creative engine

the first 30 ads, on a hook framework.

we don't ship a brand deck as an ad. we ship 30 native units across formats — ugc, talking-head, hook-driven static, motion variants — built off a hook framework (problem-first, contrarian, demonstration, social-proof, list). every unit gets a tracked id so we can see which angle pays back.

hook framework 30 ad units tracking taxonomy
03
— week 03–04measurement layer

your numbers, in one place, daily.

we set up a single source-of-truth dashboard that pulls every channel + your shopify/stripe/ga4 + a marginal-cac model on top. by friday, you can open one tab and see what's pacing, what's burning, what shipped today, and which creative concept is breaking the line.

unified dashboard marginal-cac model creative tracker
04
— week 04go live, small

50% of normal spend, learning mode.

we don't ramp on day one. we go live at half your usual spend for two weeks — just enough to validate the model, calibrate the kill thresholds, and see which creatives have legs. then we scale into what's working, not what was working last quarter.

learning budget kill calibration scale plan
05
— week 05–06scale + cadence

the weekly read becomes the rhythm.

monday: weekly read — pacing, MER, creative refresh signals, three calls (scale / hold / kill). thursday: creative shipout — new units live, old units paused. monthly: incrementality holdout review, scoreboard v2. by week six, the program runs on rails — and you stop checking the ads manager at midnight.

monday read thursday ship monthly board

forty-seven million in managed media. flat-fee, every dollar.

— 01 / under management
$47M
media managed across portfolio in the last 12 months — flat-fee, no % of spend
— 02 / output
~28
ads per account per month — built for the feed, not retrofitted from brand work
— 03 / efficiency
+38%
avg MER lift in first 90 days vs. previous engagement
— 04 / discipline
~17%
of accounts have recommended a spend cut in the last quarter
"

first agency that's ever told us to spend less. and revenue went up.

D
deepak r.
— head of growth · dtc supplements
05. faq

questions, answered straight.

can't find it here? email hi@socialmafia.agency — replies in under 24 hours.

do you charge a percentage of media?
no — ever. we charge a flat monthly fee scaled to scope (channel count, creative volume, attribution complexity). a percentage-of-media model puts our incentive against yours; we'd never recommend you spend less because we'd earn less. flat fee fixes that.
what's the minimum media spend?
$25k/month in working media. below that, the creative engine, measurement layer, and weekly read aren't worth the overhead — you'd be better off with a freelance buyer + good creative templates. we'll happily refer you.
do you work on commission or in-house ad accounts?
your accounts, your name on every contract, your billing. we operate inside your ad accounts with vendor access — never our own. when the engagement ends, you keep every campaign, every audience, every learning. zero handover friction by design.
how do you handle attribution post-iOS / cookie deprecation?
we run a three-layer stack: platform-reported (with appropriate skepticism), MER vs. forecast (the truth), and quarterly incrementality holdouts (the validation). we don't rely on platform ROAS alone — that number is a marketing aid, not a financial one.
who makes the creative? do we need a separate creative agency?
we make the creative. our paid team includes ad-native creators, editors, and a copy lead — built for performance, not brand campaigns. if you have a brand agency, we work alongside them: they make hero work, we make the 30 units that test it in market.
how fast do you typically see results?
creative volume + measurement layer takes 4–6 weeks to stabilise. the read is honest by week six. material MER shifts usually show inside 90 days. if nothing has moved by month four, the strategy is wrong — and we'll tell you, not paper over it.
what platforms do you actually work on?
meta, tiktok, youtube, google, amazon ads, snapchat, pinterest, linkedin (b2b), reddit, applovin (mobile), netflix/prime ad-tier (enterprise). we'll refuse to run channels that don't fit your category — so don't expect a wishlist. expect a thesis.

stop spending. start buying.

— flat fee · channel-agnostic · weekly read · kill-switch first