Meta Ads

Creative is the targeting.
Everything else is settings.

Meta’s machine learning now finds your buyers better than any manual audience you could build. What it cannot do is make people stop scrolling. We treat creative as the primary performance lever, because on Facebook and Instagram it is.

Shop NowREACH1.2MROAS5.2xCPA−44%
Creative production
Broad targeting
Conversions API
Incrementality
01
Why accounts stall

Interest stacking stopped working. Most accounts never noticed.

Meta’s delivery system now identifies buyers from conversion signals more accurately than any interest combination a media buyer can assemble. Yet accounts still run dozens of narrow ad sets fragmenting the same budget, each too small to exit the learning phase, all served the same three creatives that stopped working two months ago.

Too many ad sets, too little data

Fragmenting budget across narrow audiences keeps every ad set in the learning phase, where delivery is least efficient. Consolidation usually improves performance immediately.

Creative fatigue read as rising costs

Performance decays with frequency. Without a production and retirement cadence, accounts pay progressively more for the same declining attention and blame the auction.

Browser-only tracking

Since iOS privacy changes, pixel-only setups lose a large share of conversion signal. The algorithm optimises on partial data and reported results understate what actually happened.

Retargeting mistaken for performance

Retargeting reports the best returns in nearly every account because it reaches people who already decided. It is also where over-investment almost always hides.

5.1×
Target ROAS achieved
−36%
Typical CPA reduction
40+
Creatives tested monthly
8+
Years running Meta Ads
02
What we actually do

Consolidate the structure. Industrialise the creative.

Broad targeting with strong conversion signals now outperforms hand-built interest stacks in most accounts. That frees the work to go where it matters — producing enough creative variety for the algorithm to optimise against.

We run a standing production and testing calendar rather than commissioning assets when performance dips. By then you have already paid for weeks of decay.

  • Consolidated account structure — fewer, larger ad sets so budget concentrates and campaigns exit the learning phase quickly.
  • Creative production system — modular assets in volume — hooks, formats and angles varied systematically rather than one polished hero video.
  • Conversions API — server-side events so the algorithm optimises on complete data rather than what survives browser restrictions.
  • Fatigue monitoring — frequency and creative decay tracked with retirement rules, so assets are replaced before costs climb.
  • Incrementality holdouts — geo tests establishing what Meta spend genuinely caused, separating real lift from claimed credit.
Interest-stack era
How Meta works now
30 narrow ad sets
Few consolidated ones
Manual interest targeting
Broad + strong signals
3 creatives per quarter
40+ tested monthly
Pixel only
Conversions API
Retargeting-heavy
Prospecting-led
Platform ROAS trusted
Holdout tested

The uncomfortable finding: in most incrementality tests we run, retargeting produces far less additional revenue than its reported ROAS suggests. Those people were coming back anyway. The spend was buying credit, not customers.

03
The Sachkhand blueprint

How we rebuild a Meta Ads account

Six workstreams, signal and structure first — because creative testing on broken tracking teaches you nothing reliable.

01

Conversions API setup

Server-side event tracking with proper deduplication and event match quality, so delivery optimises on complete rather than partial conversion data.

02

Structure consolidation

Narrow ad sets merged so budget concentrates and campaigns exit the learning phase, where delivery efficiency is at its worst.

03

Creative production

A standing calendar of modular assets — varied hooks, formats and angles — produced at volume rather than commissioned reactively when results dip.

04

Testing framework

Structured creative tests with defined win conditions, sample sizes and retirement rules, so results are decisions rather than opinions.

05

Fatigue management

Frequency and decay curves monitored per asset, with replacement triggered before costs climb rather than after they already have.

06

Incrementality testing

Geo holdouts on prospecting and retargeting to measure genuine lift, because attribution windows credit Meta for conversions it did not cause.

04
What is changing through 2026

Four shifts in Meta advertising

Automation absorbed targeting and bidding. Creative volume and signal quality are what separate accounts now.

01

Broad targeting beats manual audiences

Advantage+ and broad delivery consistently match or exceed hand-built interest stacks when conversion signals are strong. The buyer's job shifted from finding the audience to feeding the system.

02

Creative volume is the constraint

Delivery needs variety to optimise against. Accounts producing a few assets a quarter starve it, then attribute the resulting decline to rising auction costs.

03

Server-side tracking is mandatory

Browser-only pixels lose a substantial share of conversions. Conversions API is no longer an enhancement — it is the difference between the algorithm learning and guessing.

04

Short vertical video dominates delivery

Reels and vertical placements now take a large share of impressions and cost less per result. Repurposed landscape assets underperform natively produced vertical creative.

05
Why Sachkhand

We produce creative, not just manage settings

On Meta, the account settings are quickly optimal and then stay there. Creative is where performance is actually won or lost.

Creative at volume

Modular production so dozens of variations ship monthly without a full shoot each time. Volume is what the delivery system needs to work.

Conversions API done properly

Server-side events with deduplication and high event match quality, so optimisation runs on complete data instead of fragments.

Incrementality over attribution

We run geo holdouts. Attribution windows credit Meta for conversions it did not cause, and holdouts are the only way to know the difference.

Fatigue caught early

Decay curves monitored per asset so creative is replaced before costs climb, rather than after a month of paying more for less.

We will tell you to cut retargeting

If holdouts show it is not incremental, we say so — even though it reduces the budget we manage and worsens reported ROAS.

Planned with other channels

Meta budgeted against blended efficiency rather than platform ROAS, so it complements search rather than competing for the same credit.

06
The Honest Comparison

Why Businesses Choose SDM for Meta Ads

Same budget, very different outcomes. Point by point, here’s how a specialist partner compares to a typical agency or going it alone.

What actually matters With SDM Typical Agency In-House / DIY
Senior specialist on your accountAlwaysOften a juniorStretched thin
Custom strategy built for your goalsTailoredTemplatedGuesswork
Deep audit before any work beginsSurface-levelSkipped
White-hat, penalty-safe methodsGuaranteedVariesHigh risk
Plain-English reporting tied to revenueMonthlyJargon PDFsNone
Direct access to your specialistAccount-manager relayN/A
Creative produced, not just managedClient suppliesAd-hoc
Conversions API configuredPixel onlyDefault
Structure consolidatedFragmentedFragmented
Fatigue monitored per assetReactiveUnnoticed
Incrementality holdouts runPlatform ROASNever
Ongoing competitor gap analysisOne-offManual
Conversion-focused, not just trafficTraffic-firstUnclear
Premium tools included (Ahrefs, SEMrush)SometimesCostly extra
No long lock-in contractsFlexible6–12 mo lock-inN/A
Established agency, operating since 2017VariesLearning curve
Fast onboarding & early quick wins~2 weeksSlowTrial & error
Human, SEO-led content (no AI spam)Outsourced / AI spamTime-heavy
Focus on compounding, long-term ROICore promiseShort-term winsSlow & ad-hoc
Recovery from Google penaltiesSometimesVery hard

20 reasons growing brands make the switch. See the difference for yourself →

07
Measurement

Platform ROAS is a diagnostic, not a result

Meta counts conversions it touched within its attribution window. We report what it actually caused.

Incremental ROAS

Holdout-tested return, which is consistently lower than platform-reported ROAS and consistently more useful.

Blended CPA

Total spend against total new customers from your own systems, so no attribution window can inflate it.

Creative win rate

Share of tested assets that beat control, and how quickly winners decay under frequency.

Frequency & decay

How fast each asset loses efficiency, which sets the replacement cadence rather than guesswork.

New vs returning

How much spend acquires customers versus re-reaching existing ones — where over-investment almost always hides.

Meta Ads performance
All campaigns — rolling 90 days
LIVE
ROAS
5.1x
CPA
−36%
Creative win
47%
Prospecting
74%
Match qual.
8.9/10
↑ 5.1x ROAS with 74% of spend on prospecting
08
The engagement

Four stages, repeated every month

Fix signal, consolidate structure, then run creative as a production line.

01

Signal & structure

Conversions API implemented with proper deduplication, then narrow ad sets consolidated so budget concentrates and campaigns exit the learning phase.

→ Delivery optimising on real data
02

Creative baseline

Current assets audited for fatigue, winning angles identified and a modular production system built so variations can ship weekly.

→ A production line, not a shoot
03

Structured testing

Hooks, formats and angles tested with defined win conditions. Losers retired, winners varied further, learnings documented rather than remembered.

→ Compounding creative library
04

Test lift & scale

Geo holdouts on prospecting and retargeting to establish real incrementality, then budget scaled against marginal CPA.

→ Growth verified, not assumed
09
Outcomes

The targets we set for Meta Ads engagements

Targets we plan against for ecommerce, DTC and lead generation clients after six months.

5.1×
Platform ROAS
From 2.4× at start
−36%
Cost per acquisition
At higher volume
47%
Creative win rate
Assets beating control
74%
Spend on prospecting
Rebalanced from retargeting

Targets rather than best cases. Note that incremental ROAS is always lower than the platform figure — we report both, and budget decisions are made on the incremental one. These are planning targets rather than a record of past client averages; we agree a realistic range against your own baseline before an engagement starts.

10
Scenarios

Meta Ads in action — three sectors

Different models, the same lever: more creative, better signal, honest measurement. The scenarios below are composites drawn from situations we encounter repeatedly — they illustrate method, not the account of any single named client.

Ecommerce

Thirty-one ad sets, three creatives

Challenge

Budget was fragmented across narrow interest audiences, none exiting the learning phase, all serving the same three ageing assets.

What we did

Consolidated to four ad sets on broad targeting, implemented Conversions API and started a weekly creative production cadence.

5.4×
ROAS
−41%
CPA
DTC brand

Retargeting that was not incremental

Challenge

Retargeting reported an 11× return and had been scaled aggressively on that basis, while overall business growth stayed flat.

What we did

Ran a geo holdout. Most of that revenue occurred without the ads, so budget shifted into prospecting creative.

+83%
New customers
0
Extra spend
Lead generation

Optimising toward the cheapest leads

Challenge

Campaigns optimised for form fills produced high volume at low cost, but sales reported that almost none were qualified.

What we did

Fed CRM outcomes back through Conversions API so only qualified leads counted, then rebuilt creative around qualifying rather than maximising response.

+96%
Qualified leads
−27%
Cost per SQL
11
Questions

The questions we get asked most

Straight answers on creative, targeting and what Meta reporting actually means.

Should we still use detailed interest targeting?+

Rarely. With strong conversion signals, Meta’s broad delivery consistently matches or beats hand-built interest stacks, and narrow ad sets fragment budget so nothing exits the learning phase. We test it rather than assume, but consolidation onto broad targeting improves performance in most accounts we take over.

How much creative do we actually need?+

Considerably more than most brands produce — typically dozens of variations monthly rather than a few per quarter. Delivery needs variety to optimise against, and every asset decays with frequency. We build modular production so variations ship without a full shoot each time.

Why is our ROAS falling when nothing changed?+

Usually creative fatigue. Frequency climbs, the same people see the same ads, response drops and cost per result rises. It reads as an auction problem but is almost always a creative supply problem, and it is fixed by production cadence rather than bid adjustments.

Is the Conversions API really necessary?+

Yes. Browser-only tracking loses a substantial share of conversions, which means delivery optimises on partial data and reporting understates results. Server-side events with proper deduplication and high match quality typically improve both actual and reported performance.

How much should we spend on retargeting?+

Less than most accounts do. Retargeting reaches people who already decided, so it reports excellent returns while frequently adding little genuinely incremental revenue. A geo holdout settles it in a few weeks, and the result usually justifies moving budget into prospecting.

Can you produce the creative or do we supply it?+

We produce it. Managing Meta without controlling creative means managing settings that are already close to optimal. We can work with your existing brand assets and guidelines, but the production and testing cadence needs to sit with whoever is accountable for performance.

How quickly will we see results?+

Signal and structure fixes show within two to four weeks as delivery relearns. Creative testing compounds from the second month as the library of proven angles builds. Incrementality tests need a few weeks each, so a complete picture usually takes a quarter.

Next step

Ready to find out what Meta actually earns you?

We will audit your creative fatigue, tracking setup and account structure, and design a holdout test that shows how much of your reported ROAS is genuinely incremental.

Free Meta Ads audit
Creative fatigue analysis
Conversions API review
Incrementality test design
PLATFORM ROAS5.1x
Chat With Us
🚀 Book Free Audit & consultation!
45 min · Free audit + strategy + consultation · No obligation