Performance marketing
Paid acquisition is an engineering problem before it is a creative one. We verify the measurement, build an account structure that keeps tests separable, move budget on evidence rather than on a good week, and reconcile platform reporting against your own data.
- Conversion tracking
- Account architecture
- Creative testing
- Incrementality
ADVERTISE
Digital advertising & media
A working advertising practice: strategy, media planning, campaign management, creative testing and reporting across search, social, video and app platforms — run against your own conversion data rather than platform defaults.
Performance marketing is the part of growth where money leaves the business every day, so it deserves the discipline you would apply to production code. A change is written down before it is made. One variable moves at a time. The result is read at a point agreed in advance rather than whenever someone opens the dashboard. Accounts drift when every week brings an unrecorded adjustment, and a drifted account cannot be diagnosed later.
Nothing gets spent until the measurement holds. We trace each conversion from the user action through the tag or SDK to the platform, and then to your own database, checking that the numbers agree closely enough to make decisions on. Where revenue exists, conversions carry a value rather than a flat count, because a bidding system optimising towards undifferentiated events will reliably find the cheapest and least valuable ones.
Structure follows the questions you need answered. Campaigns are separated where a decision depends on the separation, and consolidated where splitting only starves the system of signal. Budgets, geographies, audiences and creative sets are chosen so that a result can be attributed to something specific. When the evidence says the constraint sits in the landing page, the pricing or the onboarding step rather than the ad account, we say so.
What usually brings people here
These are the states an account is normally in when someone decides the current arrangement is not working.
- Spend rises, revenue does not follow
- Reported conversions climb while the bank balance stays flat. Usually the account is optimising towards an event that is easy to fire and easy to duplicate: a page view counted as a lead, a form submission recorded twice, a conversion window long enough to claim credit for demand that already existed. The platform is doing exactly what it was told to do.
- Nobody can explain what changed last month
- Bids, budgets, audiences and creatives were all adjusted in the same week, so no outcome can be traced to a cause. Performance moved and the account holds no record of why. Reconstructing the reasoning afterwards is guesswork, and the same failed test gets proposed again two quarters later as though it were a fresh idea.
- Platform numbers and your dashboard disagree
- The ad platform reports one conversion count, analytics reports another, and finance recognises a third. Each is measuring something slightly different: modelled against observed, click-time against conversion-time, gross against net of refunds and cancellations. Without a documented reconciliation, review meetings become arguments about whose number is real instead of decisions about where money should go.
- Creative decisions made on opinion, not evidence
- Ads are swapped whenever someone senior dislikes one, or refreshed on a calendar rather than on a result. There is no inventory of what has already run, no record of which angle or format won, and no hold period long enough to read anything reliably. Every cycle restarts from zero and the account never accumulates knowledge.
What the engagement covers
The work spans measurement, account structure, creative process, the page the click lands on, and the reporting that ties them together. Scope is agreed in writing; nothing below is assumed.
Measurement and conversion tracking
Before structure or creative, the signal has to be right. We define the conversions, implement or repair the tracking, and verify each one end to end rather than trusting a green tick in a settings screen. Value is attached wherever revenue or margin can be known at conversion time.
- Conversion and event definitions agreed in writing before implementation
- Tag-based or server-side implementation, verified in a debug view
- Conversion values attached where revenue or margin exists
- Deduplication across web, app and offline sources
- Consent state and its effect on measurement documented
- A reconciliation between platform reporting, analytics and your own records
Account architecture
Campaigns are separated where a decision depends on the separation and consolidated where splitting only fragments the signal. Naming, budgets, bidding and exclusions are set deliberately and documented, so someone who inherits the account later can read the reasoning out of it.
- Campaign separation driven by the decisions you actually need to make
- Naming conventions that survive a change of hands
- Bid strategy chosen per campaign, with the assumptions recorded
- Geography, language and device decisions made explicitly
- Negative keyword, placement and audience exclusions maintained, not set once
- Shared budgets and conversion goals mapped so tests stay separable
Creative testing as a structured programme
Once the structure and the tracking are sound, creative is the lever that is left, which is precisely why it should not be pulled on instinct. Each test carries a written hypothesis and a read-out point set before it launches, and every retired asset keeps its result attached.
- A creative inventory: what exists, what has run, what won and what did not
- One hypothesis per test, stated so it could be proven wrong
- One variable per test wherever the volume supports it
- Static, video and copy variants briefed from the same underlying angle
- Read-out point fixed in advance, including the minimum hold period
- Retired creatives archived with their result, so ideas are not recycled blindly
Landing page and product path alignment
A campaign cannot outperform the page it sends people to. We measure the path from click to conversion on the devices and connection speeds your traffic actually uses, then change the parts that are demonstrably costing conversions rather than the parts that are easy to redesign.
- Load performance and layout stability measured on real devices
- Message match between the ad, the page and the offer
- Form, signup and checkout friction reduced where the effect is measurable
- Mobile-first layouts tested against the traffic mix being bought
- Instrumentation added to every step between click and conversion
Budget management, pacing and incrementality
Budget moves in steps with a check between them. We also ask the question an ad account cannot answer about itself: how much of this would have happened anyway. Where volume and geography allow it, we test holdouts or staged regional changes instead of assuming attributed revenue is incremental revenue.
- Stop conditions agreed before a test begins
- Staged budget increases with a defined check between steps
- Brand and generic demand separated so they can be judged differently
- Geo holdouts or staged rollouts where volume makes them readable
- Seasonality and known demand spikes accounted for in the read
Reporting that reconciles
One report, naming its sources and the definition behind every metric. Platform figures sit next to analytics and next to whatever your own system considers a paying customer, with the differences explained rather than averaged away. The underlying data lands somewhere you own.
- A single report that names its sources and their definitions
- Cohort views that follow spend into product behaviour and repeat revenue
- A written read each cycle: what moved, what did not, what remains unknown
- Raw exports into a warehouse or sheet under your control
- Variance between platform and internal numbers stated, not smoothed over
How the work runs
The sequence matters more than the tactics. Spending before the measurement holds produces expensive data you cannot use.
- 01
Audit what already exists
We read the account, the tracking and the path from click to purchase before proposing anything. That includes change history where the platform retains it, current conversion definitions, the pages actually receiving traffic, and what your own database says happened. The audit is written down and shared, including the parts that reflect badly on the plan you arrived with.
- 02
Fix the measurement
Tracking is corrected, re-implemented or built from scratch until a user action appears where it should, once, with a value attached where revenue exists. We test in a debug environment, then against live traffic, then against your own records. Spend does not scale until those three views agree closely enough to decide on.
- 03
Rebuild the structure
Campaigns are restructured so each one answers a question worth asking. Naming, budgets, bidding, geography and exclusions are set deliberately and documented. Where the existing account holds useful conversion history, we migrate rather than reset, because discarding a working learning history is costly in ways that only surface several weeks later.
- 04
Run the cycle
Each cycle follows the same loop: read, hypothesise, make one structured change, hold it, reconcile the numbers, decide. Changes are logged with their reasoning and their date. A cycle that produces no clear answer is reported as inconclusive rather than dressed up as a directional win.
- 05
Scale what holds, stop what does not
Budget rises in steps, with a check between steps, and only on the segments, geographies and creatives that held under scrutiny. Where a channel cannot be made to work at a price the business can afford, we recommend stopping it rather than quietly rebalancing spend so the aggregate report still looks acceptable.
The loop we run every cycle
Six moves, in the same order, every cycle. The value comes from the discipline rather than from the cleverness of any single step.
- 01
Read
We start from the data rather than from a hunch. Spend, conversions, values and cohort behaviour are pulled for the period and compared against the previous one, then checked for anything that is a reporting artefact rather than a real movement.
- 02
Hypothesis
One sentence, written down, stating what we believe is limiting performance and what we expect to happen if it changes. If the hypothesis cannot be phrased in a way that could be proven wrong, it is not ready to be tested with money.
- 03
Structured change
One variable moves. Budget, bid strategy, audience, geography or creative, not several at once and never across every campaign simultaneously. The change is recorded with its date and reasoning so the next reader can trace a performance shift back to a cause.
- 04
Hold period
The change runs untouched for a period agreed before it started, long enough for the bidding system to leave its learning phase and for the conversion window to close. Reacting inside the hold period destroys the reading and leaves the account carrying a change nobody can evaluate.
- 05
Reconcile
Platform reporting is compared against analytics and against your own records. Where they disagree we explain why before drawing any conclusion. A result that exists in only one system is treated as unconfirmed, not as a win worth scaling.
- 06
Decide
Keep, expand, revert or park. The decision is written next to the hypothesis it tested, failures included, so the same idea is not reintroduced next quarter as though it were new. Parked ideas keep their reasoning attached for whoever picks them up.
What has to be true before we spend anything
If an item here is unresolved, we resolve it first. Launching around a gap in this list is how accounts end up unexplainable three months later.
- Conversion tracking verified end to end, from the user action through the platform to your own database
- Conversion values attached wherever revenue exists, so bidding can tell a large order from a small one
- The landing page or in-product path tested on the devices and connection speeds your traffic actually uses
- Budget, pacing and stop conditions agreed in writing before any spend is committed
- Account ownership confirmed: ad, analytics and tag accounts sit under your identity, with us added as a user
- Billing arrangement documented, with media spend charged by the platform to you and kept separate from our fee
- A creative inventory in place: what exists, what has already run, and what can realistically be produced this cycle
- Geography and language decisions made deliberately rather than inherited from a default setting
- A reporting destination agreed, owned by you, that continues to work after the engagement ends
- One named person on your side who can approve a change without convening a committee
What we will not promise
No one can guarantee a return on ad spend, a cost per acquisition, a cost per install, a conversion rate, or that a platform will approve a given ad or account. Anyone offering those is guessing. What we commit to instead is measurement you can verify, an account you own, every change recorded with its reasoning, an honest read each cycle including failures, and a recommendation to stop when the evidence says more spend will not help.
Tools we work in
These names are descriptive references to products we work with day to day. They imply no partnership, sponsorship, certification or endorsement of any kind, and no relationship with the companies that own them.
- Advertising platforms
- Google Ads
- Performance Max
- Google Display Network
- YouTube Ads
- Meta Ads Manager
- Microsoft Advertising
- Apple Search Ads
- Measurement and tagging
- Google Tag Manager
- Server-side Tag Manager
- Google Analytics 4
- Firebase
- Meta Conversions API
- Enhanced conversions
- Consent Mode
- Data and reporting
- BigQuery
- Looker Studio
- SQL
- Python
- Google Sheets
- Google Cloud Storage
- Landing pages and testing
- Next.js
- TypeScript
- Vercel
- WordPress
- Lighthouse
- PageSpeed Insights
- Chrome DevTools
What you are left holding
- Conversion tracking you can verify yourself, not a tag someone installed once
- An account structure where a result can be traced back to a cause
- A creative programme with a record of what has already been tried
- Reporting that reconciles platform numbers against your own data
- Ad, analytics and tag accounts owned by you, with access you control
What people ask before starting
01Can you guarantee a return on ad spend?
No. Return on ad spend depends on your pricing, margin, product, competition and underlying demand, and most of that sits outside an ad account. What we commit to is measurement you can check, structured tests, decisions recorded with their reasoning, and an honest read every cycle. If a channel cannot reach a price the business can afford, we will say so rather than keep spending.
02Is media spend included in your fee?
No. Media spend is charged by the advertising platform to your own payment method and stays entirely separate from what we charge for the work. You see exactly what was spent and on which campaign, without it passing through us. Our fee covers measurement, account structure, creative testing, landing page work and reporting, and it is agreed in writing rather than taken as an unstated margin on media.
03Who owns the ad accounts and the data?
You do. Ad accounts, analytics properties, tag containers and conversion history are created under your business identity where they do not already exist, and we work inside them as a user with the access the job requires. If the engagement ends, we remove our access and everything stays where it is, including the tracking implementation, the reporting and the learning history.
04What drives the cost of this work up or down?
Three things mostly. The number of platforms being run, the state of the existing measurement, and how much creative and landing page work is required. An account with clean tracking and a working page costs less to run than one where every conversion has to be rebuilt first. Multiple geographies, languages and product lines each add structure, and structure adds review time.
05What happens if paid acquisition simply does not work for us?
We say so and recommend stopping. Some products cannot be acquired profitably through paid channels at their current price, margin or conversion rate, and no amount of bid tuning changes that arithmetic. When the evidence points there, the useful advice is to fix pricing, onboarding or retention first, or to move budget to a channel that suits the product better.
06What will you not do?
We will not run spend against tracking we cannot verify, buy traffic from sources that inflate clicks, or present modelled platform conversions as confirmed revenue. We do not claim partnership, certification or preferred status with any advertising platform, because we hold none. We will not make account changes without recording them, and we will not keep a test running because stopping it would look like an admission.
What this work usually connects to
Talk about your paid acquisition
Send us the account you are running now, or the product you are about to start spending against. We will read what exists first and tell you plainly whether paid acquisition is the right next move.
Directcontact@mnfinfotech.com