Paid Media
Paid media managed for return, not spend
Paid media buys attention now. We plan, build and manage campaigns across Google, Meta, LinkedIn, TikTok, YouTube and Amazon, and we judge everything by the return it brings, not the budget it burns.
What Pay-Per-Click Advertising covers
Most wasted ad spend comes from the same causes: broken tracking, weak structure, no testing plan and reports nobody can read. We fix those first, before scaling anything.
We also cover the surfaces beyond the big platforms: retargeting across web and social, programmatic buying, Connected TV and streaming audio, plus full audits and rebuilds of underperforming accounts.
- Google Ads: Search, Shopping, Display and Performance Max
- Meta, LinkedIn, TikTok and YouTube campaigns
- Amazon and marketplace advertising
- Retargeting and remarketing across web and social
- Programmatic, Connected TV and streaming audio
- PPC audits and account restructuring
Writing on PPC & Paid Media
- How much do Google Ads cost, and what actually drives the number?
Google Ads has no fixed price. Here is the auction mechanic that sets your cost per click, and the levers that move it.
How this is delivered
One person leads every project. Where a job genuinely needs a specialist, I bring in people I have worked with before and manage them, so you get one point of contact and one invoice rather than three suppliers blaming each other.
- You talk to the person responsible for the work, not an account manager
- Specialists are briefed and managed by me, and their work is checked before it reaches you
- One contract, one invoice, one place to chase
Explore each service
PPC & Paid Media explained
How an ad auction actually works
The highest bid does not simply win. Every major ad platform runs an auction that combines what you are willing to pay with how likely the platform thinks your ad is to get the response it is being optimised for. Google calls the combined figure Ad Rank. Meta calls it total value. The names differ, the principle does not: bid multiplied by predicted quality and relevance, plus the expected experience for the person seeing it.
That is why a smaller advertiser with a sharper ad and a matching landing page can outrank a bigger budget. It is also why raising a bid is usually the least efficient fix available. If your predicted click-through rate and relevance are weak, you are paying a surcharge on every impression to compensate for an ad that people do not want. Improving the match between query, ad and page lowers what you pay for the same position.
Two further details change how you should read results. Auctions are per impression, not per day, so your average position and average cost move constantly as competitors enter and leave. And what you pay is set by the competition below you rather than by your own bid, so your maximum bid is a ceiling, not a price.
- Bid sets your ceiling, not the price you actually pay
- Predicted relevance and expected experience act as multipliers on that bid
- A better ad and a matching landing page beat a bigger bid on cost per result
- Costs move by the hour as competitors change their own bidding
Why conversion tracking is the foundation of everything else
Modern ad platforms are optimisation machines pointed at whatever you tell them to value. If the signal you send is wrong, the machine works hard in the wrong direction and the reports look fine while the business does not improve. This is the single most common reason paid media underperforms, and it is almost always fixable.
Broken tracking rarely looks broken. It looks like a page view counted as an enquiry. A thank-you page that fires again on refresh. A conversion counted every time rather than once per person, so one enquiry appears three times. Two tags on the same page double counting. A form that submits without a redirect, so nothing fires at all. In every case the platform learns to buy more of the behaviour you accidentally rewarded.
Getting it right means one primary action that reflects real business value, secondary actions recorded but not bid on, and a reconciliation against your own CRM or order system rather than trusting the platform interface. For anything with a sales process, sending qualified leads or closed deals back to the platform matters most. Optimise towards form fills and you get form fills. Optimise towards deals and the picture of which campaigns are working often changes completely.
- One primary conversion that matches real business value
- Deduplicate so a single enquiry is counted once
- Reconcile platform numbers against your CRM or orders, monthly
- Feed qualified leads and closed deals back where a sales process exists
Campaign structure: why sprawling accounts underperform
Automated bidding needs volume to learn from. Every time you split a campaign, you divide the conversion signal that the bidding depends on. An account with two hundred thinly spread ad groups gives the algorithm two hundred weak signals instead of a handful of strong ones, and performance suffers even though the account looks thorough.
The rule we use is simple. Create a campaign when you genuinely need separate control of budget, geography, bidding or schedule. Group beneath it by shared intent and shared destination. If two queries or two audiences deserve the same message and the same page, they belong together. Sprawl also has a human cost: nobody can manage two hundred ad groups well, so most of them get ignored and quietly waste money.
One structural split is worth making every time: brand and non-brand kept apart. Brand traffic is cheap and converts well because those people were already coming. Blend it into the same campaign as cold traffic and you get a flattering average that hides a poor result on the demand you are actually trying to buy.
- Split only where you need separate budget, location, bidding or scheduling control
- Group by shared intent and shared landing page beneath that
- Keep enough conversion volume per group for bidding to learn
- Always separate brand from non-brand so the reporting stays honest
Demand capture and demand creation are different jobs
Search, Shopping and marketplace ads capture demand that already exists. Somebody types a query, you appear, and the intent was there before you arrived. Social and video ads create demand. Nobody opened Instagram to buy your product, so the ad has to do the work of making them want it. The two behave differently and should never be judged by the same yardstick.
Capture converts faster and reports better, but it is capped by how many people are searching. You cannot buy more demand than the market is generating this month. Creation is slower to prove, harder to attribute and often looks worse on a last-click report, but it is the only way to grow beyond the existing search volume. Accounts that only ever buy capture eventually plateau and then start bidding themselves up on the same finite pool of queries.
Budgeting across the two is a judgement call, not a formula, and anybody quoting you a universal split is guessing. What we do is fund capture to the point where it stops returning well, since that is the cheapest money in the account, then put the growth budget into creation and hold it long enough to see whether capture volume rises as a result. If branded searches and direct traffic climb while the creation campaigns run, that is your answer.
- Capture: search, shopping, marketplaces. High intent, fast feedback, capped by market size
- Creation: social, video, display. Slower proof, under-credited by last click, the route to growth
- Fund capture until returns flatten, then move growth budget into creation
- Watch branded search and direct traffic as a read on whether creation is working
On social platforms, creative is the real lever
Targeting on the big social platforms has been largely automated. The algorithms find the responsive people themselves given a broad pool and a clear conversion signal. What still separates a campaign that works from one that does not is the creative. It is the variable you fully control and the one with the widest range of outcomes.
Testing creative properly means testing concepts, not cosmetics. Changing a button colour or swapping one stock photo for another tells you nothing. Real tests compare different angles: a problem-led hook against a proof-led one, a founder speaking to camera against a customer speaking, a demonstration against a promise. Each of those is a different argument for buying, and the winner usually beats the loser by a margin that no bid adjustment could ever produce.
The discipline matters as much as the ideas. Change one thing at a time, give each test enough budget and time to produce a result you would bet money on, and resist calling a winner after a day. Also plan for fatigue. On social, performance decays as the same people see the same ad repeatedly, so a working ad has a shelf life and the pipeline of new concepts is part of the job rather than an occasional project.
- Test concepts and angles, not colours and crops
- One variable at a time, or you learn nothing you can reuse
- Let tests run long enough to be believable before acting
- Build a steady pipeline of new creative, because winners fatigue
Audience strategy, and its honest limits
The most valuable audience data you have is your own. Customer lists, purchase history, email subscribers and site visitors are first-party data, and they power the two things that reliably work: retargeting people who already showed interest, and building lookalike or similar audiences from your best customers rather than from anyone who ever visited.
The quality of what you upload decides the quality of what comes back. A lookalike built from all purchasers is weaker than one built from repeat purchasers or from your highest value accounts. A retargeting pool of everyone who touched the site is weaker than one segmented by what they actually did. Small, clean and specific beats large and vague, as long as the list is big enough for the platform to work with.
Now the honest part. Audience matching has become less reliable, and it will not go back. This is not about cookies disappearing. Google kept third-party cookies in Chrome and retired most of the Privacy Sandbox proposals. Signal degrades for more ordinary reasons: ad blockers, browser tracking prevention that shortens how long identifiers survive, mobile app tracking permission prompts, and consent requirements that mean a share of visitors are never measurable at all. Retargeting pools fill more slowly and match rates are lower than the numbers people remember from years ago. Server-side tracking and consented first-party data recover some of that. Nothing recovers all of it, and anybody promising otherwise is selling.
- First-party lists are the strongest input you control
- Build lookalikes from your best customers, not all customers
- Segment retargeting by behaviour rather than treating all visitors alike
- Expect lower match rates than historic benchmarks, and plan budget accordingly
Budget, bidding and why constant fiddling hurts
Every automated bidding strategy has a learning period. When you launch a campaign, or make a significant change to it, the system spends a stretch of budget working out who responds before it settles. During that window performance is unstable and unrepresentative. Judging a campaign inside its learning period, then changing it because the early numbers look bad, restarts the learning period and guarantees you never see the settled result.
The changes that reset learning are the big ones: bid strategy swaps, large target moves, budget changes of a serious size, new creative in some setups, and edits to what counts as a conversion. The fix is not to stop optimising. It is to batch changes, make them deliberately, and then leave the campaign alone long enough to produce a verdict. Small target adjustments beat large ones for the same reason.
Budget size matters more than people expect. A campaign needs enough weekly conversions for the bidding to have anything to learn from. Spread a modest budget across six campaigns and none of them ever gets there, so all six underperform. Concentrating the same money into two campaigns usually beats it outright. If the budget cannot support a platform properly, the right answer is to run fewer platforms well.
- Expect an unstable learning period after launch or any major change
- Batch changes, then hold, rather than adjusting something every day
- Move targets in small steps to avoid triggering a reset
- Concentrate budget so each campaign gets enough conversions to learn from
How to read a paid media report honestly
Every attribution model is a simplification of something messier. Last click gives all credit to the final touch, which systematically flatters brand search and retargeting while making the campaigns that created the demand look worthless. Cut the campaign that last click says is useless and the campaign it made look good often gets worse. That is the clearest sign the model was misleading you.
Platform reporting also grades its own homework. Each platform counts conversions it believes it influenced, using its own window and its own rules, so the totals across Google, Meta and your analytics will never agree, and adding them together produces a number larger than your actual sales. View-through conversions deserve particular scepticism: somebody who saw an impression, did not click, and later bought may have been influenced or may simply have been a customer already.
The measurements worth trusting are the ones that survive the platform being switched off. Incrementality testing, whether by geographic holdout, by pausing a campaign and watching total revenue, or by a structured lift test where the platform supports one, answers the only question that matters: what happened that would not have happened anyway. Alongside that, track blended figures. Total spend against total revenue, and cost per acquisition across everything, is harder to argue with than any single platform dashboard.
- Treat last click as one view, not the truth
- Never sum conversions across platforms, they overlap by design
- Discount view-through conversions rather than banking them
- Use holdouts and blended numbers to sanity check platform claims
When paid media is the wrong answer
Paid media amplifies whatever you already have. If the offer is unclear, the pricing is uncompetitive, or the page turns away the traffic it receives, ads make that failure faster and more expensive. Sending more people to a page that does not convert is not a media problem, and no amount of bid management fixes it.
We will tell you to wait when the fundamentals are not ready. That includes sites where nothing is tracked and nobody could tell whether the spend worked. Products with no clear reason to choose them over the obvious alternative. Businesses that cannot handle the enquiries the campaign would generate. And budgets too small to gather a meaningful read on the market, where the money buys noise rather than an answer.
Sometimes the better first move is organic. If your category has steady search demand and your site has no visibility in it, that traffic compounds and does not stop when the invoice does. Sometimes the better first move is fixing the page, the offer or the follow-up process. Ads still have a real advantage worth using: they buy market feedback in days rather than months, which tells you what messages and which queries convert before you commit to a longer content programme. That is a good reason to run a small paid test. It is not a good reason to scale spend before the basics work.
- Fix tracking, offer and landing pages before adding budget
- A small paid test is a fast way to buy market feedback
- Organic often beats paid where search demand is steady and you are invisible
- No agency controls the platform algorithms, so treat any promised outcome with caution
What to expect
- Tracking you can trust, so decisions are grounded
- Budget concentrated on the campaigns that pay
- Plain monthly reporting on return, not vanity metrics
Common questions
How much budget do I need for paid ads?
Enough to gather real data on your market, which varies by platform and competition. We size a test budget honestly on the call, and we will tell you if paid is not the right first move.
Which ad platform is best for my business?
It depends on where your buyers are and how they decide. Search captures existing demand; social creates it; marketplaces win the shelf. We recommend the mix from your goals, not our preferences.
Can you fix an underperforming account?
Usually yes. We audit the account, find the structural and tracking problems, and rebuild it properly. Sometimes the honest answer is a fresh structure, and we will show you why.
Do ads work together with SEO?
Well. Ads bring immediate traffic and fast market feedback while SEO compounds over months. The keyword and conversion data from ads also sharpens the SEO targeting.
Ready to get found?
Book a free visibility call. I will show you where you stand and what to fix first.