Online Advertising Explained: What It Is and How It Works

Online advertising is the practice of paying to place promotional messages in front of audiences across digital channels, including search engines, social platforms, websites, and apps. It operates on a range of pricing models, from cost-per-click to cost-per-thousand impressions, and spans formats from text-based search ads to video, display, and native placements. If you are spending money to reach people online, you are doing online advertising.

That definition sounds simple. The execution is not. I have managed hundreds of millions in ad spend across more than 30 industries, and the gap between understanding what online advertising is and knowing how to make it work commercially is wider than most people expect when they start.

What Online Advertising Actually Covers

When most people say “online advertising,” they mean Google and Facebook. That is understandable, since those two platforms account for the majority of digital ad spend globally. But the category is considerably broader, and understanding the full landscape matters if you are making channel allocation decisions.

Search advertising sits at one end of the spectrum. You bid on keywords, your ad appears when someone searches for those terms, and you pay when they click. It is the closest thing online advertising has to intent-based targeting, because the user has already signalled what they want. Google Adwords, now rebranded as Google Ads, pioneered this model and it remains the backbone of paid search for most businesses.

Display advertising works differently. You pay to place banner or rich media ads across a network of websites, targeting by audience profile, context, or behaviour rather than search intent. It is better suited to building awareness or retargeting people who have already visited your site than to capturing bottom-of-funnel demand.

Social advertising has grown into its own substantial category. Meta, LinkedIn, X, Pinterest, and Snapchat all offer paid placements, each with different audience profiles and ad formats. Then there is TikTok Ads, which has moved from niche to mainstream fast enough that ignoring it is now a real strategic choice rather than a default.

Video advertising, programmatic buying, native advertising, podcast sponsorships, connected TV, and in-app advertising round out the broader category. Each has its own mechanics, pricing logic, and audience behaviour. The fact that they all sit under the umbrella of “online advertising” does not mean they behave the same way or require the same skills to run well.

If you want a broader framework for how paid channels fit together strategically, the Paid Advertising Master Hub covers the full picture, from search to social to programmatic.

How Online Advertising Pricing Models Work

Before you can evaluate whether a campaign is working, you need to understand what you are actually paying for. Online advertising uses several pricing models, and they are not interchangeable.

Cost-per-click (CPC) means you pay each time someone clicks your ad. It is the standard model for search advertising and is used across many social platforms. The appeal is obvious: you only pay for engagement, not just eyeballs. The risk is that clicks do not equal conversions, and without proper tracking, CPC campaigns can look productive while delivering very little revenue.

Cost-per-thousand-impressions (CPM) means you pay for every thousand times your ad is shown, regardless of whether anyone clicks. This model is common in display and video advertising, and it makes sense when your goal is reach and awareness rather than direct response. The challenge is that impressions are easy to buy and hard to evaluate. A CPM campaign that reaches the wrong audience at scale is just expensive noise.

Cost-per-acquisition (CPA) is the model most performance marketers prefer in theory, because you only pay when a specific action occurs, whether that is a purchase, a lead form submission, or a sign-up. In practice, CPA bidding on platforms like Google Ads is a target rather than a guarantee. You are telling the algorithm what you want to pay per acquisition, and the platform optimises toward that goal. It works well when you have enough conversion data for the algorithm to learn from. It works poorly when you are starting from scratch or operating in a low-volume category.

There are also hybrid models. Return on ad spend (ROAS) targets are common in e-commerce, where you set a revenue multiple you want to achieve for every pound or dollar spent. Value-based bidding is the more sophisticated version of this, where you feed the platform data about the relative value of different customer types rather than treating all conversions equally.

Understanding Google advertising fees in detail is worth doing before you commit significant budget to paid search, because the relationship between bid strategy, quality score, and actual cost-per-click is not always intuitive.

The Difference Between Demand Capture and Demand Creation

This is the distinction that most online advertising conversations skip over, and it is the one that matters most when you are deciding where to put your budget.

Search advertising, at its core, is demand capture. Someone already wants something. They type it into Google. Your ad appears. You intercept that existing intent and try to convert it. The category has to exist for this to work. If nobody is searching for what you sell, paid search will not manufacture that demand for you. It will just return very little traffic at whatever you bid.

Social and display advertising, by contrast, are demand creation or demand amplification tools. You are reaching people who have not necessarily expressed intent to buy. You are interrupting their experience to make them aware of something, or to remind them of something they looked at before. The conversion rates are lower, the path to purchase is longer, and the measurement is harder. But for businesses in categories with limited search volume, or for brands trying to grow beyond their existing audience, these channels are essential.

Early in my career at lastminute.com, I launched a paid search campaign for a music festival. The campaign was structurally simple: a small set of tightly matched keywords, a clear offer, and a landing page that did not get in the way of the purchase. Within roughly a day, we had generated six figures of revenue. That result was not a miracle of creative brilliance. It was the result of placing a relevant offer in front of people who were already looking for it. Search advertising at its most efficient is almost mechanical when the intent is there.

The lesson I took from that, and from hundreds of campaigns since, is that the channel choice should follow the demand landscape. Start with search if the search volume exists. Use social and display to extend reach, build consideration, and retarget. Do not use social advertising to do the job of search advertising, and do not use search advertising to build brand awareness. They are different tools for different jobs.

This distinction also shapes how you measure success. A search campaign that does not convert is a problem. A brand awareness campaign on social that does not drive immediate conversions might still be working perfectly well, if you are measuring the right things.

How Search Advertising Works in Practice

Paid search is an auction. Every time someone performs a search, Google runs an instantaneous auction to determine which ads appear and in what order. Your position is not determined solely by how much you bid. It is determined by a combination of your bid and your Quality Score, which is Google’s assessment of how relevant your ad and landing page are to the search query.

This means that a well-structured campaign with highly relevant ads and a good landing page can outrank a competitor who is bidding more. It also means that a poorly structured campaign can spend a lot of money for mediocre results, even with a generous budget. Keyword match types are central to this, controlling how closely a search query needs to match your keyword before your ad is eligible to show.

Broad match, phrase match, and exact match each carry different trade-offs between reach and precision. Broad match gives you scale but can serve your ads against searches that have little to do with your business. Exact match keeps you precise but limits volume. Most campaigns use a combination, with negative keywords doing the essential work of filtering out irrelevant traffic.

Geographic targeting adds another layer of control. The ability to target ads to specific regions, cities, or even postcodes has been part of the platform for a long time, and regional targeting in Google Ads remains one of its most commercially useful features for businesses with defined geographic markets.

Campaign structure matters more than most people give it credit for. I have audited dozens of accounts over the years where the core problem was not the budget or the bids but the architecture. Ad groups containing hundreds of loosely related keywords, ads that did not match the search intent of the group, landing pages that sent everyone to the homepage regardless of what they searched for. These structural problems compound over time, and they are invisible in headline metrics until you dig into the account properly.

Automation has changed how campaigns are managed. Smart bidding, responsive search ads, and Performance Max campaigns have shifted significant control from the advertiser to Google’s algorithm. This is not inherently bad, but it requires a different kind of oversight. The algorithm optimises toward the goal you set. If the goal is wrong, or if your conversion tracking is broken, the algorithm will optimise toward the wrong thing with considerable efficiency.

AI-assisted campaign management is increasingly the norm rather than the exception. Running better Google Ads campaigns with AI is a practical reality now, but the quality of the output still depends on the quality of the inputs, the goals you set, the data you feed in, and the strategic judgement you apply to the results.

Social Advertising: Reach, Targeting, and the Attention Economy

Social advertising operates on a fundamentally different premise from search. You are not responding to expressed intent. You are placing content in front of people based on who they are, what they have done, or what they look like relative to your existing customers. The targeting sophistication is considerable, and the creative demands are higher, because you are asking people to pay attention to something they did not ask for.

Meta remains the dominant platform for most consumer advertisers, with audience scale and targeting depth that is difficult to match. LinkedIn is the default choice for B2B advertisers targeting by job title, seniority, or industry, despite its comparatively high cost-per-click. TikTok has built a genuinely different advertising environment, one where the creative format is more demanding but where engagement rates in the right categories can be significantly higher than on older platforms.

The creative brief for social advertising is different from search. In search, the ad is a response to a query. Relevance is built in. In social, the ad is an interruption. It has to earn attention before it can deliver a message. This means the first frame of a video, the first line of copy, and the visual treatment of an image all carry disproportionate weight. Many social campaigns underperform not because of targeting errors but because the creative does not stop the scroll.

Retargeting sits across both search and social and is one of the most commercially efficient uses of online advertising budget for most businesses. You are reaching people who have already expressed some level of interest, either by visiting your site, engaging with your content, or abandoning a cart. The conversion rates are higher, the CPCs are often lower on a relative basis, and the message can be more specific because you know something about where the person is in their consideration process.

Lookalike audiences extend this logic. You take a list of your existing customers, feed it to a platform, and ask it to find people who share similar characteristics. When it works well, it is one of the more efficient prospecting tools available. When it works poorly, it is usually because the seed audience was too small, too broad, or not representative of your best customers.

What Good Campaign Management Actually Looks Like

There is a version of online advertising management that is largely administrative: checking dashboards, adjusting bids, writing new ad variations, reporting on last week’s numbers. There is another version that is genuinely strategic: asking whether the campaign is structured correctly for the business goal, whether the attribution model is telling a coherent story, whether the budget is allocated to the highest-return activity, and whether the channel mix makes sense given where demand actually lives.

The administrative version is common. The strategic version is rarer and more valuable. When I was growing the agency at iProspect from around 20 people to over 100, one of the consistent differentiators we tried to build was the habit of asking the commercial question first. Not “how do we optimise this campaign?” but “what is this campaign actually supposed to do for the business, and is it doing that?”

That question sounds obvious. In practice, it gets lost quickly, especially when clients are focused on platform metrics and agencies are focused on retention. A campaign that is performing well on click-through rate but generating no profitable revenue is not a good campaign. A campaign with a mediocre CTR that is delivering consistent, profitable customer acquisition is doing its job.

For businesses that are not ready to build this capability in-house, working with a specialist is often the right call. Understanding what PPC management services actually include, and what they cost, is a useful starting point before engaging any external partner.

If you are considering bringing in an agency specifically, it is worth understanding how these relationships are structured. What a PPC agency does, how they charge, and what questions to ask before signing anything will save you significant time and money.

The practical mechanics of good campaign management include regular search term reviews to catch irrelevant traffic, bid adjustments based on device and time-of-day performance data, landing page testing to improve conversion rates, and audience segmentation to ensure messaging is matched to intent. None of these are glamorous. All of them compound over time.

The Measurement Problem Nobody Talks About Honestly

Online advertising is supposed to be the measurable channel. Compared to TV or outdoor, the feedback loop is faster and the data is richer. But measurable is not the same as accurately measured, and this distinction gets glossed over constantly.

Attribution is the central problem. Most businesses use last-click attribution as their default, which means the final touchpoint before a conversion gets all the credit. This systematically overstates the contribution of bottom-of-funnel channels like branded search and retargeting, and understates the contribution of everything that built awareness and consideration earlier in the process. It is a convenient fiction that makes performance look better than it is for certain channels and worse than it is for others.

I have judged the Effie Awards, which evaluate marketing effectiveness with real rigour. The campaigns that stand up to scrutiny are the ones where the advertiser has thought carefully about what they are measuring and why, and where the measurement methodology matches the campaign objective. The campaigns that fall apart are the ones where impressive-looking platform metrics are presented as evidence of business impact without any connecting logic.

Platform attribution and business attribution are different things. Google Analytics, Meta Ads Manager, and Google Ads will each tell you a version of the same story, and those versions will often disagree with each other. They disagree because they are each measuring from their own perspective, using their own attribution windows and methodologies. Neither is definitively right. Both are perspectives on reality, not reality itself.

The practical response to this is not to abandon measurement but to be honest about its limits. Use a consistent methodology. Track the metrics that matter to the business, not just the ones that make campaigns look good. Supplement platform data with revenue data from your own systems. And when someone presents you with a dashboard full of green arrows, ask what the business actually sold.

There is also the question of incrementality. Did the ad cause the purchase, or would the person have bought anyway? This is genuinely hard to measure at scale, but it is the right question to ask, particularly for branded search campaigns and retargeting. Some of what looks like performance marketing is actually just intercepting people who were already going to convert.

Online Advertising Across Different Business Types

The channel mix that works for an e-commerce retailer is not the same as the one that works for a local service business, a B2B software company, or a professional services firm. Online advertising is not a one-size solution, and treating it as one is where a lot of wasted budget originates.

For local businesses, the geographic precision of paid search is often the most valuable feature. A beauty salon in Manchester does not need national reach. It needs to appear when people in Manchester search for the services it offers. Google Ads for beauty salons illustrates this well: the keyword strategy, the geographic targeting, and the conversion goal (a booking, a phone call, a direction request) are all tightly defined by the local context.

For e-commerce businesses, the combination of Google Shopping, paid search, and social retargeting tends to form the core stack. Google Shopping puts product listings directly in front of people searching with purchase intent. Paid search captures branded and category queries. Social retargeting brings back people who browsed without buying. Each layer serves a different part of the funnel.

For B2B businesses, the calculus is different again. Search volumes for specific B2B queries are often low, and LinkedIn’s targeting by job function and seniority commands a premium that is hard to justify unless the lifetime value of a customer is high enough. Content-driven advertising, where you promote a piece of genuinely useful content rather than a product page, often outperforms direct response in B2B because the buying cycle is longer and trust is built incrementally.

For subscription businesses, the unit economics matter more than almost anything else. The cost to acquire a customer has to be recoverable within a reasonable payback period relative to the average subscription length. I have seen subscription businesses running paid campaigns that looked profitable on a per-acquisition basis but were destroying value because churn was so high that the payback period was never reached. Online advertising cannot fix a product problem, and it cannot fix a retention problem. It can only bring people in.

The Innovation Trap in Online Advertising

Every year there is a new format, a new platform, or a new targeting capability that agencies present as the next frontier. Some of these are genuinely useful. Many are solutions in search of a problem.

I have sat in enough new business pitches and strategy presentations to recognise the pattern. The slide deck leads with innovation. There is always something new and shiny, a format that has just launched, a targeting capability that nobody else is using yet, a creative approach that is “significant.” The implicit message is that doing the new thing is the same as doing the right thing.

It is not. The question that should precede any channel or format decision is straightforward: what business problem does this solve? Not “is this interesting?” or “are our competitors doing it?” but “does this help us reach the right people with the right message at a cost that makes commercial sense?”

VR-driven advertising experiences. Shoppable video formats that require custom production budgets. Programmatic audio on platforms with limited measurement infrastructure. These things may have their place in specific contexts for specific businesses. But they are not inherently better than a well-structured Google Ads campaign with strong keyword coverage and a landing page that converts. The boring fundamentals done well will outperform the innovative experiment done poorly almost every time.

This does not mean ignoring new channels. TikTok was a new channel not long ago, and the businesses that tested it early when CPMs were low and competition was thin got a genuine advantage. The point is that novelty is not a strategy. Testing a new channel because it might work, with a defined hypothesis and a clear success metric, is different from adopting it because it looks impressive in a deck.

The history of online advertising is littered with formats that were going to change everything and then quietly disappeared. Online ad spending forecasts from 2010 were full of confidence about formats and channels that never delivered at scale. The platforms that survived and grew were the ones that solved a real problem for advertisers: reaching the right people efficiently and measuring the result with enough accuracy to make decisions.

Budget Allocation: The Decision That Matters Most

Most online advertising conversations focus on optimisation within a channel. How do you improve your Quality Score? How do you reduce your cost-per-click? How do you write better ad copy? These are legitimate questions, but they are second-order questions. The first-order question is where to put the budget in the first place.

A 20% improvement in conversion rate on a campaign that is targeting the wrong audience is still a campaign targeting the wrong audience. A 15% reduction in CPC on a channel where the customer lifetime value does not justify the acquisition cost is still an unprofitable channel. The allocation decision, across channels, across campaign types, and across funnel stages, has more leverage than almost any optimisation you can make within a channel.

The allocation should follow the evidence. Where are you seeing the best return on ad spend? Where is there untapped capacity, meaning more budget that could be deployed profitably? Where are you spending money because you always have, rather than because the data supports it? These are the questions that a serious quarterly review of online advertising spend should answer.

There is also the question of budget floors. Some channels require a minimum level of investment to generate enough data for the algorithm to optimise effectively. Running a Performance Max campaign with a daily budget of £10 is not a test. It is noise. If you cannot afford to fund a channel properly, it is worth considering whether to run it at all or to concentrate budget in the channels where you have enough scale to learn something useful.

Seasonality matters more than most budget plans account for. Online advertising costs are not fixed. CPCs and CPMs fluctuate based on competitive pressure, which itself fluctuates based on time of year, category, and market conditions. A budget that is distributed evenly across twelve months may be underfunding the periods when demand is highest and overfunding the periods when demand is lowest. Aligning budget to demand curves rather than calendar months is a straightforward improvement that most businesses do not make.

The broader paid advertising landscape, including how different channels complement each other and how to think about total investment across the mix, is covered in more depth at the Paid Advertising Master Hub.

What Has Changed and What Has Not

Online advertising has changed substantially since the early days of text ads and banner placements. The platforms are more sophisticated, the targeting is more granular, the automation is more capable, and the creative formats are more varied. The ecosystem that existed when Google Adwords changed its interface and auction mechanics in its early years looks almost unrecognisably simple compared to what exists today.

What has not changed is the underlying commercial logic. You are paying to reach people who might buy from you. The price you pay should be less than the value you get from the customers you acquire. The message you deliver should be relevant to the person receiving it. The measurement should be honest enough to make real decisions.

Privacy changes have made some of the targeting that advertisers relied on harder to execute. Third-party cookie deprecation, iOS privacy changes, and increasing regulatory scrutiny of data practices have all reduced the fidelity of audience targeting and attribution in ways that are still working through the industry. First-party data, the information you collect directly from your own customers and prospects, has become more valuable as a result.

AI and machine learning have accelerated the shift toward automated campaign management. This is genuinely useful when the automation is working toward the right goal with good data. It is genuinely problematic when it is optimising toward a proxy metric that does not reflect real business value. The advertiser’s job has shifted from manual bid management toward goal-setting, data quality, and strategic oversight. That is a better job in many ways, but it requires different skills than the ones that made people effective performance marketers ten years ago.

The volume of ad spend flowing through online channels has grown consistently over the long term. Double-digit growth forecasts for online ad spend from over a decade ago proved broadly correct, and the trajectory has continued. This growth has brought more competition, higher CPCs in many categories, and more sophisticated buyers on both sides of the market. The easy wins from the early days of paid search are largely gone. What remains requires more rigour, more patience, and more commercial honesty than the channel’s reputation for quick results sometimes suggests.

About the Author

Keith Lacy is a marketing strategist and former agency CEO with 20+ years of experience across agency leadership, performance marketing, and commercial strategy. He writes The Marketing Juice to cut through the noise and share what works.

Frequently Asked Questions