Performance Marketing

Buying advertising once meant buying page space and airtime. Buy a full page in a newspaper and you could learn how many copies shipped, but not how many people looked at that page, and certainly not how many of them walked into a store. The money was paid up front, all at once, and after that it couldn't be stopped or moved.

Budget meetings kept jamming at the same point. The chief executive would ask what $45,000 of advertising bought this quarter, and the person responsible would answer that awareness went up. Neither side had a way to check whether that was true, so the meeting tilted toward whoever spoke with more conviction. The old joke about knowing half your advertising budget is wasted but not which half describes exactly that condition.

Online advertising broke the premise. How many times an ad appeared, who clicked it, what the clicker did on the site, how many of them signed up and how many paid — all of it started being recorded as one connected line. Because the record accumulated daily, there was no reason to wait for the next quarter either. If two campaigns show costs per signup of $9 and $30 on Monday morning, the budget can move that same afternoon.

The name comes from there. What separates this from other marketing isn't what gets bought but what gets measured and what changes as a result. A day in this job holds more time spent reading yesterday's numbers and redividing the budget than time spent making ads. The trade-off is plain. Put money in and people arrive that day; switch it off and the traffic stops with it. No asset accumulates the way search rankings or a body of published writing does, so a company running on this channel alone has to put in the same amount every month just to stand where it stood yesterday.

Running a single ad means settling five things: what shape it takes, what unit the money is paid on, who sees it, what the results get read as, and the work done daily while it runs. The reason for the five-way split is that it tells you where to reach when the results are bad. Impressions happen but nobody clicks: look at the format and the creative. Clicks happen but no signups: look at the targeting and the page people land on. Signups happen but no money is left: look at pricing and the metrics.

Ad format is the split by what shape the ad takes and where it appears. A few lines of text pinned above the search results, an image or video wedged between posts in a feed, a strip along the bottom of an app screen, a clip that runs before a video starts — each is a different format. What decides the choice is not what you want to show but what the person in that spot is doing right now. Someone who typed "meeting notes tool" into a search box is already looking, so a headline and a link are enough. Someone scrolling a feed wasn't looking for anything, so a picture that stops the thumb has to come first. Paste the same copy into both spots and one of them dies.

Ad pricing model is the rule for what gets counted when the bill arrives. There are three broad kinds: paying per thousand impressions, paying each time someone clicks, and paying only when an action like a signup or a purchase occurs. The difference between them is who carries the risk. Buy on impressions and the money goes out even if nobody clicks, so the advertiser carries it. Buy on actions and nothing is owed without a result, so the platform carries it. That's why platforms rarely open action-based pricing, and why the unit price is high when they do. The middle one, per click, is what most teams use. At $0.60 a click, bringing in 1,000 people costs $600, and if 30 of them sign up, each signup cost $20.

Ad targeting is the setting that picks who sees the ad. It divides into filtering on conditions the platform prepared in advance, such as location, age, and interests; uploading a list of existing customers and letting the platform find people who resemble them; and showing the ad again to someone who visited the site once and left. Narrower looks better, but it isn't. Shrink the audience and the number of advertisers competing for the same slot stays the same while the people to show it to go down, which pushes the unit price up. When Pagedeck narrowed from "people interested in document tools," about 300,000 of them, to "people doing planning work at companies of 20 to 100," about 40,000, the cost per click rose from $0.60 to $1.05. The cost per signup, however, fell from $20 to $8.50. Expensive clicks bought only people who would sign up. Targeting is settled at the end of that conversion, not at whether the clicks were cheap.

Ad metrics is the cluster of numbers that reads the result. They sit in order, so you can work down from the top to find where the block is.

1. Impressions — how many times the ad appeared. If even this number is missing, the budget or the bid is too low to win the auction.
2. Click-through rate — the share of impressions that turned into clicks. Plenty of impressions with a low rate means the copy and the image don't fit that spot.
3. Cost per click — what one click cost. It rises with competition for the slot.
4. Cost per acquisition — what one signup or purchase cost. When clicks keep coming and only this number spikes, the thing to examine is the landing page rather than the ad.
5. Return on ad spend — how much revenue one dollar of ad spend produced. It suits products where payment happens on the spot. For subscription products, where money arrives over months after signup, the right comparison is cost per acquisition against the total a customer leaves behind.

Operations is the work done daily while an ad runs, and three terms interlock there.

1. Bidding strategy — ad slots have no list price; they're sold at auction. Several advertisers bid for the same slot at once, and the winner is decided by the bid together with how the ad performs. You can set a ceiling yourself, or hand the platform a goal like "up to $15 per signup" and leave the bid to it. The second is easier, but the platform spends on the wrong places until it has enough data to learn from.
2. Creative testing — making several versions of copy and imagery and running them at once under the same conditions to compare. When Pagedeck ran "your team's documents in one place" against "the meeting notes are done before the meeting ends" on the same audience with the same budget, the click-through rates came out at 0.9 percent and 2.1 percent. A gap of more than double came out of one sentence, which is why the return here usually beats adjusting targeting or bids.
3. Ad fatigue — the drop in response as the same person sees the same ad repeatedly. The creative that was pulling 2.1 percent settles to 0.8 percent by its fifth impression on one person. The creative didn't get worse; everyone who was going to see it already has, so you widen the audience or swap the creative.

The five branches feed each other in order. Format fixes the slot and the pricing model follows; targeting fixes the pool and the unit price follows. Good creative with misplaced targeting means showing a well-received ad to people who won't buy, and precise targeting with flat creative means sitting in front of the right people while nobody clicks.

Wrapping the whole job from outside the five is advertising regulation. These are the rules against exaggerated or deceptive advertising, so certain language is off limits — claims that state an outcome as certain, discount figures inflated beyond what's real. Push the copy harder because the numbers look good, get caught, and a fine erases the quarter's results outright.

"I'm killing this campaign if cost per signup goes past $25. Our payback line is $19."

If bringing one person in costs more than that person will ever pay, spending more loses more. This is where the off switch gets set in advance.

"Click-through is fine but signups aren't landing. I don't think it's the creative."

The ad itself is delivering people, so what to examine is the landing page, not the ad.

"Frequency went over 5 this week. We need new creative."

The same people have seen the same ad five times, so swap it before the response falls.

At Pagedeck this work belongs to Ravi, the growth marketer. In the quarter when spend climbed toward $15,000 a month, Ravi listed cost per signup by campaign every Monday and moved budget on the spot. He started with half in search and half in feed, and four weeks later the two had separated to $9 and $30. Moving the feed budget into search then pushed the search unit price up, because the number of people searching "meeting notes tool" is finite and anything past that flowed into unrelated queries. That's when Ravi learned search could absorb about $2,300 a week and no more.

At the end of the quarter Grace asked what 180 signups a week had cost, and Ravi brought cost per acquisition by campaign alongside how many of those people were still around three months later. What changed the judgment was what Sofia added from sales: customers who arrived through ads needed the product explained from scratch at the start of a call, which made conversations slower. On the numbers alone search was the cheapest channel, but putting the quality of the people it brought into the same view changed the order.

The most common mistake is believing that because everything is recorded, the right answer gets calculated. The screen fills with metrics carried out to decimal places, which makes that an easy read. But those numbers only show the paths that stayed connected to a record. Someone who saw an ad, didn't click, and searched the company name directly three days later never appears in ad performance, and someone who crossed several channels hands all the credit to whichever one was clicked last. Leave this misreading in place and the budget concentrates on whatever is measurable. The activity that was actually generating demand gets cut first for not producing numbers, and only six months later, when ad efficiency falls too, does it emerge that the cut activity was holding up the front of the funnel.

Reading cost per signup as better when lower goes wrong just as often. Bringing in more people for the same money looks like the better job, so the reading is natural. But the unit price almost always drops when the audience widens, because widening it means cheaply acquiring people whose interest is thin, and what happens after they arrive is different. At Pagedeck, people who came in through a $6 campaign were 6 percent retained after three months; people from the $20 campaign were 41 percent. Move budget on unit price alone and you get a quarter where signups rise and revenue doesn't. Because the report reads as a metric improvement, this state usually goes unnoticed for a few quarters, until someone counts paid conversions separately.