Bidding Strategy

Ad slots don't have list prices. A line at the top of a search results page, a card wedged between two posts in a feed — none of it sits on a shelf with a number attached. The slot is sold at the instant someone opens the screen, so an advertiser never looks up a price and pays it. You name a figure and find out afterward whether it was enough.

For a while after this became the norm, marketers adjusted those figures by hand every day. You came in, looked at yesterday's positions, nudged the keywords that had slipped up by eight cents, and pulled the ones burning budget without a single signup down by fifteen. Miss a Saturday and a Sunday and Monday morning greeted you with a spent budget and three signups to show for it.

The manual labor wasn't the worst of it. Winning the top slot meant nothing if the people who arrived through it never signed up. Buying clicks at thirty cents meant nothing either if the bid was so low the ad barely appeared. Cheap and profitable turned out to be two different things, and the gap showed up in the numbers every day.

A bidding strategy moves that judgment call forward in time. Write down the ceiling and the conditions once, and the same reasoning applies at three in the morning when nobody is watching the account.

The moment a slot opens, the platform gathers every advertiser who wants it, runs an instant auction, and weighs each bid against how likely that ad is to get a response. Bidding the most does not automatically win, which is what separates this from a normal auction. A platform that shows an ad nobody clicks earns less, so an ad bidding $0.75 with a 3 percent click-through rate (CTR) beats one bidding $1.15 at 0.5 percent. The winner also pays less than the figure it named, just enough to clear the runner-up, so a generous ceiling doesn't drain at that rate.

What the money is charged against is decided separately, by the ad pricing model. The bidding strategy sets the amount attached to whatever unit that model picked. Four approaches cover most accounts.

1. Manual cap — you name the highest cost per click (CPC) you'll accept. Set it at $0.75 and nothing above that gets bid. Costs stay predictable, but somebody has to look at the account daily.
2. Target CPA — you give the platform a target cost per conversion and let it choose the bids. Set $15 per signup and the platform estimates each person's odds of signing up and bids accordingly.
3. Target ROAS — the bid is set by how much revenue each dollar of ad spend returns. This fits products where the purchase happens on the spot, right off the ad.
4. Spend-the-budget — no ceiling at all, just instructions to use the full daily budget and bring back as many clicks or conversions as possible. Useful for events with a fixed end date.

The two approaches that hand control to the platform run on data. Until roughly fifty conversions a week have accumulated, the platform is guessing at who will sign up and bidding as though the guesses were good. When Pagedeck set a $15 target CPA, the first ten days cost $33 per signup, and only after three weeks did the figure settle at $14.

Which strategy you pick depends on where the campaign sits in its life, and the test is a single question: has it collected enough conversions to hand over?

1. A campaign you just turned on — with no history, start with a manual cap. Work the ceiling backward from the conversion rate of an earlier campaign. If a signup is worth up to $15 and three of every hundred visitors sign up, $0.45 per click is the break-even line.
2. A campaign with history — once it clears fifty conversions a week, move it to target CPA. Setting the target far below what the campaign currently delivers makes the platform cut impressions until the campaign quietly stops, so lower it 10 to 20 percent at a time.
3. A campaign with a narrow audience — where ad targeting has already gathered people who are close to buying, bidding high is fine. Someone who reached the pricing page and left converts at five or six times the rate of a stranger, so even at $1.50 a click the cost per signup lands lower.
4. A campaign with an end date — a three-day promotion whose budget disappears if unspent goes on the spend-the-budget setting.

Before touching a bid, check whether bidding is the problem at all. Thin impressions do point at a bid that's too low. But impressions arriving with no clicks is a job for creative testing, where the copy and the image get swapped, and an ad whose response fades week over week is ad fatigue — raising the bid won't bring it back.

"The target's set at $15 and it's been running at $27 for three weeks. Should we switch to manual?"

The learning period is over and the target still isn't being hit, so a person should take the bids back.

"Raising the bid on this keyword won't do anything. We're already at 90 percent impression share."

The campaign is buying nearly every slot available, so there's nothing left to outbid anyone for.

"It's still learning, so let's leave it alone this week and watch."

The platform is collecting data, and changing the settings resets that collection.

Ravi brings this up at Pagedeck's Monday performance review. Cost per signup on the free trial campaign came in at $18 last week, and Grace asks why that week was expensive. The answer is that a competitor entered the same keywords on Wednesday, cost per click jumped from $0.70 to $1.15, and the target-based campaign followed the market up to hold its position. Two options — hold the target and keep buying expensive slots, or lower it and buy only the slots that still pay for themselves. Ravi chose the second. Signups that week dropped by half, and cost per signup came down to $14.

The most common mistake is assuming a lower bid is always the cheaper deal. Bidding is spending, so pulling the number down feels like saving. What a low bid actually buys is the slots nobody else wanted, and those slots are cheap for a reason. When Pagedeck dropped its cap to $0.45, cost per click fell to around $0.30 while cost per signup climbed from $15 to $29. Squeezing the budget by watching the click price alone makes the number that matters worse.

The other one is reading automated bidding as permission to stop looking. The platform optimizes toward the target a person typed in, not the economics of the business. Tell it $38 per signup is acceptable and it will faithfully spend all $38. Leave a wrong target in place for a few weeks and the campaign looks healthy in every chart while the invoice says otherwise.