What is eCPC?

A campaign bought on a thousand-impression basis spends A$6,000 for 800,000 impressions and receives 9,600 clicks. The effective cost per click is 6,000 ÷ 9,600 = A$0.625, even though no click price was ever agreed. Comparing that with a click-priced campaign paying A$0.90 makes the impression buy the cheaper source of traffic, which is a comparison the two buys' own pricing could not have produced.

Effective cost per click is used for two different things, and a conversation in which each party means a different one can run for some time before anybody notices.

The first meaning is arithmetic. Take what a campaign spent, divide by the clicks it produced, and you have what a click effectively cost, whatever pricing model the buy was struck on.

eCPC = total spend ÷ clicks

The second meaning was a feature. In Google Ads, Enhanced CPC took a manually set click bid and adjusted it in each auction according to how likely the platform thought that click was to convert — bidding above the manual figure where conversion looked probable, below it where it did not. Google has since retired it and migrated the campaigns that used it to manual cost-per-click bidding.

The two share nothing beyond the letters. This page covers both, because the ambiguity outlives the setting and still causes the most common confusion about the term.

Media is sold in incompatible units. One buy is priced per thousand impressions, another per click, another per completed view or per install. Clicks are a unit most of them produce regardless of how they were priced, which makes the derived click price a usable common axis for the traffic half of a mixed portfolio.

An impression-priced campaign never agreed a click price, and it still delivered clicks at some cost. Working that cost out lets it be compared with a campaign that did agree one. Frequently the impression buy is cheaper for traffic and the comparison is not available any other way.

The same calculation run on a conversion-optimised campaign is a useful drift detector. An automated strategy buys whatever produces conversions cheaply and is indifferent to the composition of the traffic. Over weeks, delivery migrates toward cheaper inventory: broader placements, audience networks, retargeting pools where the conversions were already likely. The campaign's headline cost per conversion holds steady, and the effective click price drops. That divergence is visible from outside the platform's own reporting and is worth a look whenever it appears.

Every cost-per-click figure inherits the question of which clicks were counted, and the derived version inherits it twice because it is frequently used to compare across platforms that count differently.

Platforms that report both an all-clicks figure and an outbound or link-clicks figure produce two effective click prices from the same spend, and the gap between them can be a third or more. A comparison that uses the wide definition on one side and the narrow one on the other is measuring a reporting setting.

The rule is simple and routinely ignored: pick the narrower definition, use it on both sides, and say so.

For comparing suppliers, the numerator should be what the buyer actually paid — media plus platform fees, ad serving, data costs and any managed-service margin. That is the number that reconciles with the invoice, and the difference between it and the media-only figure is precisely what a supplier comparison is trying to find.

For diagnosing delivery inside a single platform, media cost alone is cleaner, because the fees are constant and only add noise.

Either is defensible. Mixing them within one comparison is not, and it happens often enough that stating which version is in use should be routine.

Enhanced CPC sat between manual bidding and fully automated strategies. The advertiser set a click bid; the platform adjusted it up or down in each auction on its prediction of that click's chance of converting.

Its appeal was that it gave an account some of the benefit of automation without handing over the bid entirely. The advertiser kept a baseline, the model did the per-auction adjustment, and the result stayed intelligible in a way a fully automated target is not.

Its limitations followed from the same design. The prediction depended on conversion data, so an account with thin conversion volume gave the model little to work with and the adjustments were close to noise. Where a campaign had enough conversions for the model to be confident, a fully automated strategy optimising directly toward cost per acquisition or return on ad spend generally did better, because it was not anchored to a manual bid at all — which is most of the reason the feature no longer exists.

Google has retired it, and campaigns that ran on it were migrated to manual cost-per-click bidding. Accounts that want a bid adjusted by predicted value now choose a fully automated strategy and give it a target, and the sensible reading of the retirement is that the halfway position stopped earning its place.

For the derived figure: use it whenever campaigns bought on different models have to be compared on traffic cost, and as a drift detector on automated campaigns. Do not use it as a performance metric in its own right — cheap clicks that do nothing are the easiest thing in advertising to buy.

For the bidding feature: nothing, because it is no longer offered. An account that used to rely on it is now on manual cost-per-click bidding, and the choice in front of it is between staying there and moving to a strategy with a target derived from what a conversion is worth.

The derived click price belongs in the same panel as the effective impression price, and the two are joined by the click-through rate: the effective click price is the effective impression price divided by a thousand, divided by the click-through rate.

That identity is the diagnostic. When the derived click price moves, one of the other two moved, and knowing which one separates an auction story from a creative story in about a minute.

• Say which meaning you are using before the conversation starts. One is a derived cost figure and the other is a retired bidding setting, and the abbreviation is identical.
• Compute the derived version for every buy, however it was priced. It is the only way to compare the traffic cost of an impression buy against a click buy.
• Use the same click definition on both sides of any comparison. An all-clicks denominator against an outbound-clicks denominator produces a difference that is purely a reporting choice.
• Watch it fall while conversions stay flat. That pattern usually means delivery moved to cheaper, weaker inventory rather than that the buying improved.
• Include the fee stack when the purpose is comparing suppliers, and exclude it when the purpose is diagnosing delivery inside one platform.