Your ad was seen 40,000 times. Was it?

An ad impression is the most quoted number in advertising and the least examined. It sounds like a person seeing something. It is closer to a delivery receipt: a record that a server sent a creative towards a browser, under conditions that vary enormously between platforms and are rarely stated on the invoice. At forty thousand of them the difference is easy to ignore. At four hundred, which is what a small budget actually buys, it decides whether your test meant anything. (Disclosure: published by BidSurvivor, an ad board that counts things differently — the last section says how, including where our counting is weaker.)

What the word covers

The industry definition of an impression has been through several revisions, and the modern one is narrower than most people assume. The widely used viewability standard counts a display ad as viewable when at least half its pixels are in the browser viewport for at least one continuous second. Video asks for two seconds. That is the bar for viewable — and a plain impression, the kind most reporting defaults to, does not require even that.

Read that again with a buyer's eye. Half the ad, one second, in a viewport that may be behind another tab, below a scroll the reader never made, or on a page abandoned mid-load. Nothing in the definition requires the page to have finished rendering, the reader to have looked at that part of the screen, or a human to have been present at all.

The practical consequence: "impressions" and "times a person saw your ad" are different quantities, and the gap between them is not a rounding error. Typical display viewability rates sit well below 100%, which means a meaningful share of what you paid for was never eligible to be seen under the industry's own definition.

Why this bites harder at small budgets

A large advertiser absorbs this. They buy across many placements, average the waste, negotiate on viewable impressions and audit the difference. None of that is available to someone spending fifty dollars.

Your sample is small enough that composition matters. With four hundred impressions, whether three hundred of them were below the fold changes your click rate by a factor you cannot see and cannot correct for. The benchmarks you compare against were computed on inventory with a different viewability profile, so the comparison is not measuring what you think.

You cannot buy your way out with an audit. Verification vendors exist and they are priced for advertisers whose monthly spend has more digits than your annual one. The honest small-budget move is to stop trusting the impression figure rather than to try to clean it.

The metric you are paying by may be the one most affected. If you are buying on a CPM basis, you are paying per impression, which means you are paying for the ambiguity directly. Buying per click moves the risk onto the seller — which is exactly why sellers with weak inventory prefer to sell impressions.

Three checks worth making before you buy

Ask what counts as an impression here, in one sentence. A seller who can answer plainly — "the card is rendered in the viewport" or "the ad request was served" — is telling you something useful either way. A seller who cannot, or who redirects to audience size, has answered.

Ask for the click number beside it. Clicks are hard to inflate accidentally: someone either arrived at your site or did not, and your own analytics will confirm or contradict the seller's figure within a day. An impression count with no click count beside it is unfalsifiable, and platforms that publish their numbers are distinguished mainly by their willingness to be checked.

Compare the ratio, not the totals. Forty thousand impressions and eight clicks is a worse placement than four hundred impressions and eight clicks, and the second one costs less. The ratio between what was delivered and what was acted on is the only part of the impression figure that survives contact with your own logs.

What to use instead

For a small advertiser, the impression count is best treated as a denominator you did not choose and cannot verify. The numbers that survive scrutiny are further down the funnel.

Visits, from your own analytics. They are yours, they are countable, and no seller controls them. If a placement claims ten thousand impressions and your analytics shows six visitors, you have learned everything you need about that placement without ever resolving what its impressions meant.

Cost per visit, computed by you. Divide what you paid by the visits you observed. This single number makes wildly different pricing models comparable and is immune to definitional games, because both of its terms are things you can see.

Value per visit, from your own funnel. What a click is worth to you is arithmetic over your conversion rate and margin, and it sets the ceiling above which no impression count, however impressive, makes a placement worth buying.

Together those three replace the impression entirely for decision-making purposes. The impression figure remains useful for one thing: comparing two placements on the same platform under the same definition, where the ambiguity is at least constant.

What we count, and where we are weaker

Our board counts two things per slot and publishes both. A card open is the brand's card rendered on the board; a click-through is a visitor leaving for the brand's own site. They are counted separately and shown separately, on the front page and on each brand's own page, before anyone bids.

The honest limitation: an open on our board is subject to the same physics as anyone else's impression. A card rendered in a viewport is not proof that a human looked at it, and we do not claim otherwise. What is different is that the click number sits directly beside it and is published whether the ratio flatters the placement or not — so a buyer can compute the open-to-click rate themselves rather than take a delivery receipt as evidence of attention.

Slots are two a day, twelve hours each. The first brand into an empty slot pays nothing; a held slot can be taken from a $1.00 floor, capped at $5,000; every visitor starts with $100 of house credit. If the ratio on a slot is bad, it is on the page. That is the point of publishing it.