Incremental lift

증분 효과

Measure the extra outcome beyond what would happen without exposure.

···
html
<div class="lift"><div class="head">EXPERIMENT <b id="gap">+0.0 pt LIFT</b></div><div class="plot"><div class="line treat"><span>EXPOSED</span><i id="t"></i></div><div class="line ctrl"><span>CONTROL</span><i id="c"></i></div><div class="brace" id="brace"></div></div><div class="base"><span>START</span><span>AFTER CAMPAIGN</span></div></div>
css
.lift{width:min(86%,600px);font:850 clamp(10px,max(2vmin,1vw),14px)/1.2 sans-serif}.head,.base{display:flex;justify-content:space-between;align-items:center}.head{border-bottom:2px solid var(--fg);padding-bottom:8px}.head b{color:var(--accent)}.plot{position:relative;height:clamp(85px,23vmin,170px);margin-top:clamp(8px,2vmin,17px);border-left:1px solid var(--line);border-bottom:1px solid var(--line)}.line{position:absolute;left:8%;right:7%;height:2px;transform-origin:left center;transition:transform .7s}.line span{position:absolute;right:0;top:-19px;color:var(--fg);font-size:clamp(10px,max(1.5vmin,1vw),12px)}.ctrl span{top:7px}.line i{display:block;width:100%;height:2px;background:currentColor}.treat{top:65%;color:var(--accent)}.ctrl{top:65%;color:var(--accent-2)}.brace{position:absolute;right:7%;top:25%;height:40%;border-right:2px solid var(--fg);transition:height .7s,top .7s}.base{padding-top:6px;color:var(--muted);font-size:clamp(10px,max(1.5vmin,1vw),12px)}
js
let opened=false;function draw(){const t=document.querySelector('#t'),c=document.querySelector('#c'),b=document.querySelector('#brace');t.parentElement.style.transform='rotate('+(opened?-19:0)+'deg)';c.parentElement.style.transform='rotate('+(opened?-7:0)+'deg)';b.style.top=opened?'29%':'63%';b.style.height=opened?'24%':'2%';document.querySelector('#gap').textContent=opened?'+2.0 pt LIFT':'+0.0 pt LIFT'}draw();setInterval(()=>{opened=!opened;draw()},1700);

Incremental lift asks what changed beyond outcomes that would have happened without advertising. Randomized exposed and control groups provide a basis for estimating that counterfactual.

In the demo, purchase rates start together and diverge; only the gap is highlighted. The example values illustrate the concept, not a campaign result.

Selection bias, outside exposure, and small samples can distort estimates. Inspect experiment design and uncertainty before acting.

When to use

Use it to test whether media caused additional purchases.

Open as page ↗