Probably, but nobody can tell you by how much - and anyone who gives you a precise number is quoting a study that measured something other than your situation.
That is an unsatisfying answer, so this article does two things instead. First, it looks honestly at what the commonly cited personalization statistics actually measured, because most of them have been quietly mangled in retelling. Second, it sets out how to measure the effect on your own list, which is the only number that can inform a real decision.
If you are trying to justify the spend to a client or a finance team, the second half is what you need.
What the commonly quoted statistics actually say
Search for personalization statistics and the same handful of figures appear everywhere, usually stripped of the context that makes them meaningful.
"344% higher engagement." This comes from OneSignal's 2024 State of Customer Engagement Report. The actual finding is that push notifications using personalized content have a 344% higher engagement rate than those that do not. It is about push, not email, and about personalized content in general - the report's own example is including a user's name or their favourite team's logo. It is genuine evidence that personalization works. It is not a measurement of personalized images in email, and it is not a number you should expect to reproduce.
"130% CTR increase." From the same report: Betmate saw click-through rise 130% after introducing a user_name tag for personalization. One company, one implementation, push notifications again. Useful as an illustration, useless as a forecast.
The older email statistics. Several widely quoted figures about personalized email transaction rates trace back to studies from the early 2010s. Inbox behaviour, filtering, and subscriber tolerance have all changed considerably since. Treat anything without a recent date and a linked methodology as folklore.
None of this means personalization does not work. The direction of the evidence is consistent and it points the same way. It means the magnitude is not transferable, and quoting a percentage as though it were a promise sets up a disappointment.
Why a reliable benchmark does not exist
The variance is not measurement noise. It is structural.
Use cases differ enormously. A name in a welcome banner and an abandoned-cart image showing the exact product someone left behind are both "personalized images" and they are not remotely the same intervention. One is a courtesy; the other is a reminder of something the person already wanted.
Audiences differ. A list of engaged customers who bought last month behaves nothing like a cold prospecting list. Personalization tends to amplify existing relationship strength rather than manufacture it.
Baselines differ. If your current click-through rate is 1%, there is room to move. If it is 8%, you are already doing most things right and the incremental gain will be smaller.
Industry norms differ. B2B SaaS, ecommerce, events, and recruitment have different inbox competition, different expectations, and different tolerances for personalization before it reads as intrusive.
Implementation quality differs. A well-designed image with the name integrated naturally performs differently from one with a name obviously pasted into a gap. The technique is not self-executing.
Any single benchmark number has averaged all of that away. That is why the honest answer to "how much will this lift my clicks" is "run the test - here is how."
How to actually measure it
The good news is this is a straightforward A/B test, and the mechanics are easier with personalized images than with most changes, because the only difference between your two variants is one image.
Set it up properly
Change one thing. Variant A uses your existing static image. Variant B uses the personalized version. Same subject line, same body copy, same send time, same call to action, same everything else. If you change the headline too, you will not know which change did the work.
Split randomly, not by segment. Your email platform's built-in A/B split does this. Do not send variant A to one list and variant B to another, and do not split by signup date or region - you will measure the difference between the groups rather than between the images.
Pick your metric before you start. Click-through rate on the image or its surrounding link is the direct measure. Open rate will not move, because the image renders after opening. Conversions further down the funnel are what actually matter but are noisier and take longer to reach significance.
Get the sample size right
This is where most in-house tests go wrong. Detecting a small difference requires more traffic than people expect.
As a rough guide, if your baseline click-through rate is around 3% and you want to reliably detect a lift of half a percentage point, you need on the order of ten thousand recipients per variant. Detecting a large effect - a doubling - takes far fewer, perhaps a thousand per variant. Small lists can only detect large effects, and that is fine as long as you know it going in.
If your list is small, run the same test across several sends and pool the results rather than declaring a winner after one campaign of 400 people.
Do not stop the test when it looks good. Checking daily and stopping the moment variant B is ahead is the most common way to manufacture a result that does not hold. Decide the sample size in advance and let it run.
Interpret it honestly
A result that is not statistically significant is a real result. It means the effect, if any, is smaller than your test could detect. That is useful information, not a failure.
Watch for novelty. The first personalized image an audience receives may outperform partly because it is unfamiliar. Retest after a few months to see what persists.
Segment the result afterwards. The average can hide the interesting part. Personalization often performs very differently for new subscribers than for long-standing customers, and knowing which is which tells you where to use it.
What to test first
If you are going to run one test, choose the flow where personalization has the most to say:
Abandoned cart is usually the strongest candidate. The image can show the actual product, which is genuinely informative rather than decorative.
Welcome emails work well because engagement is already high and the personalization signals attention at exactly the moment someone is deciding whether to pay attention to you.
Win-back campaigns are worth testing because the audience is disengaged by definition, so anything that interrupts the pattern has room to work.
Routine newsletters are the weakest starting point. Engagement is habitual and a name in the header rarely changes established behaviour.
How to have this conversation with a client
If you are an agency proposing this to a client, the borrowed-statistic approach is weaker than it looks. Clients have seen the same numbers, and a precise figure invites a precise expectation you cannot control.
A better framing:
Be straight about the evidence. The research consistently shows personalization improves engagement. The size of the effect varies widely by audience and use case, and anyone quoting an exact number for your list is guessing.
Propose a test rather than a rollout. One flow, one A/B split, an agreed metric and sample size, a defined date to review. This is a much easier approval than "let's personalize everything."
Agree in advance what success looks like. Decide together what lift would justify continuing before you see the numbers. Setting the bar afterwards is how tests become arguments.
Model the value, not just the percentage. A 0.4-point lift in click-through means nothing on its own. On a list of 50,000 with a known conversion rate and order value, it is a number the client can weigh against the cost. That is the calculation that gets budget approved.
This positions you as someone measuring rather than someone selling, which is a considerably better place to be when the results arrive.
Frequently asked questions
Do personalized images actually improve click-through rates? The evidence consistently points that way - OneSignal found push notifications with personalized content see 344% higher engagement than those without, and similar studies show the same direction for email. But the size of the effect varies enormously with audience, use case, and baseline, so the only number that can inform your decision is one from your own list.
How much lift should I expect? Nobody can tell you honestly. Published figures range from single-digit percentages to multiples, and they measured different channels, audiences, and implementations. Run a controlled test and use your own result.
Why do published personalization statistics vary so much? Because they measured different things. Some cover push notifications, some email, some the whole customer journey. Some are a decade old. Some come from a single company's implementation. Very few isolate personalized images specifically.
How do I A/B test a personalized image? Send two variants that differ only in the image - static versus personalized - split randomly using your email platform's built-in test, and measure click-through rate. Decide the sample size in advance and do not stop early.
How big does my list need to be? To detect a small lift reliably, on the order of ten thousand per variant. To detect a large one, roughly a thousand. Smaller lists can still test, but only for large effects, or by pooling results across several sends.
Does personalization affect open rates? Not from the image. The image renders after the email is opened, so it cannot influence the decision to open. It affects what happens next. Subject-line personalization is the lever for opens.
Is it worth it if my list is small? Possibly, but you will not be able to prove it statistically. On a small list the argument is qualitative - the experience is better and the effort shows - rather than measured. Be honest with yourself and your client about which case you are making.
How do I know the personalization is working technically? Before measuring performance, confirm the mechanics: send tests to profiles with complete data, missing data, and unusually long values, and check the rendering in Gmail, Outlook, and on mobile. A test that measures a broken implementation tells you nothing.
Test it on your own list
The most useful thing you can do with this article is stop looking for a benchmark and generate your own number. One template, one flow, one controlled test, and a fortnight later you have a figure that is actually about your audience.
OKZest generates personalized images from a single template - one URL that renders differently for every recipient - so setting up the personalized variant of an A/B test takes about ten minutes and no design work per recipient. It works with any email platform that accepts a custom image URL.
Create a free account, no credit card required.
Related reading: put every recipient's name inside your email image, personalized images in Klaviyo, personalized images explained.
Sources
- OneSignal, 2024 State of Customer Engagement Report, page 11 - the 344% notification engagement figure and the Betmate 130% CTR case study. Report PDF