HomeBlogPaid media costs

How many new ad creatives a store needs per month: the refresh pace of real online stores

Every month the same question sits on the design schedule: how many new ads to make. We opened 55 store-months of Israeli online stores, counted what went up each month, how much of it caught budget, and what that did to purchases per shekel.

A loose fan of thin cards of frosted pale green glass lying on a bright studio surface, with one card of polished green marble standing upright on its edge in front of them, in diagonal window light
Twelve ads go up. One or two will stay.
15
new image ads an online store launched in a month. Median of 55 store-months
92%
purchases per shekel in the months a store launched many new ads, against the months it launched few. Within the same store
16%
of new ads reached 1,000 shekels of spend, that is, became a real ad. One in 6

What refresh pace is, and what we measured

Creative refresh pace is how many new ads a store launches in a month relative to its budget: how many new images enter the ad account, what share of the month's spend goes to them, and how many of them get any budget at all. The usual advice is "keep uploading": more ads, more variations, every week. We wanted to know whether the stores that refresh faster really get more for every shekel.

I plan the monthly design schedule for online stores, and at the start of every month the question is how many ads to make: ten, twenty, forty. Usually the answer is set by how many hours there are, not by what the store needs. In the article on ad fatigue we checked how long one image keeps selling. Here the question is at the level of the store: not when to replace an image, but how many new ones a month needs.

We took every image ad on Meta in the Israeli online stores we work with, from May to October 2026, month by month. A new ad is an ad whose first day with spend falls in that month. The sample: 12 stores, 55 store-months, 1,281 new image ads and 970 thousand shekels of cumulative spend on image ads, which brought 12,945 purchases, from May to October 2026. A store's month is in only if at least 3,000 shekels were spent on image ads that month. Video ads are out, as in the previous article: this is about images. Purchases are what Meta reports on the ad, not the store's revenue. And every comparison of a fast month with a slow one is made inside the same store, against that store's own median, so a large store is never compared with a small one.

How many a store launches in a month

In a median month a store launched 15 new image ads, 13.5 new ads for every 10,000 shekels it spent, and 40% of the month's spend went to ads that started that month. In other words: in an ordinary month, about two fifths of the money ran on images that did not exist a month earlier. In all, 38 image ads ran in a median month, so the new ones are roughly four of every ten active ads.

MetricLower quarterMedianUpper quarter
New image ads in the monthan ad whose first day with spend falls in the month101526
New ads per 10,000 shekelsthe refresh pace relative to budget8.113.519.7
Share of spend on the new adshow much of the month's money went to ads launched in it25%40%57%
New ads that caught budgetspent 300 shekels in their first 14 days258
Active image ads in the monthevery ad that spent money in the month243865
What a month of image ads looks like. 55 store-months, May to October 2026. Preneuriat data, 12 stores, 55 store-months, May to October 2026

The range is wide, even inside the same store. Between a store's fastest and slowest month, the refresh pace relative to budget moved by a median of 3.1x. No store works at a steady pace: there is a month with a big sale and twenty new images, and after it a month in which the veterans run. That is exactly what can be measured: the same store, a fast month against a slow one.

More new ads did not bring more purchases per shekel

In every store we took the months in which the refresh pace was above the store's median and the months in which it was below, and compared purchases per shekel. In its fast months a store got 92% of the purchases per shekel of its slow months. In 4 of the 9 stores the fast months were better, in the rest they were not. The picture is the same when the pace is measured differently: by the share of spend that went to new ads, 91%, by the number of new ads, 91%, and by how many of the new ads caught budget, 92%.

Purchases per shekel, by the month's refresh pace

Every store-month is measured against its own store's median, in pace and in result. 100% is the store's median month. In the fast quarter the pace was 2.7x the slow quarter.
The store's median month
0% 50% 100% 104% Slow quarter 100% Below median 92% Above median 103% Fast quarter
0% 50% 100% 104% Slow quarter 100% Below median 92% Above median 103% Fast quarter
Preneuriat data, 55 store-months in 12 stores, May to October 2026

The chart is flat. In the quarter of the slowest months, purchases per shekel were 104% of the store's median month, and in the fastest quarter, where the pace was 2.7x, 103%. In between, 100% and 92%. A store that doubled its refresh pace in a given month did not get more purchases per shekel for it, and not fewer either.

The months with many new ads looked exactly like the months with few. The money worked the same.

The reason is in the next comparison. Inside the same month, the month's new ads brought 102% of the purchases per shekel of the veteran ads running beside them, across 47 store-months. The new ads beat the veterans in 53% of the months and lost in the rest. 40% of the spend went to the new ads, and 40% of the purchases came from them. A new ad, as a group, is not a better ad. It is an ad nobody yet knows is good.

How many of the new ads catch budget at all

This is the number that changes the plan. Of 1,237 new image ads that had a full 14 days in the data, 65% spent 100 shekels or more in their first two weeks, 32% spent 300 or more, and only 17% reached 500. Meta decides within days who gets budget, and most new images do not get it.

How many of the new ads reached each spend level

Image ads launched from May to September 2026, so each had at least a month. Share of the ads that spent at least that amount over their life.
0% 50% 75% 100 ILS 42% 300 ILS 29% 500 ILS 16% 1,000 ILS 7% 2,000 ILS
0% 50% 75% 100 ILS 42% 300 ILS 29% 500 ILS 16% 1,000 ILS 7% 2,000 ILS
Preneuriat data, 1,055 new image ads in 12 stores, May to September 2026

Over its whole life, a new image ad reached 1,000 shekels of spend, the bar from which we measured an ad in the previous article, in 16% of cases. One in 6. 7% reached 2,000 shekels, one in 14. Between stores the range is wide: in the median store 14% of new ads became a real ad, and between 6% and 39% at the extremes. Inside the month, the three strongest new ads took a median 60% of all the money that went to new ads.

Two things we checked that did not happen. First: uploading more does not dilute. In the months a store launched more ads than usual, 31% of the new ads caught budget, and in the months it launched fewer, 33%. The same rate. Each ad is judged on its own, and launching ten instead of five simply gives twice the hits. Second: an ad that caught budget is not necessarily an ad that wins. Of 381 new ads that caught budget, 46% brought purchases per shekel in their first month at or above the store's veterans. The rest got money and were worse than what was already running.

Catalog against designed image

A store's ad account has two kinds of image ads. A catalog ad pulls the product images from the feed, and Meta picks what to show each person. A designed image is design work: product, background, message. A new catalog ad becomes a real ad more often:

New ads that reached 1,000 shekels of spend, by kind

Image ads launched from May to September 2026. 293 catalog ads and 762 designed images.
Preneuriat data, 1,055 new image ads in 12 stores, May to September 2026

That makes sense: a catalog ad is really dozens of images, and Meta finds what works inside it. A designed image is a single bet. So when you look at designed images only, the planning numbers look like this: 14 new designed images in a median month, 18.1 per 10,000 shekels, 46% of designed-image spend went to the new ones, 4 of them caught budget, and 14% became a real ad. One in seven.

What to do with this on the monthly schedule

This is how we now plan a store's design month.

  1. Plan by how many have to work, not by how many go up. One in 3 new images catches budget, and one in 6 becomes a real ad. A store that needs 5 working ads a month, which is the median, needs to launch about 15. That is the number that goes on the schedule, and hours follow from it.
  2. Do not double the pace to "refresh". The months in which a store launched 2.7x its usual number brought 103% of the purchases per shekel of an ordinary month. More ads buy more bets, not more efficiency. While the veterans still sell, they stay, and the new ones go up beside them, not instead of them.
  3. Launch in small groups, and check after two weeks. The ad that did not reach 300 shekels in 14 days probably will not. Close it and make the next one. What happens in those two weeks is in the article on Meta's learning phase.
  4. Measure the hit against the veterans, not against itself. An ad that caught budget beat the store's veterans in only 46% of cases. The comparison is the new ad's purchases per shekel in its first month against the ads that were already running. If it does not pass them, the budget it took is budget we pulled from something better.
  5. Catalog always runs, the designed image is the bet. A new catalog ad becomes a real ad in 25% of cases, a designed image in 13%. The catalog gives the base, and design tries to beat it with a single image. Every month we also make a few variations of the image that won the previous month, the same product on another background or in another crop: those are the cheapest bets there are.
  6. When the hit starts to drop, the replacement already has to be up. How long one image keeps selling and when to replace it is in the article on ad fatigue. The two articles together are the schedule: how many to launch in a month, and when to replace what is already running.

Questions we get

How many new ads should you launch per month?

In the stores we measured, an online store launched a median of 15 new image ads a month, 13.5 per 10,000 shekels of spend. But only 32% of them caught budget in their first two weeks, and 16% reached 1,000 shekels. So plan the other way round: decide how many working ads the month needs, a median of 5, and launch about three times that.

Does launching more new ads improve results?

Not in our data. In the months a store launched new ads at a pace above its own median, it got 92% of the purchases per shekel of its slow months, and split into quarters every quarter came out between 92% and 104%. The month's new ads brought 102% of the purchases per shekel of the veterans running beside them. More ads are more bets, not more efficiency.

How many new Facebook ads actually get budget?

Of 1,237 new image ads, 65% spent 100 shekels or more in their first two weeks, 32% spent 300 or more, and 17% reached 500. Over their whole life, 16% reached 1,000 shekels. The rate did not fall in the months a store launched more: 31% against 33%.

What is better to launch, a catalog ad or a designed image?

Both, in different roles. A new catalog ad reached 1,000 shekels in 25% of cases, a designed image in 13%, because a catalog is dozens of images and Meta picks among them. The catalog gives the base, and the designed image tries to beat it. Variations of the image that won last month are the cheapest bet.

How do you know whether a new ad succeeded?

In two steps. After 14 days: did it spend 300 shekels or more. If not, Meta did not pick it and probably will not. After the first month: its purchases per shekel against the store's veteran ads in the same month. An ad that caught budget beat the veterans in only 46% of cases, so budget it took without winning is budget that came off a better ad.

Fit check

Want to see your real ROAS?

Every Preneuriat client gets a live dashboard: store revenue, all media spend, real ROAS, updated all day. The first call goes both ways.

Check the fit