01 / The questionDo people click polish, or do they click a person?
We sell two digital products of our own through paid ads: a property investing ebook for Australians and a doctor's appointment kit for women in the United States. Before spending properly on either, we wanted to answer one question with real money rather than opinion: does a polished, designed ad card get more clicks, or a plain ad that looks like something a real person made?
Most small business ads look like ads: a logo, a stock photo, a headline in a box. People have learned to scroll straight past them.
We started the polished card at A$2.20 a click and brought it down to A$0.24 by changing the ad, not the budget. Paid campaigns like these are run and reported separately: the traffic across our own sites, the 21,253 unique visitors a month, is mainly organic search and content, not paid ads.
02 / Test oneAustralia, property ebook, one of the major ad platforms, seven days
The plain version was an unbranded, personal looking artefact. The designed version was a proper ad card. Same product, same platform, same week.
| Ad | Impressions | Clicks | Click rate | Cost per click |
|---|---|---|---|---|
| Plain personal artefact | 41,764 | 983 | 2.354% | A$0.24 |
| Designed ad card | 636 | 1 | 0.157% | A$2.20 |
The plain ad drew 983 clicks in seven days. Its click rate was fifteen times the designed card's, and each click cost about a ninth as much. The designed card was shown far less, so its figure rests on a small sample, which is exactly why we ran the test again somewhere else.
03 / Test twoUnited States, appointment kit, the same platform
A different product, a different country and a different audience. This time both versions were shown almost exactly as often, which makes it the cleaner comparison.
| Ad | Impressions | Clicks | Click rate | Cost per click |
|---|---|---|---|---|
| Handwritten artefact | 1,191 | 42 | 3.526% | A$0.56 |
| Designed ad card | 1,166 | 10 | 0.858% | A$1.85 |
Same result, about four times the click rate at under a third of the cost. Across the whole campaign it drew 191 clicks from 8,729 impressions, a 2.19% click rate. The finding held in two unrelated markets.
04 / What we threw outBad data is worse than no data
Not every run counted. The first night of the property campaign ran 12,273 impressions at a 0.114% click rate, and we marked it invalid because the targeting had not actually limited who saw the ad. An early card test drew zero clicks on 801 and 767 impressions. Both were logged and excluded rather than averaged in, because averaging a broken test into a good one gives you a number that describes nothing.
05 / Where the visitors wentBuilt for the phone, because that is where they were
The property ebook's landing page recorded about 1,140 real visits in its first two months. 69% of them were on a phone and 78% were in Australia, so the page was built phone first rather than shrunk down from a desktop design.
06 / The same test for your businessWhat we would do for you
- Run your polished ad against a plain one that looks like a person made it: a photo of your own work, a handwritten note, a real job.
- Show both evenly, on a small capped budget, before scaling anything.
- Throw out broken tests instead of averaging them in.
- Judge every ad on the enquiries and sales it brings in, then scale the winner hard.
