eBay’s mission is to be the world’s favourite destination for discovering great value and unique selection
Do Promoted Listings on eBay Work?
– The Trial
David Brackin is the managing director of Stuff U Sell, the leading eBay trading assistant in the UK and a regular Tamebay contributor. Today, in the first of a three part series and the first published rigorous wide scale scientific test of eBay Promoted Listings, David sets out the basis for the experiments that Stuff U Sell carried out. Come back tomorrow when we will reveal the results of the tests and on Friday for the conclusions and actionable insights that you can use in your business.
Do Promoted Listings on eBay work? – The Trial
The “Promoted Listings” feature is now well-established on eBay as an additional paid-for service to market your listings. For anyone who hasn’t yet tried the service, it is relatively simple. You set up a campaign, identify the listing numbers you’d like to include and choose an additional percentage of the sale price that you are prepared to pay on top of the final value fee. In return, your listings compete to be shown in premium positions across the site – for example currently slots 1, 4 and 5 in the search results page are all promoted listings. The higher the percentage you are prepared to pay, the more likely you will appear in the top spot.
The benefits are clear. You only pay when you actually sell an item, and each sale then increases the sale history of the listing, so you can use it to give a kick-start to your best match position. Initially launched as a trial in 2015 and then expanded across further categories in 2016, I remember speaking with the product manager and being impressed by the idea but wondering one thing: will it work?
The inventory we have at Stuff U Sell tends to be one-off second-hand goods that we’re selling for other people, so the benefits of sales history don’t really apply to our listings, and in any case we relist our listings weekly to surf the Sunday wave of buyers. If it works for our listings then it will be a no-brainer for anyone else. We had run a few small trials and we noticed that we would always get some sales and pay some fees. However, sales move up and down each week at random so it might just be a rainy weekend and not the promoted listings that gave us the boost. How could we tell if we were going to get those sales anyway?
We were about to give up on Promoted Listings altogether when I spotted a book at the house of a friend who is a healthcare expert.
I learned how the gold standard of clinical research is a blind randomised test against a control group, and it was obvious this is what we needed to do with Promoted Listings. However, it wouldn’t be cheap to run the kind of extensive trial that would produce a definitive answer, so I asked eBay if they would cover the costs, provided we had independence to design it properly and share the results. They agreed and over the next week I’m going to share our results and what they mean for your business.
In clinical terms, we decided to test whether a “dose” of Promoted Listings spend would increase sell-through on a randomly selected group of our listings.
The trial would start with 13 weeks of no promoted listings (“zero dose”) to find our average sell-through rate how much it moves up and down randomly. This would enable us to use statistics to determine whether any effect was significant or just random noise.
Then over the next nine weeks, we would randomly select either all the even-numbered listings, or the odd-numbered listings or both and apply a “dose” of Promoted Listings to them and leave the other control group “untreated”. We chose PL dosages from 0% to 20%. Because we relist our listings every Sunday, the even/odd listings are jumbled up again each week, so each is a randomised half of our inventory unrelated to the previous week. We only applied the trial to our Buy-It-Now/Best Offer listings, not the auction format.
Furthermore, to make the test “blind”, my business partner was to pick which trial we were doing each week and not tell me, and I would analyse the relative sell-through rate.
The results were eye-opening. Come back tomorrow to see our initial findings.
Looking forward to seeing these results. For me personally like you I sell lots of one off items for clients. And have a few businesses where I sell more items and have multiple stock.
When I run promoted listings it increases sales not only on the items I promote but I see a marked increase in sales other listings I run.
Hi cannot really see how any trial can prove conclusive where the majority of goods are second hand one offs. Surely you need the same goods to get a true reflection.
Second hand shoes against 2nd hand car has no comparison.
Secondly the strange thing is when we signed up to eBay we were explained to by eBay that they are a selling platform offering to display your goods in a shop window for buyers to see.
Now they are saying unless you pay us more money your products are still in the shop window but we have boarded your window up.
With sales on eBay plummeting week on week, month on month I would have thought maybe removing these extra ways to grab money and keep their books balanced may be the best way to keep the sellers that are deserting the site to other platforms.
Instead of making it like a game of 3 card brag.
The biggest wallet always wins. (Tesco Argos etc)
As they go blind and raise the stakes so high so you have to stack.
Just leave the choice of who to buy from up to the buyer and stop blinding them with the highest payers often not the best option as between just those two they have 6,000 negative and neutral feedbacks in one year now that according to so called eBay research means those 6,000 buyers do not shop on eBay again. So 500 per month go to another platform and their friends may follow. I know that is a small percentage of sales to those companies but in 10 years that’s 60,000 who will be shopping elsewhere plus that’s just 2 companies there are many more i have not mentioned
Yes of course your sell through rate will increase if you PAY for more coverage. People in this day and age will only ever look at the first few options.
It is very much PAY to play. Anything you sell now on ebay is with stealth fees that is clear. The only saving grace is your not paying for clicks but for transactions.
eBay now is maybe for the 1 of items and the large companies, where you have lot of margin in say some book.
However say in our industry it is not so, we have used the promoted listings since the trial back when they started when it showed up in the sidebar back in 2016.
The real problem is listings like this eBay item number: 223258933099 , which are counterfeit and pretty much dominate every single slot…Why we have had it with them.
The issue is not Promoted listings or (stealth fees), it is the fact they are abused and ebay have let is go unchecked as the are so narrow focused on this one stream of revenue….
If everyone on tamebay report that item one after the other hopefully it will have an effect
I’m a sceptic on promoted listings. Like David, our items are one-offs in collectables, but theoretically higher viz in those categories could be useful. But I can’t identify any consistent benefit so far. Maybe it’s the fear of being driven down the searches if you don’t promote that’s driving this on Ebay.
Interested to see what the results are.
promoted listings can be dangerous in the wrong hands. some of my clients had the bar way up at 20% – assuming they were giving away 20% of their PROFIT.
Its 20% of their turnover! Some of them were only making a 5% net so it threw them into a loss and they dint realise. Common sense aint very common.
we tested this on one of our own shops for 9 months. we increased prices to help pay for the promoted listings and put the bar at max.
the result was more sales but less profit. not a bad tool with the correct products and profit margins but doesnt really work well for us – from a profit perspective. Good if you want to shift old stock I suppose.
I sell used professional electronics on the US site.
I’ve found Promoted Listings to be worthwhile.
I don’t usually bid anything like the eBay-suggested figure for that category as I think I get a better ROI bidding low across everything, but my stuff is relatively obscure so I don’t think that I have much competition for the assigned promo spots.
My gut feel is that most of the PL sales are ‘incremental’ (defined as selling much faster or at a higher price than if I just let the listing run unpromoted). Often sales are to newer or less savvy buyers who are more easily directed into making a purchase – and/or the item is something that has been sitting around for a while, or is relatively expensive (that ordinarily I would consider as a candidate for a markdown in a few months’ time).
Ball park numbers – bid at approx 3%, PL sales are 25% of revenue, 80% of which is incremental (estimated).
So with those numbers the ‘incremental’ sales cost me 3.75% approx in PL fees (less the allowance included in my store subscription) which I think is worth it in return for a quicker stock turn or not having to lower the price to make a sale. I don’t think that I would get a lot of uplift running a 5% markdown which would be my next best option, so PL seems a good choice.
I’ve also tested going with a higher bid percentage on a handful of items that are under a lot of price competition – as an alternative to joining the race to the bottom on price. I think that may also be a valid approach but it depends very much on whether the competing sellers are also bidding on the same SKU.
I’ll be interested to see how the results of the test compare.
My recommendation would be to test the feature and see how it performs with your stuff. I would start with a low percentage and only raise it by enough to start showing up in the promo spots, which is easy to observe.
Interesting experiment, looking forward to viewing the results.
The experiments I’ve done suggest severely diminished returns on increased spend – especially on more unique items with lesser competition (naturally).
One of the most irritating effects of promoted listings placement in search is the very close proximity to the “original” item. I’ve never seen a promoted item directly next to the source item, but I’ve often seen it only one row away which results in buyers being more likely to arbitrarily click on either surfaced item. Promoted listings need a minimum required placement away from the source item, but again my belief is this is much more likely to happen the less competition the source item has.
There are noticeable flaws in the methodology, which limits the usefulness of the results.
-There is no true control group in David’s assumption (13 and then 9 weeks are over different timeline). What should instead happen is 2x almost identically rated account e.g. Stuff_U_Sell 1 and Staff_U_sell 2 – selling the same items. One has the ads – one doesn’t. You can argue there is substitution effect – but this will show raw visibility.
-Also, it sounds like goods are tracked across different timeframe (market on eBay often have cyclic performance across the year). In essence, David will be measuring that seasonal cycle partially, not fully the performance of the ads programme. It’s possible to adjust for the seasonal trend year on year – but that results in more noisy data.
-Also, given David’s comment, they mostly do one-off items – the output will be limited in usefulness for the different type of sellers, who have fewer listings – but massive quantity of each (David’s test doesn’t doesn’t represent the scaled performance for ads – if you’ve multiple quantities). Because David isn’t doing large quantity on single listings – arguably a listing will never reach its peak performance (e.g. a good selling item in large quantity – might benefit more / or less – we don’t know)
Interesting non-the-less. Keep it up.
Hi Josh — thanks for the comment and interest. You highlight some of the methodological issues which we considered when designing the trial. It may help to allay your concerns if I explain in a little more detail.
We looked at the 13-week control group to establish the distribution of weekly sell-through rates. By looking over this period we are able to identify if something is an unusual week. For example, if you stood next to a motorway for a week and measured car speeds, you might decide the average is 70mph, but not be surprised by anything going between 40 and 100. If something came past at 200mph, you would sit up and take notice. By looking at the distribution of the sell-through rates we could decide what the natural random variation looked like, and so identify if a result were outside that distribution.
Furthermore when we ran the tests, at least half of them were on an even/odd basis, taking a random half of our inventory for the week and dosing it, while leaving the other half untreated. (We picked even listing numbers or odd listing numbers to achieve this random selection). The goal here was exactly to remove the weekly random fluctuations. The sell-through rate in the dosed listings was compared with the untreated listing *in that same week*. Where we treated the entire inventory, we compared with the 13-week control average, taking a two-standard deviation band (95% confidence interval) to test whether we thought the change was significant or not.
I think we’ve taken the question of a control group very seriously and thought about how to mitigate for variations in time.
The mix of inventory also troubled us. We aren’t selling identical items but we are selling a lot, which gives us some statistical overview. And the types of items that we sell are similar in categories — your second-hand shoes aren’t identical to mine, but they are both second-hand shoes and have similar sales characteristics if there are enough of them. Furthermore the repeated random assignment of which half of the inventory was dosed meant that even if there were random variations in sell-through because of the mix of good, that was equally likely to affect the control group as the treatment group. This is identical to the issue faced in medical trials — no two people or treatment periods are the same, you just have to control for that by making is random whether they are included in the control or the treatment groups.
Finally you are quite right that the effect is multiplied for sellers with deep inventory who promote listings to gain sales history and then see a best-match benefit on future sales. We do not have that effect, thus making a trial on our inventory a very pure test of whether the PL program increases sell-through rate directly without the second-order effects. It makes us a harsh example. If it works for us, then it works better for deep inventory. If it doesn’t work for us, then there won’t be the initial spark increased sales history on deep inventory either.
I have tried it few times and despite being a top rated seller and offering 6% more than the trending rate, it didn’t work. Branded popular car battery charger was on page 5 and never moved up. In fact my other chargers were showing in first 3 pages. I called ebay twice and they had no clue and didn’t care. Case was forwarded to “relevant department” and this was two months ago!
I believe we already have the results of this trial from our own experience of promoting “random” items using Promoted Listings consistently for the last 12 months.
What we see is X amount of IMPRESSIONS and X amount of CLICKS per % increase of spend, but SELL-THROUGH rates do not increase in the same way. Sell-through rates as a result of increased spend do not seem to follow the same pattern in any sense or form, thereby in our opinion, casting doubts on the “quality” of the “extra” traffic being received as a result of using promoted listings. We find this is not too dissimilar in ways to our experience with running Display Network ads with Google without any negative keywords. The net result of which (as many who have used Google Display Network Ads in the past) will know, is a ton of wasted spend.
In essence you get lots of extra traffic which simply increases with the more you “offer” to spend, but this does little to increase overall sales volume.
On the plus side whilst we already know “poor quality traffic” will never convert regardless how much you increase your spend, with Promoted Listings this will not cost you any extra money (unless it generates extra sales) due to the CPA (Cost per action) model, but it will mean you’ll pay extra fees on the products that we’re already selling at the lower % that you were previously paying… so be careful here.
We believe that Promoted Listings could be significantly improved however with the ability to define a set of “negative keywords” but obviously we’d need some proper analytics tools from eBay in order to get that working efficiently. Perhaps this was the longer term purpose of the Terapeak aquisition?
We’ve also recently read elsewhere that eBay are proposing to introduce a CPC (Cost per click) model in 2019 and we hope this doesn’t replace Promoted Listings entirely. If it does, we won’t be using Promoted Ads on eBay anymore, that’s for sure. CPC models are usually difficult to profit from with tight margins, especially since we’re already paying eBay/PayPal Fees, there’s nothing left to spend on “clicks without sales” which is what you’ll get alot of in the beginning of any CPC campaign.
When compared to Google or Amazon with whom you are forced into (for the most part) a CPC (Cost per click) method of advertising, and whilst they both have excellent analytical reporting tools available and enable you to filter the traffic to increase your ROI over-time, you’re still going to have to expect to spend a bit (usually, quite a bit) to get the desired results. eBay’s Promoted Listings on the otherhand is a much easier and safer option, especially if you’re new to paying for Sponsored Ads.
So the short answer is YES, Promoted Listings will generate sales, even with a 1% spend, but with little way to filter the traffic every extra % that you offer to “give away” will not increase your overall sales by any amounts substantial enough to qualify the extra spend. Don’t be fooled by the massive spikes in traffic and clicks as you increase your % spend. If you’re going to use Promoted Listings focus only on the sales, and of course, your profit after paying the extra fees.
And one more thing… Don’t run campaigns for long periods of time. We have found that a campaigns performance would decrease in effectiveness the longer it remained active. By simply ending the campaign and restarting it a day later, we found that this would respark the performance and allow us to see more consistent sales overall.
using one off unique items is a basic flaw
if testing asprin you dont use paracetamol
Hi we ran a trial a while ago on some items. 160 to be exact and we saw a slight rise in sales. Yes!!!! we thought that is great we sold 2 extra items.
We were at the time operating at 23% down on this time last year and 19% down on previous months sales.
All items were currently making 14% profit as we had not increased any prices in 3 years to try and hold falling sales and guess what they now made only 8% so we returned back to the basic listing and sales remained the same.
This idea of pay us more and we will show more people your item is just a way of getting more money for old rope as the saying goes.
Just like the TV advert they are currently running.
Two people riding bikes saying.
“I used to think shipping on eBay was expensive but i got these bikes with free shipping”.
Apparently I may have not heard it correctly
What it really said was;
“I used to think shipping on eBay was expensive but I got these bikes with free shipping it was just added on the cost to try and fool me. Because eBay think we are all stupid