Imagine the following situation: It’s early November, Black Friday is getting closer, and you have figured out that you have more than 5000 products whose prices are not competitive enough, because you have never compared them to the prices of your competitors.
You contact your repricing company, asking them if they could match these 5000 SKUs across 5 most important competitors – and your vendors responds: Sure, would love to do that, we can deliver first results by the end of January. By then, Black Friday, and Christmas sales would be long gone, the whole project loses the point, and you drop the idea all together.
So, what can you do to prevent such situations? This is the main question we will be tackling in this post.
Why does product matching take so long?
There can be many reasons for product matching in eCommerce to take too long, we will now mention the most common ones:
- Your vendor cannot scale their operation: usually this means that your vendor is performing product matching manually, and they cannot increase the size of their team performing the matching
- Your vendor is not using automated product matching
- Your products cannot be matched fully automatically
- Your competitor Websites do not utilize same product identifiers (EAN, ASIN, UPC) as you do
What is the expected speed of manual product matching?
The answer depends on several parameters
- The nature of products that need to be matched (technical products (like consumer electronics) are easier to match)
- The speed of competitor Website, where your products need to be matched. If the Website is slow, that will impact the speed of manual work performed
- The experience of the product matching team. Naturally, more experienced agents will perform matching tasks faster (and with higher accuracy) than newbie agents
Although there are so many factors behind the calculation, we can say that in slower cases, you can expect between 30 and 60 matches to be attempted per hour. (Please note that this does not mean 30-60 matches established! There will be many products with no matches (please see our article on expected matching rate), but they also take time because the agent needs to determine that there are no matching products).
Lets give it a numerical example: say that you need 5000 products matched across 5 most important competitors.
That means 5000 x 5 / 30 (or 5000 x 5 / 60) = 833 (or 417) man-hours needed. If your team consists of 5 agents, and if they work 8 hours per day – this means that your manual matching task will take 833 / (5 x 8) = 20.8 days in pessimistic scenario, while it’s going to be 417 / (5 x 8) = 10.4 days in a more optimistic scenario.
The more team members you have, the faster the execution. Of course, if your team works more than 8 hours per day, this will also positively impact the project speed. Of course, one should be careful with the amount of working hours in a day, because the quality of matches established can significantly drop if the agent performing the work feels exhausted or overworked.
What is the expected speed of automated product matching?
Automated product matching is by far the quickest method of product matching in Ecommerce. Lets now discuss some real numbers.
If we’re talking about several thousand products (lets say 5000) – it would be reasonable to expect the results to be delivered in 7-10 days time. For more products, the expectation goes higher, but not linearly (for example: an Automatch task of 100 000 SKUs should take no more than 20 days time.
The above calculations are strongly impacted by number of proxies your Automatch provider has in place. In case of fewer number of proxies, the time required goes up, because your provider should not risk that his IP addresses get banned by competitor’s Website, because of too much traffic generated.
What is nice about Automatch is that several competitors can be automatically matched in parallel – meaning that for 5 competitor Websites (with 100 000 SKUs to be matched) – the total delivery time still stays at 20 days.
Of course, the principal problem with Automatch still remains – it is not always applicable (better to say, it’s applicable in no more than 30% cases, depending on client’s industry)
What is the expected speed of AI-assisted product matching?
To elaborate this question we should first remind ourselves of the nature of AI-assisted product matching.
- Step 1: Automatch delivers potential matching candidates. As we have seen, this should take between 7 and 20 days time, depending on the number of products (SKUs)
- Step 2: AI processes the candidates and establishes correct matches. This process typically very quick, and takes no more than 2-3 days
As you can see, AI-assisted product matching is superior when it comes to the time needed to execute the project.