How Good Data Creates Scalable SEO

Analytics drive decision making for most large organizations. We have more systems in place that can capture information from different touches: like a scanner at a shipping station or point of sale terminal. Much of this data can be aggerated and poured over by data scientists to answer real questions like “If we changed this process, we could ship 10% more orders each day.”

The core of any analytics-driven product-based business is having a good database of item data that aids in smarter decisions around the marketing mix. The structure and accuracy of data helps one make decisions that are scalable and capitalize on new opportunities as marketers take in new information.

Why Data Structure Is Needed

Take an example of curtains which is a product line my company sells. A casual exploration of Google’s Keyword Planner Tool shows that certain colorways get more search volume than others. Grey is the most popular choice for curtains at 40k searches per month, followed by white at 33k.

If I have a database of 600 curtains, how do I quickly apply “white curtains” as a keyword to everything where appropriate? Most ecommerce companies probably have product descriptions saved somewhere – hopefully in a product information management (PIM) system or ERP rather than a Google Sheet. I could export that list, do a filter for everything that contains the word “white” and then add the keyword to every remaining item. The problem is at this number for sku’s there might be a chance that the word “white” still appears in the product description for a non-white curtain. Maybe it has “white lining” or the writer got creative and said the curtain would pair well with “white decor.” Maybe the copy is bad and it doesn’t even mention the color – just assuming that people glean that from the image. Your company might have writing standards but those can still be missed and if the item is an old one it may have been written to different standards. Simple filtering for text on non-validated fields leads to applying keywords to the wrong items or missing opportunities to apply the right ones.

How To Structure Data

This is where it is important to isolate your data and capture it in a Boolean true/false format or through validated pick lists. In the case of curtains, every color is assigned a generic mapping and a specific one. The generic fields are validated to colors like “blue, red, grey” while specific fields tell more information about the shade or specific type. That way one can use a field clearly defined for color to apply keywords (and guide research). This could also help a sales rep to quickly provide a list to customers if they asked for curtains of a certain color, or a product development team to identify gaps in a company’s offerings. We could talk all day about how this impacts non-SEO but my point is this: good data structure helps drive decisions.

In our old format, we had data structured like this:

“Detail field 2: Lined with a white cotton fabric”

We decided to break this into several fields

“Lining Material: Cotton

Lining Color: White”

The benefit of the second one is we can easily tell if an item is lined because the lining fields are filled, and we can write keywords based around the material and color. Whereas the first method can lead to errors due to addressing multiple features at once. One also gains consistency when you remove short blurbs and only input concise and validated data.

Replicate this logic over a large number of fields such as product style, pattern, motif, cleaning method, size, material, shape, etc. and you can do wonderful things like easily create keywords for “large square easy to clean area rugs”. My company has over 4,000 products, but we have a lot of repeatable information like color, style, and item type and I use these to quickly build large lists of keywords that applied to hundreds of items at once like “white farmhouse bedding” and “country red couch pillow”.

For example, the curtains in the image to the right (above if you're on mobile) are all variants based on color - so length, material, and features are all the same. This allows us to provide consistent keywords to product families without manually researching each one. Of course, it's still good to do so, but for companies trying to stay on-trend, you probably do not have the timeline to SEO research each item individually before a launch so starting with a set of trusted keywords can really boost your speed to market.

Turbo Charge SEO With Good Data By Using Excel

Where you work wonders is using excel to concatenate certain fields in excel to do keyword research. Product type, color, and style are three of the most important fields I use in research. I can take lists like this:

Type: pillow case, quilt, sham, euro sham.

Color: white, red, blue.

Style: farmhouse, Christmas, rustic

Then I can list all the values in separate columns and use formulas to build lists like “white pillow case, white quilt, white sham, white euro sham, red pillow case…farmhouse sham, white farmhouse sham…”

Formula-based concatenation can be done for every combination and then those combinations can be plugged into Google’s Keyword Planner or a tool like SEM Rush to get search volume. I did this to build over 2,000 keywords that mapped to our different products. Once all the keywords are pasted back into excel along with the search volume (and hopefully ranking difficulty if you have that info) you can create columns that relate to relevant fields and map those keywords to those fields so that “white pillow case” only appears when the color field is white and the item type field is pillow case.

This can be done to scale and is pretty much necessary for a company with a catalog of large items. If new items fit the keyword criteria, they should be automatically applied to new items – ensuring you never go to market without SEO in place.

Know Thy Product, Know Thy Data

It's only possible to create good structure if you understand your products enough to build out data requirements and have the people and processes in place to capture that information. There will be a lot of planning and work hours to ensure you are capturing the right information, and that the values you are inputting are consistent and clean.

The process of building out the right data formats and then setting up a system to create keywords and map them to items took nearly a year to complete on my first go around, and it wasn’t done alone – a lot of technical work from smart people goes into creating the formulas and scripting to map keywords. If you would like to speak with me about how you can do this for your company, I’d love to get a message from you!