UNDERSTANDING GOOGLE BARD AND ITS IMPACT ON SEO – SARMLife [updated]

Just as ChatGPT is becoming the new normal and the noise is going down, Google hits us with another AI update called Bard.

Google Bard is likely a response to Bing’s ChatGPT and its integration into the search engine, or has it always been on the horizon for Google?

Regardless, we get to experience yet another AI wonder.

The fact that this new tool is coming straight from Google has raised the question of what this means for SEO and SERPs, especially organic searches.

As much as Google hasn’t stopped releasing updates to how searches are answered, from knowledge graphs to featured snippets and rich snippets, are we set to experience major changes to Google Search?

Will Google Bard pose a threat to SEO writers and experts? How much of a change will Bard bring to how we see SEO? 

Read this blog post to find out.

What is Google Bard?

Google Bard is an experimental conversational AI chatbot by Google that can predict reasonable responses based on the prompts it receives.

It is powered by LaMDA (Language Model for Dialogue Applications), a conversational AI model trained on information obtained from web data and public dialogue. 

The model reads through trillions of words and picks up the patterns that make up human language.

These patterns allow it to predict the appropriate responses for different prompts.

The drama that birthed Google Bard

When OpenAI released ChatGPT, it took the world by storm to the extent that its platform could not contain the number of people who wanted access to the chatbot. 

It was such an amazing experience, and I remember wanting to write a blog post about it but couldn’t get access, and when I finally did, I was ecstatic.

Afterward, Bing announced that it would add ChatGPT to its search engine, which was mind-blowing. Many people started using Bing to see how ChatGPT would work with a search engine.

This created the idea that Google was falling behind in technology, which is probably why many people think Google created Bard in response to ChatGPT. 

However, Google’s CEO said Bard was created to “translate deep research and breakthroughs into products that truly help people.”

The launch of Google Bard was announced 68 days after the launch of ChatGPT on February 6, 2023. However, the announcement of Bard failed based on a factual error showcased in the demo.

According to Google Bard, at the time of the demo, it claimed that the James Webb Space Telescope took the first-ever picture of an exoplanet. However, it happened in 2004, and James Webb Space Technology took its picture of an exoplanet in September 2022.

Due to this error, Alphabet’s shares lost a hundred billion dollars in market value in just one day. People lost trust in Google’s ability to make a breakthrough in conversational AI technology.

Google, however, opened access to Bard on March 21, 2023, but access was limited to people in the US or UK; users also had to be over 18 years of age and join a waitlist. 

However, on May 10, 2023, Google removed the waitlist and granted access to users from 180 countries.

How was Google Bard trained?

Many have wondered how Google Bard works and if it was trained with the same model as ChatGPT (GPT-3.5).

Although Google doesn’t explain the training process, we are sure it involved LaMDA, a large language model (LLM) built on a neural network architecture called Transformer.

LaMDA, according to Google, can engage in a free-flowing conversation with an open-ended nature that allows it to move from one topic to another. 

It means that a conversation about a newly released American song can transition into the rise of K-pop and the division of Korea.

When LaMDA was trained, there were some important factors described in the associated research paper:

a. Factual grounding from external knowledge sources

This means Google Bard can consult external knowledge sources like an information retrieval system, a calculator, and a language translator to make its responses accurate and not just give responses that sound plausible but are nonsensical.

In other words, it can fact-check its responses, and when necessary, it can reference its sources.

According to the LaMDA research paper, it states that:

“We quantify factuality using a groundedness metric, and we find that our approach enables the model to generate responses grounded in known sources rather than responses that merely sound plausible.”

Google used three groundedness metrics to evaluate the factuality of Bard’s output. These metrics are:

• Sensibleness: This evaluates if the answer makes sense or not.
• Specificity: This assesses the contextual relevance of the response to the prompt.
• Interestingness: This measures if the answers are insightful and inspire curiosity.

b. Data annotation by crowd workers

According to the associated research paper on LaMDA, it was stated that “We find that crowd-annotated data is an effective tool for driving significant additional gains.”

Data annotation by crowd workers means Google employed the services of humans to evaluate the data before feeding it back to the model. 

How does Google Bard work?

Google Bard came amidst much controversy, and we kept getting many unanswered questions like, “How does it work?”

The way Google Bard works is closely related to its training, which is the same for most language learning models.

Google Bard was trained using a lightweight version of LaMDA. This version of LaMDA is a smaller model that does not use much computing power, making it easy for Google to scale to more users and get more feedback.

What happens when you place a prompt in Google Bard?

Google did an amazing thing with Bard by giving it internet access. The researchers behind the creation of LaMDA recognized a problem with large language models—they were mimicking answers—which led to factual errors. These errors occur because they only reproduce and predict from the data given to them.

However, there is something called the temporal generalization problem, which is changes in facts based on new technology, scientific research, etc. The large language models will still reproduce the old facts when these facts change. 

This is why the concept of freshness was introduced to Google Bard through its ability to access information retrieval systems (search engines).

It was confirmed in the Google Bard announcement, which states: 

“Bard seeks to combine the breadth of the world’s knowledge with the power, intelligence, and creativity of our large language models. It draws on information from the web to provide fresh, high-quality responses.”

So, when you input your prompts on Google Bard, it uses the data it has been trained with and fact-checks with the information retrieval system to give fresh, sensible, and specific answers.

Uses of Google Bard

There are many uses for Google Bard, and it’s impossible to list them all.

Here are some of its uses:

• Generates codes
• Generates topic ideas
• Summarizes texts
• Writes outlines
• Translates texts
• Solves math problems
• Develops a detailed plan
• Makes simple inquiries
• Gives real-life advice

The list is endless!

You can use Google Bard in English, Japanese, and Korean; it is available in over 180 countries!

How to use Google Bard for SEO (with sample prompts)

Here are some SEO tasks I’ve done with Google Bard:

1. Content idea generation

You can use Google Bard to generate ideas for any niche of your choice. It helps you to save time trying to come up with new ideas. 

The answers you get may not be perfect, but you can tweak them to fit your preferences.

I asked Google Bard to come up with five blog post ideas for an SEO agency, and it gave me some comprehensive ideas I can develop into power blog posts and regular posts as well.

Here is another prompt where I asked Bard to develop blog post ideas for a fashion brand, and the answers were fairly good.

2. Keyword Suggestions

You should know that when doing SEO, you must ensure your choice of keywords is verified and thoroughly accessed because your keyword choice can make or break your content’s visibility in search engines.

Remember, it is a KEYword!

So, no matter how great a chatbot or any large language model software is, try to do keyword research as manually as possible using the right keyword tools.

Also, I asked Google Bard to develop keyword suggestions for a blog post on ‘The Ultimate Guide to SEO for Small Businesses.’ Here’s the list it gave me.

Only a few of these keyword suggestions seem related to the blog post idea. Most of them are also too long and won’t do.

However, I can work with a few of them. The next step is to list these keywords and carry out proper keyword research on them.

I also asked Bard to separate these keywords based on their length (short, medium, and long-tailed).

3. Keyword intent

You can also use Google Bard to check for the search intent of various keywords. You can ask Bard to give you the search intent of all the keywords it listed, as I did here:

Or ask for a specific intent like this:

I discovered that the quality of my result was richer when I requested a specific type of intent instead of asking Bard to separate all the intents (i.e., navigational, transactional, informational, and commercial).

4. Headline generation

I was impressed with the optimized headline options I got from Google Bard.

It gave me five (5) different options, and this was the best of the three drafts.

To confirm the authenticity of these headline ideas, I verified them using IsItWP Headline Analyzer, and these were the headline scores they got:

Some of the headlines did fairly well, while one of the headlines even got a green color! The headline suggestion by Bard seems to be above average, and these were just five (5) options from a single draft.

5. Outline generation 

You can use Google Bard to generate an SEO-optimized outline. It is useful and saves time with content research. 

However, you still need to do basic content research to ensure you are getting all the points required for the post you are working on.

I tested out Google Bard on the blog post idea: ‘Ultimate Guide to SEO for Small Businesses,’ and these were the results I got:

This is the first draft, but I needed more detail for this particular content.

The second draft is a more appropriate outline for this particular blog post. If you want to rely on using Google Bard for SEO, go through all the drafts before exporting the answers to wherever you want to use them.

This last draft is too simple and does not hold the context of the blog post topic.

6. Content writing

You can use Google Bard to come up with your blog post content.

While Bard might not give you that 1,500-word blog post, you can use it effectively if you provide prompts for your subheadings instead of the entire post.

For example, suppose I’m writing a post on ‘Guide to SEO for Small Businesses’ instead of giving Bard a prompt on writing content for the topic. In that case, it’s more effective if I give it a prompt for each subheading, e.g., define SEO, benefits of SEO for small businesses, etc.

Here is Google Bard giving me content for my introduction:

The next picture is Google Bard giving me content for my first subheading:

If you notice, you will discover that most of the lines in the introduction are also in the definition of Local SEO I just got.

Here it is if you need another reminder to ensure you go through AI-generated content and not just copy and paste.

And although the first part of the definition was less detailed than I wanted, Bard ensured to include additional information to make up.

The last part of the answer to my prompt contained information on ‘important factors that affect Local SEO ranking.’ While I think that would have easily been another subheading, it still makes sense as the last part of the definition.

7. Code generation

Of course, I had to try using Bard to generate SEO-relevant codes. 

First, I used it to generate a hreflang code for people with multiple versions of their websites in different languages.

Afterward, I tried generating a redirection code, but Bard wasn’t having it with me, as you can see here:

Either Bard needs to learn how to generate a redirection code, or I’m using the wrong prompts.

I also tried to generate a robots.txt code, and it worked.

It also included a source for additional information on creating a robots.txt file.

Lastly, I attempted to create a schema markup code for a recipe page, and I must admit that I was impressed with the results.

Not only did I get my code, but I also got meaningful additional information.

1. Google Bard explained what schema markup was all about and shared tips for adding schema to recipe pages.
2. It indicated the case study source used to generate the schema code.
3. It gave me links to search schema markup-related topics on Google.

That was impressive, and if it gave answers for all prompts like this, I doubt we would have any complaints.

8. Statistics

Google Bard can also create statistics content for you.

At first, I got this type of response when I tried using Bard to create statistics.

Then, I changed the phrasing of my prompts and finally got an answer.

However, you have to be careful when using statistics generated by Bard because some of the statistics I got were inaccurate, nor were the sources indicated.

Google Bard can perform other tasks, such as translating keywords from one language to another. It can also summarize and paraphrase writing pieces to make them simpler or more professional.

Like every other chatbot, the results from Google Bard largely depend on your prompt.

Like the example I gave when discussing using Bard for keyword intent, I couldn’t get a more accurate output until I used specific prompts. The same goes for using Bard to write your content. 

I discovered this when using Bard: break your prompts into smaller chunks to get the best output, especially when dealing with informative or long-form content.

Pros of using Bard for SEO

Here are some pros I’ve discovered about using Google Bard for SEO purposes:

a. Integration with Docs and Gmail

Using Google Bard, you can export your files to Google Docs or as an email draft.

This is convenient if you want to create simple content. However, you cannot export different answers to the same document.

For example, if you are working on a blog post using several prompts, you cannot export all the answers to a single document. Each time you export your answers, a new document automatically opens.

b. Additional information

Using Google Bard, you get additional information as part of your answers.

For example, if you ask for local SEO tools, you can get local SEO tips to enhance your SEO strategy.

Here’s an example of me asking for powerful blog post ideas, and Google Bard gave me additional tips for writing these types of posts.

c. Answer sources

Google Bard gives you the sources of some of its answers.

However, the sources are not always accurate. 

Here’s an example:

From the keyword suggestion prompt, I asked Google Bard for its sources, and initially, it claimed to have gotten them from these platforms:

However, when I clicked on the second draft, the sources were not completely aligned with the first draft.

The differences might indicate that Bard didn’t use the sources as claimed. However, it could also mean that it used all the sources listed but put them in different lists.

When I created a content writing prompt for Bard, I scrolled down and discovered that it gave me links to some sources it used in answering that particular prompt.

Unlike the previous scenario where I had to request the sources, they didn’t have hyperlinks.

It’s safe to say that the sources cited by Google Bard are functional sites with slightly relevant content to what is being discussed on the chatbot.

Here’s an example:

This website is one of the cited sources by Google Bard on my prompt: ‘Write a 500-word introduction for a blog post on The Ultimate Guide to SEO for Small Businesses’.

While this might not be in the same context, it is still about SEO.

Here’s what Google has to say about Bard citing sources:

d. Alternative drafts

Google Bard gives you different drafts with different answers to choose from.

There are some prompts where I get two drafts, while I also get three for some.

Here’s a sample of a prompt showing three different drafts:

e. Option to further search via Google

Using Google Bard, you also get an option to further your search through Google, probably because Bard outsources answers from information retrieval systems, and Google has strong leverage over other search engines.

The button to further your search via Google is at the end of the answer to your prompt.

A click will take you to another tab to google your prompt and get relevant blog posts.

You can also search for related topics to your prompts directly from Google Bard. It analyzes the prompts you have inputted and suggests related searches.

For example, my prompt was ‘Guide to SEO for Small Businesses,’ and Bard could predict these three queries to further my search.

f. Future improvements

One of the advantages of Google Bard is that it knows what’s currently in the market, gets feedback from its wide range of users/testers, and gets more opportunities to improve its features.

The future improvements for Bard are numerous; this is an advantage because they don’t need to commercialize what has been tested and failed. Rather, Google can improve its successes and make better user options.

Google has already stated some improvements to Bard, and they include:

• Images in prompts
• More visual interactions
• Coding upgrades
• More precise citations
• Extensions

Limitations of Google Bard

Despite how great Google Bard is, there are some limitations.

Google itself states on Bard’s homepage that it has a lot of limitations, and as such, you have to be watchful when using it and make significant suggestions to improve its features.

Here are some limitations of Google Bard:

a. Inability to hold context

Bard cannot hold context (i.e., it cannot remember previous conversations). While I tried to verify this claim, I discovered that Bard could hold context adequately.

Here is a sample:

Before this prompt, Bard had given me some blog post ideas, and I had asked it some follow-up questions. I then proceeded to ask another question without including the subject. 

As you can see, I used ‘this’ to see if Bard could understand what ‘this’ in the context of my question meant.

And surprisingly, it did!

While it might be true that Bard cannot hold context in some cases, this attempt shows that it can remember previous conversations to a certain extent.

b. Hallucinations

According to Google, when prompted, Bard may give false positive and false negative answers. This can make you skeptical about trusting this chatbot mindlessly, and rightly so.

c. Biases

Google has stated that Bard can potentially create biases based on the training data it has learned over time because it is obtained from varying sources with different viewpoints and stereotypes. This learning model can incorporate specific opinions and leave out the rest.

This limitation might make Bard develop a kind of ‘personality,’ resulting in inappropriate or inaccurately biased results.

Impact of Google Bard on SEO and Search

The controversy about how Bard will change Google Search might have stemmed from the statement released by Sundar Pichai, the CEO of Google and Alphabet.

Sundar Pichai, in one of the blogs released by Google (An Important Next Step on Our AI Journey), stated:

“Soon, you will see AI-powered features in Search that distill complex information and multiple perspectives into easy-to-digest formats, so you can quickly understand the big picture and learn more from the web: either that’s seeking out additional perspectives, like blogs from people who play both piano and guitar or going deeper on a related topic, like steps to get started as a beginner. These new AI features will begin rolling out on Google Search very soon.”

First, ‘features’ refer to a part of something and not necessarily the entirety of it.

Here’s an example of an AI-powered search result from Google.

We need to understand that Google has always updated the way Search is done over the years, and it is only appropriate they integrate AI features into their search engine, but that doesn’t mean that Bard will change Search.

Yes, there might be significant changes in Search that Bard may influence; however, Bard cannot replace Google Search, just like ChatGPT did not replace Bing.

Google Bard might only be a feature in Search, just like featured snippets, knowledge panels, and PAA was added to how results were shown on Google’s SERP.

However, if Bard is made a major part of Google Search, It might affect the results, especially regarding organic traffic.

Recent Google Bard Updates to Know!

As I stated earlier, Google has the advantage of getting insight from other AI chatbots like ChatGPT, and if they leverage these insights coupled with customer feedback, they have an advantage. 

In two months, Google Bard has made several significant updates, some unprecedented changes.

I’ll explain these details with screenshots to show you how they work.

1. Images in Bard

Google Bard can now bring you images from Google Search and its response. It allows you to visualize the concepts of what you are searching for.

I tried to see this feature’s reliability and requested images along with my prompts.

Here are the results:

The incredible thing about this feature is that, with just a click on the image, Bard will take you to the source of the image.

For example, when I clicked on the images I got in the response, it took me to the source website.

2. Google Lens in Bard

Aside from the fact that you can now see images in your responses, you can also upload images along with texts, just like when using Google Lens to search the Internet.

All you need to do is go to this section on Bard.

 After clicking on the link, you will get this prompt:

Note that this feature is only available in English.

3. Location-Specific Information

According to Google, Bard can now give more specific responses based on your location, but only if you allow it to.

This feature allows Bard to respond to relevant and accurate queries, like when looking for a local business near you.

You can update your location here:

I couldn’t get a screenshot for this feature because Bard says I can’t update my location.

You can manage your preferences in the location settings.

4. Export Tables to Google Sheets

Like the Gmail and Google Docs export options, exporting to Google Sheets allows you to seamlessly export tables developed by Google Bard to your Sheets.

This is good news for people still learning to use Google Sheets/Microsoft Excel. 

You can ask Google Bard to create a personalized table for you and include the needed features. Afterward, you can export the table to Google Sheets with all the features.

When you click on the ‘export sheet’ option, it shows you when your Google Sheet has been created at the bottom of the page:

When you open the sheet, it directly opens a Google Sheet for you and redirects you to the file like this:

You can edit the table and add other data as needed.

5. Improved Math and Data Analysis

Here’s what Google says about this new feature:

“Starting with English, we’ve updated Bard to detect computational prompts and run code in the background, making Bard better at mathematical tasks, coding questions, and string manipulations.”

It means that Bard can detect questions that are mathematical problems and use the appropriate code to generate a response.

For example, if you ask Bard, “What is the square root of 16?” Bard can now detect it as a mathematical question and run the code to calculate the square root of 16. 

This will give you a more accurate response than if Bard tried to answer the question based on its understanding of the English language.

Here are some screenshots of Bard solving some mathematical problems:

6. Export Python Codes to Replit

Google Bard is improving its coding abilities and quality. Here are some recent codes generated by Google Bard:

1. A structured markup CSS code

2. A sample Python code

Note that the export option for codes is limited to Gmail and Docs. However, Google is trying to update these options, starting with Python code.

When Bard generates Python code for you, it allows you to export to Colab or Replit. 

I couldn’t take screenshots of the Replit platform because I don’t have a Replit account. However, you can create an account if necessary.

Here’s a screenshot of Python code in Colab.

7. Modify Responses

Google Bard allows you to modify your responses to make them more professional, simpler, shorter, longer, or casual.

This shows that the responses from Bard will soon become more to the user’s liking, and I hope they can add more modifications to Bard to make its output alignable with the user’s preferences.

8. Share Your Conversations

You can now share part or all of your Bard conversations with friends, teammates, or your social media audience. Google Bard has made it easy by creating a link and allowing you to choose what you want to share.

It starts with you clicking the share button.

Select the part of your chat you want to share.

And copy the link directly and share it on your platform of choice or directly on LinkedIn, Facebook, Twitter, or Reddit.

9. Pinned and Recent Threads

On Google Bard, you can pin your chats and revisit them when necessary. This comes in handy when you constantly use Google Bard and need to keep track of the content you have on the platform.

Pinning your threads will help you quickly access your important threads on Google Bard.

You can either pin, rename, or delete your threads on Google Bard:

Here’s how your pinned threads look on Google Bard:

10. Voice Output for Prompts

Voice search is here to stay, and the populace encourages it more as the days pass. This might be why Google strives to ensure that it incorporates voice technology in the recent updates to Bard.

The recent Bard update allows you to select a voice output option if you want the result read.

I tried to confirm if one gets a voice output for voice input and discovered that you must manually select the voice output option to get the results read.

It may be inconvenient, mainly because most people do voice searches to avoid the hassle of manually reading results. It would be great if the voice output had a manual option and an automatic action.

For example, if I’m using my voice to ask Bard about a particular topic, I might want the answers automatically read out.

However, I love the way the voice output for codes is read. Instead of just reading out the letters and whatnot, the voice output explains the codes logically.

11. New Places, New Languages

As of July 13, 2023, Google Bard is said to be available in over 40 new languages, including Simplified and Traditional Chinese, Hindi, Spanish, Arabic, German, etc.

Its access has also been extended to all 27 European Union (EU) countries and Brazil.

Here’s a list of some of the languages supported by Bard, as seen on the FAQ page of BARD:

• English
• Japanese
• Korean
• Arabic
• Bahasa Indonesia
• Bengali
• Chinese
• Croatian
• Czech
• Danish
• Dutch
• Estonian
• Farsi
• Finnish
• French
• German
• Gujarati
• Greek
• Hebrew
• Hindi
• Hungarian
• Italian
• Kannada
• Latvian
• Lithuanian
• Malayalam
• Marathi
• Norwegian
• Polish
• Portuguese
• Romania
• Russian
• Serbian
• Slovak
• Slovenian
• Spanish
• Swahili, 
• Swedish
• Tamil
• Telugu
• Thai
• Turkish
• Ukrainian
• Urdu
• Vietnamese. 

Google has mentioned that it hopes to include more languages, but currently, Bard covers most of the languages used on the Internet.

However you choose to use Google Bard, it is important to be careful of the information you input into the software. Human reviewers may process your Bard conversations for research and quality purposes.  

Final Thoughts

From my research into using Bard for search engine optimization, I think Google was right to emphasize that Bard is still an experiment, which means we should expect more features and improvements to how it currently works.

Google Bard presently has limitations that can impact SEO if used independently, like hallucinations, inability to hold context and incorrect information.

However, I like that it is integrated with major apps like Google Docs and Gmail, making it easy to export your answers.

The ability to provide additional information and further your queries through Google Search is a smart move, and I hope these features survive the updates that Bard might go through.

Aside from Google Bard, the use of artificial intelligence to simplify SEO is something that should be embraced but used with caution.

No matter how much data a system is trained with, it cannot accurately reproduce human emotion, which is a needed ingredient in creating content that resonates with your audience.

So, regardless of your AI chatbot of choice, ensure your final content has human input.

CTA: Are you embracing the integration of artificial intelligence into search engines? What are your thoughts?

READ MORE: How Every SEO Pro Can Use ChatGPT