SEO
6 Strategic Marketing Goals and How to Measure Them

Marketing goals can be defined as broad, long-term outcomes that a company wants to achieve via marketing efforts.
Setting clear marketing goals is important, as this can effectively focus your team on a shared vision. But the thing is, you need to choose your goals carefully. Otherwise, you may waste a lot of time on things that bring poor results or even undermine your past efforts.
In this article, we’ve curated a short list of strategic goals that are worth considering in any marketing strategy, along with some ideas on how you can measure them:
- Improve product satisfaction
- Grow organic traffic
- Generate leads
- Establish thought leadership
- Increase brand awareness
- Increase revenue
Any successful marketing needs to be founded on a good product that satisfies existing market demand. Otherwise, none of your marketing efforts will “stick.” Meaning, no matter how you promote the product, you will fail to convince your audience and build sustainable growth.
Conversely, a product that users are willing to use, buy, and recommend to others will reinforce all marketing activities. In fact, a lot of successful companies have grown solely or mainly via word-of-mouth recommendations sparked by the remarkable value of their products (e.g., Whatsapp, Tesla).
To set yourself on the right path of improving product satisfaction, you need to achieve product-market fit.
Once you know you’re in the right market with the right kind of product, you can start delighting your users with useful features and a great user experience. Keep in mind that even seemingly simple product improvements can go a long way.
We have finally released this long overdue SEO metric:
🔥 Traffic potential 🔥
Because pages don’t rank in Google for just a single keyword. They also rank for all the different variations of that keyword and get search traffic from them. https://t.co/a0ehTT5WJV pic.twitter.com/zQAXCaQTYj
— Tim Soulo (@timsoulo) November 12, 2021
How to measure
You can measure product satisfaction in two ways: ask your users what they think or gather relevant data from product usage.
Undersökningar
In surveys, you should ask all kinds of questions that help you understand how well your product satisfies users’ needs. You can also use tried and tested methods like the popular and uncomplicated Net Promoter Score (NPS) survey.
This survey comprises just one question: “How likely are you to recommend [product] to a friend or colleague?” The answers are given on a 10-point scale.
You can find multiple tools online that will help you distribute the survey and calculate your NPS (e.g., Hotjar, Survicate).
Product engagement
If you’re running an online service, consider measuring product satisfaction with product analysis tools (also called product intelligence tools) like Mixpanel or Amplitude. They work by gathering data from your users’ in-app behavior and allowing you to analyze the data to gain insights from it.
For example, by measuring how often your users reach out for particular features inside your product, you can see whether those features bring value or not. Then, you can discard unused features based on real data or conduct experiments (e.g., tweaking your features or making them more visible).
User retention
User retention (or cohort retention) is a metric used for measuring the ability to keep customers over a specified period of time.
If your customers go as quickly as they come, this is usually a huge red flag (with some exceptions, e.g., e‑commerce). If you’re not operating in a niche where a short usage period is natural, low user retention is a sign that:
- Users don’t find what you’ve promised in your marketing communication.
- Your product delivers the promise, but your competitors do a better job.
In these scenarios, it’s likely you’re wasting money and brand equity by providing something people are not willing to stay with. So you need to improve your product fast.
That said, even if you have the best product on the market, the so-called customer churn (i.e., when customers stop using your product) is a natural phenomenon to some extent. The key here is to determine whether you have a healthy retention rate.
Organic traffic, also called organic search traffic, refers to the visitors who come to your website via the non-paid search results from a search engine (e.g., Google, DuckDuckGo).
To take advantage of the organic traffic potential from search engines, you need to publicera content based on search demand and the business value of a particular topic (the so-called SEO content).
That way, whenever someone searches for a solution via a search engine, they will find your content and, consequently, your brand and product.
Here are the top reasons why you should join the majority of marketers who invest in creating SEO content to grow organic traffic:
- SEO content can influence and even drive the entire marknadsföringstratt.
- Such content brings almost free, continuous traffic.
- Compounding effects. A blog post written years ago can get you traffic now and into the future as long as you rank high.
- How much your organic traffic grows depends more on content quality and creativity rather than budget.
- The flywheel effect: Content marketing done right can be a self-reinforcing mechanism that helps you get results more easily as you go along.
Let me just add that this is not some hypothesis. At Ahrefs, we’ve been systematically developing search engine optimized articles and videos, and the articles alone bring us approximately 384K organic visits every month.
How to measure
We can recommend two types of tools here.
Firstly, measuring organic traffic is best done via Google Search Console (GSC). This is a free tool that gives the most accurate organic traffic data. GSC will show you the number of clicks coming from Google Search, Discover, or News. It’ll also show you the number of times your content has been displayed by Google (i.e., impressions):

While GSC does a great job of providing these simple metrics, it lacks features and data for comprehensive organic traffic analysis.
This brings us to other types of tools: SEO tools that fill the gaps of GSC, such as the free Ahrefs Webmaster Tools.
For example, while GSC will show you up to 1K keywords and up to 1K backlinks, Ahrefs Webmaster Tools will show you that and many more data points without any limits.
To sum up, you can use GSC for measuring organic traffic and other more advanced SEO tools for SEO analysis and finding growth opportunities.
To put things simply, the more leads you generate, the more revenue you make. This is because every lead is a potential customer. However, not every lead will become a customer, so you need a lot more leads to get your desired number of customers.
A lead is anyone who has expressed interest in a product or service by sharing their contact information (e.g., email address) with a company in exchange for some kind of value (e.g., free ebook, free tool, weekly email newsletter with educational materials).
More often than not, potential customers are not instantly ready to buy. This is especially when they have little or no acquaintance with your brand.
When there is a lot of competition in the market, your potential customers are likely to do some research and compare you to others before they make a purchase. Moreover, if your product is complex and/or expensive, people need to make sure the product will solve their problem or will be worth their money and effort.
This is where lead generation comes in. When a person gives you their contact information, you gain an opportunity to contact them directly in the future. You can use that opportunity to nurture your relationship with them to a point where they are ready to buy.
To generate leads, you will need three things:
- Traffic – In other words, visitors coming through your marketing channels.
- Offer – The value you are going to provide in exchange for contact information (e.g., free ebook).
- Lead capture – A form where people can submit their contact information (e.g., name, phone number, email address).

How to measure
How you measure your lead generation depends on your offer. This can be the number of newsletter subscribers, trial sign-ups, app downloads, or whatever else you are planning to provide.
The simplest way to measure incoming leads is via the same tool you use to capture your leads. For example, our email capture form uses Mailchimp’s functionality. It’s the same app we use to monitor the number of leads and send a weekly newsletter to people who signed up.

You can also aggregate your data in a business intelligence software like Google Data Studio or Klipfolio. Then view the data next to other important metrics for quick insights, such as the conversion rate from leads to customers.
In marketing, thought leadership is demonstrating your brand has expertise in its area of business. Effective thought leadership creates a belief among your target audience that your solutions are the best.
Through effective thought leadership, you become an authority in your industry—that status reinforces every message you send out. And so, in the classic conundrum of whether the messenger is more important than the message, you can actually have both.
The more sophisticated and technically oriented the market, the more thought leadership counts. A good example of this is the electric car market. Tesla is an undisputed thought leader in this area. That’s why it surpasses sales of other established car brands with larger advertising budgets. In fact, Tesla is famous for its anti-advertising attitude.
We don’t buy advertising
— Elon Musk (@elonmusk) April 29, 2019
How to measure
Measuring your progress in becoming a thought leader depends on where you share your ideas. Here, we’ll show examples of two effective channels and their respective metrics.
Quality backlinks
A backlink is a link on one website that links to another website. Backlinks act as votes. Even Google thinks so, treating backlinks as one of the most important ranking factors.
And so if you publish content that gets this kind of vote, you’re on the right track of becoming an authority in your industry. This is because people are digitally voting for what you say, resulting in direct traffic from those pages and higher search engine rankings.
To illustrate, one of the widely discussed subjects in the SEO community is building links through outreach. Our CMO, Tim Soulo, has joined the conversation with an article called I Just Deleted Your Outreach Email. And NO, I Don’t Feel Sorry, which explains how to do effective outreach that doesn’t feel like spam.
That article alone got over 2K backlinks (aka digital votes).
And just a quick reminder—sharing ideas through such articles brings customers:

Speaking engagements
Speaking engagements come in different shapes and sizes. These can be either big industry events like BrightonSEO (with some 4K attendees) or more cozy settings with smaller audiences like podcast interviews.
What they all have in common is getting attention from industry professionals and even industry authorities. So the more you speak at those events, the more likely you are to reach people with your ideas (and your name) and become an authority in your niche.
With speaking engagements, you can put your name on the map, attract followers to your social media channels, and communicate with these followers directly later on.
Once you have more budget, you can even up the ante by creating your own conference, especially if you want to popularize an original concept. That’s what Hubspot has done with the term “inbound marketing” and its INBOUND event.

A brand is a central concept in marketing, and it’s been this way for decades. This is because brands have powerful effects on consumers:
- A brand makes recognizing the product as easy as remembering the word or the shape of a logo.
- A brand evokes associations with positive experiences.
- A brand allows for rationalizing the cost of the product.
Building brand awareness increases the odds of consumers associating your brand or product with a specific need.
Just think about it. Starbucks is one of the most valuable brands in the world. For millions of people, Starbucks is the synonym of coffee. So essentially, it isn’t an exaggeration to say its business relies on a mental association between a logo and a need for coffee. That’s how powerful brand awareness is.
And the amazing part is, however absurd this may sound, the Starbucks logo has nothing to do with coffee. Starbucks has even dropped the word “coffee” from the logo.

How to measure
Measuring brand awareness is the domain of specialized research companies. A common method for measuring it is through surveys. However, this option has its flaws: It’s expensive and time-consuming.
Alternatively, you can gauge the overall trend of your brand awareness yourself using online tools. The only caveat is this method will give more accurate estimations for online businesses than the predominantly offline ones.
You can also use a keyword research tool to discover the search volume of your brand name. The reason is this: If your brand awareness increases, more people will want to buy from you and look up your brand on the web.
For example, if you use Ahrefs’ Keywords Explorer, you can just plug in your brand name and instantly get:
- The number of estimated monthly searches for a specific country (and globally).
- A graph of monthly searches plotted in time that offers quick insights into trends.

You can also easily measure your performance against your competitors’ (technically, this kind of comparison is called measuring the share of voice).
If you’re not an Ahrefs user or just need a point of reference without the search volume data, you can use Google Trends to gauge interest in branded queries.
So far, we’ve discussed rather indirect ways to increase revenue. Now, we’ll discuss three ideas for increasing revenue directly.
The first way is revising your pricing. If you have solid reasons for thinking you’re not charging enough for the value you provide, you can try to increase prices. Even a price increase of a few percent can result in significant returns if multiplied by hundreds or even thousands of new customers. Word of advice: A good practice here is to keep the original price for any existing customers.
A seemingly counterintuitive way (also quite risky) to increase your profits is through lowering prices (e.g., penetration pricing, loss leader strategy). This can lower the barrier enough for the arrival of new customers (you can even win your competitors’ customers this way).
Recommended reading: How to Increase SaaS Prices the Right (and Profitable) Way
The second way is adding new services and/or products. For instance, a dog food brand decides to expand its assortment by offering dog accessories like toys, dog care products, or beddings. It can even create special product bundles and call it “new dog owner essentials.”
The third way is cross-selling and upselling. Cross-selling means suggesting other products in addition to the chosen product. Upselling suggests a more expensive version of the chosen product.
Let’s learn from the best here. When you’re shopping for a new iMac, you will first see a standard price for the product:

Then you will be offered an array of upsell options:

Followed by an even wider array of cross-selling suggestions:

How to measure
The easiest way to measure revenue is to measure the number and the value of sales. But a lot of companies also need to measure recurring revenue from subscriptions, the revenue growth rate, and the value of each new customer.
Recurring revenue
Monthly recurring revenue (MRR) measures how much you’re earning each month through recurring contracts (i.e., subscriptions).
MRR = number of subscribers on a monthly plan * average revenue per user
For annual plans, you have to divide the plan price by 12 and then multiply by the number of customers on that plan.
For example: If you have 700 customers on a $9 per month plan and 100 customers on a $97 yearly plan, your MRR will be:
(700 x $9) + ($97/12 x 100) = $7,108 MRR
If you want to track annual recurring revenue (ARR) as well, all you need to do is multiply MRR by 12.
In our example, that is:
$7,108*12 = $85,296 ARR
Revenue growth rate
Revenue growth rate measures the month-over-month percentage increase in revenue. This metric is an indicator of how quickly your company is growing.
You can measure the revenue growth rate for any period you need: weeks, months, or years.
Let’s say you want to measure the annual growth rate compared to the previous year. The formula for that will be:
(revenue year 2 — revenue year 1) / revenue year 1 x 100 = revenue growth rate (%)
In our example, that is:
($170,592 — $85,296) / $85,296 x 100 = 100% revenue growth rate
Customer lifetime value
Customer lifetime value (CLV) is the total worth of a customer to a business over the whole period of their relationship. CLV can also be used as a predictive metric of how much revenue each new customer will bring on average.
There are multiple models of calculating CLV. Without going into too much detail about each alternative, here’s a fairly simple formula to calculate CLV:
customer lifetime value = average order value x purchase frequency rate x average customer lifetime
Where:
- Average order value is your total revenue divided by the number of purchases.
- Purchase frequency rate is the total number of purchases divided by the number of customers.
- Average customer lifetime is the number of days between the first and last purchase date, divided by 365.
Slutgiltiga tankar
Marketing goals, by nature, are usually grand and ambitious. Hence, they can be quite intimidating.
But no worries. You can overcome that problem by setting achievable goals and breaking your overarching goal into smaller bits. You can see how it’s done in practice using SEO goals as an example in the below article:
Har du frågor? Pinga mig på Twitter.
SEO
Optimera din SEO-strategi för maximal ROI med dessa 5 tips

Wondering what improvements can you make to boost organic search results and increase ROI?
If you want to be successful in SEO, even after large Google algorithm updates, be sure to:
- Keep the SEO fundamentals at the forefront of your strategy.
- Prioritize your SEO efforts for the most rewarding outcomes.
- Focus on uncovering and prioritizing commercial opportunities if you’re in ecommerce.
- Dive into seasonal trends and how to plan for them.
- Get tip 5 and all of the step-by-step how-tos by joining our upcoming webinar.
We’ll share five actionable ways you can discover the most impactful opportunities for your business and achieve maximum ROI.
You’ll learn how to:
- Identify seasonal trends and plan for them.
- Report on and optimize your online share of voice.
- Maximize SERP feature opportunities, most notably Popular Products.
Join Jon Earnshaw, Chief Product Evangelist and Co-Founder of Pi Datametrics, and Sophie Moule, Head of Product and Marketing at Pi Datametrics, as they walk you through ways to drastically improve the ROI of your SEO strategy.
In this live session, we’ll uncover innovative ways you can step up your search strategy and outperform your competitors.
Ready to start maximizing your results and growing your business?
Anmäl dig nu and get the actionable insights you need for SEO success.
Can’t attend the live webinar? We’ve got you covered. Register anyway and you’ll get access to a recording, after the event.
SEO
TikTok’s US Future Uncertain: CEO Faces Congress

During a five-hour congressional hearing, TikTok CEO Shou Zi Chew faced intense scrutiny from U.S. lawmakers about the social media platform’s connections to its Chinese parent company, ByteDance.
Legislators from both sides demanded clear answers on whether TikTok spies on Americans for China.
The U.S. government has been pushing for the divestiture of TikTok and has even threatened to ban the app in the United States.
Chew found himself in a difficult position, attempting to portray TikTok as an independent company not influenced by China.
However, lawmakers remained skeptical, citing China’s opposition to the sale of TikTok as evidence of the country’s influence over the company.
The hearing was marked by a rare display of bipartisan unity, with the tone harsher than in previous congressional hearings featuring American social media executives.
The Future of TikTok In The US
With the U.S. and China at odds over TikTok’s sale, the app faces two possible outcomes in the United States.
Either TikTok gets banned, or it revisits negotiations for a technical fix to data security concerns.
Lindsay Gorman, head of technology and geopolitics at the German Marshall Fund, said, “The future of TikTok in the U.S. is definitely dimmer and more uncertain today than it was yesterday.”
TikTok has proposed measures to protect U.S. user data, but no security agreement has been reached.
Addressing Concerns About Societal Impact
Lawmakers at the hearing raised concerns about TikTok’s impact on young Americans, accusing the platform of invading privacy and harming mental health.
Enligt Pew Research Center, the app is used by 67% of U.S. teenagers.
Critics argue that the app is too addictive and its algorithm can expose teens to dangerous or lethal situations.
Chew pointed to new screen time limits and content guidelines to address these concerns, but lawmakers remained unconvinced.
Sammanfattningsvis
The House Energy and Commerce Committee’s hearing on TikTok addressed concerns common to all social media platforms, like spreading harmful content and collecting massive user data.
Most committee members were critical of TikTok, but many avoided the typical grandstanding seen in high-profile hearings.
The hearing aimed to make a case for regulating social media and protecting children rather than focusing on the national security threat posed by the app’s connection to China.
If anything emerges from this hearing, it could be related to those regulations.
The hearing also allowed Congress to convince Americans that TikTok is a national security threat that warrants a ban.
This concern arises from the potential for the Chinese government to access the data of TikTok’s 150 million U.S. users or manipulate its recommendation algorithms to spread propaganda or disinformation.
However, limited public evidence supports these claims, making banning the app seem extreme and potentially unnecessary.
As events progress, staying informed is crucial as the outcome could impact the digital marketing landscape.
Featured Image: Rokas Tenys/Shutterstock
Full replay of congressional hearing available on Youtube.
SEO
Allt du behöver veta

Google has just released Bard, its answer to ChatGPT, and users are getting to know it to see how it compares to OpenAI’s artificial intelligence-powered chatbot.
The name ‘Bard’ is purely marketing-driven, as there are no algorithms named Bard, but we do know that the chatbot is powered by LaMDA.
Here is everything we know about Bard so far and some interesting research that may offer an idea of the kind of algorithms that may power Bard.
What Is Google Bard?
Bard is an experimental Google chatbot that is powered by the LaMDA large language model.
It’s a generative AI that accepts prompts and performs text-based tasks like providing answers and summaries and creating various forms of content.
Bard also assists in exploring topics by summarizing information found on the internet and providing links for exploring websites with more information.
Why Did Google Release Bard?
Google released Bard after the wildly successful launch of OpenAI’s ChatGPT, which created the perception that Google was falling behind technologically.
ChatGPT was perceived as a revolutionary technology with the potential to disrupt the search industry and shift the balance of power away from Google search and the lucrative search advertising business.
On December 21, 2022, three weeks after the launch of ChatGPT, the New York Times reported that Google had declared a “code red” to quickly define its response to the threat posed to its business model.
Forty-seven days after the code red strategy adjustment, Google announced the launch of Bard on February 6, 2023.
What Was The Issue With Google Bard?
The announcement of Bard was a stunning failure because the demo that was meant to showcase Google’s chatbot AI contained a factual error.
The inaccuracy of Google’s AI turned what was meant to be a triumphant return to form into a humbling pie in the face.
Google’s shares subsequently lost a hundred billion dollars in market value in a single day, reflecting a loss of confidence in Google’s ability to navigate the looming era of AI.
How Does Google Bard Work?
Bard is powered by a “lightweight” version of LaMDA.
LaMDA is a large language model that is trained on datasets consisting of public dialogue and web data.
There are two important factors related to the training described in the associated research paper, which you can download as a PDF here: LaMDA: Language Models for Dialog Applications (read the abstract here).
- A. Safety: The model achieves a level of safety by tuning it with data that was annotated by crowd workers.
- B. Groundedness: LaMDA grounds itself factually with external knowledge sources (through information retrieval, which is search).
The LaMDA research paper states:
“…factual grounding, involves enabling the model to consult external knowledge sources, such as an information retrieval system, a language translator, and a calculator.
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 metrics to evaluate the LaMDA outputs:
- Sensibleness: A measurement of whether an answer makes sense or not.
- Specificity: Measures if the answer is the opposite of generic/vague or contextually specific.
- Interestingness: This metric measures if LaMDA’s answers are insightful or inspire curiosity.
All three metrics were judged by crowdsourced raters, and that data was fed back into the machine to keep improving it.
The LaMDA research paper concludes by stating that crowdsourced reviews and the system’s ability to fact-check with a search engine were useful techniques.
Google’s researchers wrote:
“We find that crowd-annotated data is an effective tool for driving significant additional gains.
We also find that calling external APIs (such as an information retrieval system) offers a path towards significantly improving groundedness, which we define as the extent to which a generated response contains claims that can be referenced and checked against a known source.”
How Is Google Planning To Use Bard In Search?
The future of Bard is currently envisioned as a feature in search.
Google’s announcement in February was insufficiently specific on how Bard would be implemented.
The key details were buried in a single paragraph close to the end of the blog announcement of Bard, where it was described as an AI feature in search.
That lack of clarity fueled the perception that Bard would be integrated into search, which was never the case.
Google’s February 2023 announcement of Bard states that Google will at some point integrate AI features into search:
“Soon, you’ll 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: whether 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 soon.”
It’s clear that Bard is not search. Rather, it is intended to be a feature in search and not a replacement for search.
What Is A Search Feature?
A feature is something like Google’s Knowledge Panel, which provides knowledge information about notable people, places, and things.
Google’s “How Search Works” webpage about features explains:
“Google’s search features ensure that you get the right information at the right time in the format that’s most useful to your query.
Sometimes it’s a webpage, and sometimes it’s real-world information like a map or inventory at a local store.”
In an internal meeting at Google (reported by CNBC), employees questioned the use of Bard in search.
One employee pointed out that large language models like ChatGPT and Bard are not fact-based sources of information.
The Google employee asked:
“Why do we think the big first application should be search, which at its heart is about finding true information?”
Jack Krawczyk, the product lead for Google Bard, answered:
“I just want to be very clear: Bard is not search.”
At the same internal event, Google’s Vice President of Engineering for Search, Elizabeth Reid, reiterated that Bard is not search.
She said:
“Bard is really separate from search…”
What we can confidently conclude is that Bard is not a new iteration of Google search. It is a feature.
Bard Is An Interactive Method For Exploring Topics
Google’s announcement of Bard was fairly explicit that Bard is not search. This means that, while search surfaces links to answers, Bard helps users investigate knowledge.
The announcement explains:
“When people think of Google, they often think of turning to us for quick factual answers, like ‘how many keys does a piano have?’
But increasingly, people are turning to Google for deeper insights and understanding – like, ‘is the piano or guitar easier to learn, and how much practice does each need?’
Learning about a topic like this can take a lot of effort to figure out what you really need to know, and people often want to explore a diverse range of opinions or perspectives.”
It may be helpful to think of Bard as an interactive method for accessing knowledge about topics.
Bard Samples Web Information
The problem with large language models is that they mimic answers, which can lead to factual errors.
The researchers who created LaMDA state that approaches like increasing the size of the model can help it gain more factual information.
But they noted that this approach fails in areas where facts are constantly changing during the course of time, which researchers refer to as the “temporal generalization problem.”
Freshness in the sense of timely information cannot be trained with a static language model.
The solution that LaMDA pursued was to query information retrieval systems. An information retrieval system is a search engine, so LaMDA checks search results.
This feature from LaMDA appears to be a feature of Bard.
The Google Bard announcement explains:
“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.”
LaMDA and (possibly by extension) Bard achieve this with what is called the toolset (TS).
The toolset is explained in the LaMDA researcher paper:
“We create a toolset (TS) that includes an information retrieval system, a calculator, and a translator.
TS takes a single string as input and outputs a list of one or more strings. Each tool in TS expects a string and returns a list of strings.
For example, the calculator takes “135+7721”, and outputs a list containing [“7856”]. Similarly, the translator can take “hello in French” and output [‘Bonjour’].
Finally, the information retrieval system can take ‘How old is Rafael Nadal?’, and output [‘Rafael Nadal / Age / 35’].
The information retrieval system is also capable of returning snippets of content from the open web, with their corresponding URLs.
The TS tries an input string on all of its tools, and produces a final output list of strings by concatenating the output lists from every tool in the following order: calculator, translator, and information retrieval system.
A tool will return an empty list of results if it can’t parse the input (e.g., the calculator cannot parse ‘How old is Rafael Nadal?’), and therefore does not contribute to the final output list.”
Here’s a Bard response with a snippet from the open web:

System för konversation av frågor och svar
Det finns inga forskningsartiklar som nämner namnet "Bard."
Det finns dock en hel del ny forskning relaterad till AI, inklusive av forskare associerade med LaMDA, som kan ha en inverkan på Bard.
Följande hävdar inte att Google använder dessa algoritmer. Vi kan inte med säkerhet säga att någon av dessa tekniker används i Bard.
Värdet av att veta om dessa forskningsdokument är att veta vad som är möjligt.
Följande är algoritmer som är relevanta för AI-baserade frågesvarssystem.
En av författarna till LaMDA arbetade med ett projekt som handlar om att skapa träningsdata för ett konversationsinformationshämtningssystem.
Du kan ladda ner forskningsrapporten 2022 som PDF här: Dialog Inpainting: Förvandla dokument till dialoger (och läs abstrakt här).
Problemet med att träna ett system som Bard är att fråge-och-svar-datauppsättningar (som datauppsättningar som består av frågor och svar som finns på Reddit) är begränsade till hur människor på Reddit beter sig.
Det omfattar inte hur människor utanför den miljön beter sig och vilka typer av frågor de skulle ställa, och vad de korrekta svaren på dessa frågor skulle vara.
Forskarna undersökte att skapa ett system som läste webbsidor och använde sedan en "dialog inpainter" för att förutsäga vilka frågor som skulle besvaras av en given passage inom det som maskinen läste.
Ett avsnitt på en pålitlig Wikipedia-webbsida som säger "Himlen är blå" skulle kunna förvandlas till frågan "Vilken färg har himlen?"
Forskarna skapade sin egen datauppsättning med frågor och svar med hjälp av Wikipedia och andra webbsidor. De kallade datamängderna WikiDialog och WebDialog.
- WikiDialog är en uppsättning frågor och svar som härrör från Wikipedia-data.
- WebDialog är en datauppsättning som härrör från webbsidedialogen på internet.
Dessa nya datauppsättningar är 1 000 gånger större än befintliga datauppsättningar. Vikten av det är att det ger samtalsspråksmodeller en möjlighet att lära sig mer.
Forskarna rapporterade att denna nya datamängd hjälpte till att förbättra konversationsfrågesvarssystem med över 40%.
Forskningsdokumentet beskriver framgången med detta tillvägagångssätt:
"Viktigt, vi finner att våra målade datauppsättningar är kraftfulla källor för träningsdata för ConvQA-system...
När de används för att förträna standardretriever- och rerankerarkitekturer, avancerar de toppmoderna över tre olika ConvQA-hämtningsmätvärden (QRECC, OR-QUAC, TREC-CAST), och ger upp till 40% relativa vinster på standardutvärderingsmått...
Anmärkningsvärt nog finner vi att bara förträning på WikiDialog möjliggör en stark noll-shot-hämtningsprestanda – upp till 95% av en finjusterad retrievers prestanda – utan att använda någon ConvQA-data inom domänen. "
Är det möjligt att Google Bard tränades med hjälp av WikiDialog- och WebDialog-datauppsättningarna?
Det är svårt att föreställa sig ett scenario där Google skulle vidarebefordra träning av en konversations-AI på en datauppsättning som är över 1 000 gånger större.
Men vi vet inte säkert eftersom Google inte ofta kommenterar dess underliggande teknologier i detalj, förutom vid sällsynta tillfällen som för Bard eller LaMDA.
Stora språkmodeller som länkar till källor
Google publicerade nyligen en intressant forskningsartikel om ett sätt att få stora språkmodeller att citera källorna för sin information. Den första versionen av tidningen publicerades i december 2022 och den andra versionen uppdaterades i februari 2023.
Denna teknik kallas experimentell från och med december 2022.
Du kan ladda ner PDF-filen av tidningen här: Tillskrivna frågesvar: Utvärdering och modellering för tillskrivna stora språkmodeller (Läs Google abstrakt här).
Forskningsdokumentet anger avsikten med tekniken:
"Stora språkmodeller (LLM) har visat imponerande resultat samtidigt som de kräver liten eller ingen direkt övervakning.
Vidare finns det allt fler bevis för att LLM kan ha potential i informationssökande scenarier.
Vi tror att förmågan hos en LLM att tillskriva texten som den genererar sannolikt kommer att vara avgörande i den här miljön.
Vi formulerar och studerar Attributed QA som ett viktigt första steg i utvecklingen av tillskrivna LLM.
Vi föreslår ett reproducerbart utvärderingsramverk för uppgiften och jämför en bred uppsättning arkitekturer.
Vi tar mänskliga anteckningar som en guldstandard och visar att en korrelerad automatisk metrik är lämplig för utveckling.
Vårt experimentella arbete ger konkreta svar på två nyckelfrågor (Hur mäter man attribution? och Hur bra presterar nuvarande toppmoderna metoder på attribution?), och ger några tips om hur man kan lösa en tredje (Hur man bygga LLM med attribution?).
Den här typen av stora språkmodeller kan träna ett system som kan svara med stödjande dokumentation som teoretiskt säkerställer att svaret är baserat på något.
Forskningsdokumentet förklarar:
"För att utforska dessa frågor, föreslår vi Attributed Question Answering (QA). I vår formulering är input till modellen/systemet en fråga, och utdata är ett (svar, attribution) par där svaret är en svarssträng och attribution är en pekare till en fast korpus, t.ex. av stycken.
Den returnerade tillskrivningen bör ge stödjande bevis för svaret."
Den här tekniken är speciellt avsedd för frågor som svarar på frågor.
Målet är att skapa bättre svar – något som Google förståeligt nog skulle vilja ha för Bard.
- Attribution tillåter användare och utvecklare att bedöma "tillförlitligheten och nyansen" av svaren.
- Attribution tillåter utvecklare att snabbt granska kvaliteten på svaren eftersom källorna tillhandahålls.
En intressant anteckning är en ny teknik som kallas AutoAIS som starkt korrelerar med mänskliga bedömare.
Med andra ord kan den här tekniken automatisera mänskliga bedömares arbete och skala processen för att betygsätta svaren som ges av en stor språkmodell (som Bard).
Forskarna delar:
"Vi anser att mänskligt betyg är guldstandarden för systemutvärdering, men finner att AutoAIS korrelerar bra med mänskligt omdöme på systemnivå, och erbjuder lovande som ett utvecklingsmått där mänskligt betyg är omöjligt, eller till och med som en bullrig träningssignal. "
Denna teknik är experimentell; den används förmodligen inte. Men det visar en av riktningarna som Google utforskar för att producera pålitliga svar.
Forskningsuppsats om redigering av svar för fakta
Slutligen finns det en anmärkningsvärd teknologi som utvecklats vid Cornell University (också från slutet av 2022) som utforskar ett annat sätt att källattribut för vad en stor språkmodell ger ut och till och med kan redigera ett svar för att rätta sig själv.
Cornell University (som Stanford University) licensierar teknik relaterade till sök och andra områden, tjänar miljontals dollar per år.
Det är bra att hänga med i universitetsforskningen eftersom den visar vad som är möjligt och vad som är spjutspets.
Du kan ladda ner en PDF av tidningen här: RARR: Undersöka och revidera vad språkmodeller säger, med hjälp av språkmodeller (och read the abstract here).
Sammanfattningen förklarar tekniken:
"Språkmodeller (LM) utmärker sig nu i många uppgifter som inlärning av få skott, svar på frågor, resonemang och dialog.
Men ibland genererar de innehåll som inte stöds eller vilseledande.
En användare kan inte enkelt avgöra om deras utdata är tillförlitliga eller inte, eftersom de flesta LM:er inte har någon inbyggd mekanism för att tillskriva externa bevis.
För att möjliggöra attribution samtidigt som alla kraftfulla fördelar med den senaste generationens modeller bevaras, föreslår vi RARR (Retrofit Attribution using Research and Revision), ett system som 1) automatiskt hittar attribution för utdata från vilken textgenereringsmodell som helst och 2) efterredigerar utdata för att fixa innehåll som inte stöds samtidigt som originalutdata bevaras så mycket som möjligt.
...vi finner att RARR förbättrar attributionen avsevärt samtidigt som den ursprungliga indata bevaras i mycket högre grad än tidigare utforskade redigeringsmodeller.
Dessutom kräver implementeringen av RARR bara en handfull träningsexempel, en stor språkmodell och standardwebbsökning.”
Hur får jag tillgång till Google Bard?
Google accepterar för närvarande nya användare för att testa Bard, som för närvarande är märkt som experimentell. Google lanserar åtkomst för Bard här.

Google är på posten och säger att Bard inte är sök, vilket borde lugna dem som känner oro inför AI:s gryning.
Vi befinner oss vid en vändpunkt som inte liknar någon annan vi sett på kanske ett decennium.
Att förstå Bard är till hjälp för alla som publicerar på webben eller utövar SEO eftersom det är bra att veta gränserna för vad som är möjligt och framtiden för vad som kan uppnås.
Fler resurser:
Utvald bild: Whyredphotographor/Shutterstock
-
WORDPRESS2 dagar ago
Intern länkning för SEO: Den ultimata guiden för bästa praxis
-
AMAZON4 dagar ago
De 10 bästa fördelarna med Amazon AWS Lightsail: Varför det är ett utmärkt val för företag
-
SÖKMOTORER7 dagar ago
Uppdatering av Google AdSense Auto Ads Publisher Console
-
WORDPRESS6 dagar ago
De bästa webbhotelllösningarna för din personliga webbsida eller företagswebbplats
-
WORDPRESS6 dagar ago
ActivityPub för WordPress går med i Automattic-familjen – WordPress.com News
-
SÖKMOTORER1 dag ago
Google Search Status Dashboard Lägger till Google Ranking Updates
-
PPC4 dagar ago
PPC Campaign Testing: Dos & Don'ts för att förvandla risker till belöningar
-
MARKNADSFÖRING5 dagar ago
Hur man beräknar kundens livstidsvärde och maximerar det för ditt företag