Chat-first search, where people use conversation interfaces like chatbots to find information, is poised to overtake traditional search engine usage sooner than you might expect. Some estimates indicate that in as little as 3-5 years, chat-first could account for 50-70% of all searches (When Will ChatGPT Replace Search? Maybe Sooner Than You Think | PCMag). The last big shift in search was the rise of mobile-first searches which overtook desktop searching by May of 2015 after technically being around since 1997. While nonprofits had a decade plus to prepare for mobile-first changes, chat-first changes will come far faster. Nonprofits that are reliant on organic search traffic need to start optimizing for this new reality now to mitigate the drop in engagement and support as the sea change happens.
With the public launch of Google’s Bard chatbot which uses their LLM and search data as of May 2022 the clock has started on how the largest search engine will be shifting user behavior (Bard FAQ on how it works).
Jump to something more interesting:
A Visual Example of Chat-First Search
Sometimes a picture (of a search result) is worth a thousand words, so let’s show exactly what we are seeing in a chat-first search.
The following is a search for the longtail term “should my nonprofit use AI tools?” on Google present day (Q2, 2023). Whole Whale’s carefully created AI guide shows up after a featured area, driving a decent amount of clicks.
This screenshot is from the Google Bard chat results for the same question (May 2023). Note the complete lack of any citation to resources, removing any chance of traffic back to a website creator:
This next screenshot shows an example of the same search on Bing’s new chat-first search interface. Notice how Whole Whale’s same content is now pushed into a footnote, while the full answer is provided.
This next example shows this same search result with Perplexity.ai, one of the more advanced chat-first search solutions. Again, Whole Whale’s content is shown as a footnote in the context of a full answer.
How to create an AI Chat-first search traffic risk audit?
Search behavior is essentially divided into 3 buckets: informational, navigational, and transactional queries, each with different risks in a chat-first world:
- Informational: Highest risk of reduced traffic. As chatbots and voice assistants provide direct answers, people may not click through to content. However, an opportunity exists to rank for long, complex informational queries if you publish helpful, authoritative content.
- Navigational: Moderate risk of reduced traffic. While some voice searches for your organization’s name may decrease over time, improving your chatbot or voice experience can build new entry points and relationships.
- Transactional: Higher opportunity for increased traffic and conversions*. A smooth transactional experience via chat or voice can improve satisfaction and lead to more donations, petitions, or event signups. Focus on simplifying transactions through these interfaces.
*However, there may be increased risk to conversions on site as the transaction point may move toward the actual chat interface as tools like ChatGPT Plugins and AutoGPT advance. IE if someone can make a donation to your charity through a chat interface because the AI assistant has access to the user’s wallet.
Here is an audit you can take to gauge your nonprofit’s traffic risk:
- Create a CSV list of your inbound keywords with traffic #s from Google Analytics or Search Console for the past year.
- Try to categorize your keywords into the 3 buckets above.
- Apply a discount rate from 10% to 90% for each category accordingly.
And now the scary part for nonprofits, let’s talk Ad Grant…
The Google Ad Grant is possibly the most generous in-kind donation to the nonprofit sector in history, offering $10k per month in text search ads to qualifying nonprofits.
- Take a look at your Google Ad Grant and pull a past year’s keyword traffic report
- Organize and categorize the keywords with the buckets above.
- Apply a discount rate of anywhere from 10% to 100% reduction of traffic. Yes, Whole Whale believes this chat-first search shift may make the Ad Grant as it is currently designed obsolete.
Why will Chat-first search happen faster than Mobile-first search?
The rise of chat-first search overtaking traditional search inside the next 3 years will be spurred by the following factors:
1. ChatGPT was the fastest product adoption in history. Billions of people already use chat apps like Facebook Messenger, WeChat and WhatsApp. Integrating search into platforms users know and love reduces friction and encourages adoption.
2. Voice assistants like Alexa or Google Assistant have introduced tens of millions of people to conversational interfaces. Interacting via voice is conditioning us to speak more naturally to get what we need.
3. AI and natural language processing have enabled far more sophisticated chatbots and voice assistants. Advancements in NLP are key to providing the seamless, “human” experience needed for chat-first to become mainstream.
4. Chat-first offers instant information on a given topic based on a summary of the top 3-5 links. For example here were top searches in 2022 According to Google which all could be answered by a GPT tool.
What should nonprofits do to prepare for Chat-first Search?
A move to chat-first search means nonprofits will need to make some changes to navigate the transition:
1. Perform the Chat-first search risk audit outlined above to start the conversation internally. Or work with a firm like Whole Whale to do this 🙂
2. Establish your organization as an expert. Create authoritative, well-researched resources that establish your expertise. Backed by evidence and expertise, your content is more likely to be recommended by AI systems and cited if there are unique opinions. Avoid flooding your content with generic GPT, Grey Jacket content that looks like everyone else’s, because it will be wiped out and worse decrease audience trust.
3. Focus on high-quality user experiences including video. Aim for highly relevant, useful and satisfying media content that keeps users engaged in ways that text just doesn’t offer (yet). Remember that the second(ish) largest search engine is YouTube and as misinformation rises it may push more people to get information from verified sources they can see. Start building your video-adjacent content library now.
4. Track upstream search-side metrics. Analyze metrics like number of turns, queries per session and bounce rate within chat interfaces to better understand how people are engaging with your content so you can optimize accordingly. Tools like Bing Webmaster tools and Google Search Console will both evolve to show websites how their content may have been shown in referenced footnotes and potential transactions. Start familiarizing yourself with them now.
5. Build content on very long-tail keywords and natural language questions. Optimize for the types of specific complex queries common in chat to rank higher in results. Provide comprehensive, helpful answers to common questions.
6. Begin planning for your chatbot. The future of your site actually being a chat-first experience is rapidly approaching as users become normalized to interacting with the UX. Begin developing and identifying the content for your chatbot will need to be trained and pushed into a vector embedding. The content experience in conversational interfaces can be created for internal use cases to begin before external ones are tried. Note that we don’t think current interactions of GPT LLM models are safe for any user-facing interactions involving sensitive issues.
Whole Whale’s CauseWriter.ai is currently helping many nonprofits navigate this type of internal chat solution.
Building a chatbot will involve continuously optimizing your content and chatbot engine. Keep tweaking your content and conversation interfaces based on feedback and insights. Optimize for the types of natural language questions that drive the most traffic to stay ahead of changes.
Don’t Blindly Adopt this Tech
Our concern about public-facing chatbots is important so we’ll say it another way, current chat models are fine
if when they ‘hallucinate’ when giving you a fun crepe recipe and call for too much flower. It is not safe when, for example, they do this when talking with a teen in crisis situations. There is huge potential for custom chats to be built for internal teams that need to respond to a variety of situations and content creation demands.
Consider the ethical implications of any models that your nonprofit chooses to use and create an AI policy that can evolve with the tech. Invest in how your team understands and then implements this tech to work across the range of daily tasks that happen internally and externally.
While chat-first search presents disruption, with preparation and the right focus on content and user experience, nonprofits can build productive new relationships and gain support through these emerging interaction types. Take steps now to adapt to this conversational future to generate new opportunities instead of scrambling to catch up. With strong expertise and helpfulness powering your conversational interfaces, nonprofits can achieve impact at scale as chat-first becomes the norm.
There are times for a wait-and-see approach, this is not one of those times. Whole Whale is actively working on planning for content scenarios and strategies that harness the tremendous potential of this tech for our clients. We happy to talk with anyone interested in starting a plan.