Loading...

4 Companies Show you their Key Lessons Learned in Building a Knowledge Base

It turns out that the number one way in which customers like to fix a problem
is by fixing it themselves. Live chat support is the next fall back with phone
support rapidly becoming an emergency escalation channel. If that's the case
then our knowledge bases and other self-support channels need to be our no.1
focus. Below we'll cover self-help channels, key lessons learned by companies
building knowledge bases and what the future might hold.

Skip straight to the lessons learned building knowledge
bases.

Why customers prefer self service


There are plenty of statistics covering why self-service is the go to channel
for many customers.

“Customers prefer knowledge bases over all other self-service
channels” Hubspot

“Instead of calling customer representatives, the digitally savvy generation
prefers to do things themselves”
RetailMinded

“According to research, 69% of millennials claim to feel good about
themselves
and the brand when
they can solve a problem on their own.”

We run 2 companies with a total of around 300 sign ups a day between them.

When we added explanatory videos to one of them the number of live chat
queries dropped dramatically. People don't want to message us and ask us
questions, they want to use the software and solve the problem.

Self-service is faster than waiting for someone to reply and there may well be
a feeling of taking up someone's time when you can find the answer yourself to
what might be a simple query.

What types of self service are there?


FAQs

This is the simplest example of self-help. Companies add a list of frequently
asked questions under their pricing, features, and linked directly on their
home page. They are useful and act equally as sales and marketing copy to
assure customers as well as solve problems.

Knowledge bases

This is a common and rapidly growing form of self help. It can include the
trouble shooting instructions, how-to guides, FAQs, overviews and more. It's a
library of documents the customer can visit to help themselves.

At Upscope we use Intercom Articles as our knowledge base. It's integrated
directly into our live chat so that we can send a knowledge base article to a
customer with an issue. They can search for that article themselves.

The most interesting current evolution within knowledge bases is how they work
with bots.

Bots

Self-help bots come in many forms but one of them is within automated live
chat.

The customer asks a question and the bot finds the appropriate answer by
surfacing an article from the knowledge base.

Bots are everywhere and can independently manage a query or help filter
questions in order to figure out whether a human agent should take over. They
can handle the simple queries and then route the others.

What's even more interesting is how bots are starting to integrate with
on-page onboarding and guidance systems to show the customer how to do
something.

On-page guidance and answers

A great example of on-page guidance working with bots is Intercom's onboarding
system integrated with its help bots.

Intercom developed one of those typical guidance systems you see when you sign
up to a new website. You know, the tool tip that appears over a button saying
"click here" and then it puts some explanatory text on the page. Then it
highlights another button saying "now click here".

Bit by bit you learn the software. This can also be used to solve problems as
most of them are a combination of navigating to a page and having something
explained to you. You ask the bot a question and it kicks off an automated
guidance flow.

Other forms of on-page help are knowledge bases built into the page so you
don't have to go to a separate page to get your answer.

AnswerDash does this. You click on a Q&A tab directly on the page and see
a search box and list of popular questions people ask on that page.

Guides, blogs, sales docs and self-educational materials

This is a slightly different form of self-help that companies provide to
customers or potential customers and you might not classify it in the same
area as a knowledge base but the overlap is immense.

The highest number of queries from customers, of a web application for
example, are often in the early days of use. Before they buy they want to
understand the features and how they're different to competitors.
They want
to understand the pricing, the team management options, the possible future
cost and lots more.

You could ask them to book a demo with your sales team and some people prefer
that. We've noticed that those buying enterprise plans will be very familiar
with getting a demo from our sales team, negotiating pricing and more. In
fact, they have systems and structures around this including procurement leads
who specialise in negotiating deals.

However, in evaluating a product a company, or even an end consumer who is
buying a product like a Roomba for the home, does a lot of research on the
product. A seller can help these buyers by creating a guide giving them an
overview of the industry, the product, how to compare products and make
decisions.

Guides can come in the form of blog posts, white papers, comparison sheets and
more. They also set expectations which, if not met, result in more support
queries and higher customer churn later.

A lot of evolution still to come in the format of knowledge delivery

Besides bots, knowledge bases, FAQs and their variants the other forms of self
help might be 3rd party websites you Google for because you don't like the
answer you got on the company's own website.

Right now the knowledge base and bot market seems to be expanding rapidly and
I expect the integration into onboarding flows and more will evolve to our
collective benefit.

Most companies don't appear to have this integration of bot + base + flow
running and I suspect that it's due to the lack of knowledge on options but
also due to the effort required in planning, implementing and managing that
combination of writing articles or short answers, educating bots, integrating
onboarding flows and then updating it all when things change.

What sort of problems do people need help with?


For perspective I'll start with general Google queries and then drill down
into individual apps we all know.

Most common non-app related questions on Google

As you can imagine the most popular questions are very human. They're often
looking for very specific information.

3 Most common 'How' Questions people Google for

Just for the hell of it and because I'm confused about number 2, here are the
3 most popular "how" questions on Google.

  1. How to draw 1,500,000

  2. How to make slime 1,500,000

  3. How to screenshot on mac 1,500,000

I really can't say why "how to make slime" is so popular.

Popular app related questions

How to delete instagram account 165k

How to use snapchat 165k

How much is netflix 33k

How to delete a facebook account 33k

Questions people enter into Google about Youtube each month in the USA

Here's where things get interesting.

The general queries above give perspective on overall global numbers but now
we'll see queries from one country on one specific app.

Below I've used the service SEMRush to look at Googled
queries in the USA.

SEMRush is a great tool for keyword research is fairly accurate in somehow
figuring out queries people search for.

This is for the USA only and specifically questions asked about Youtube each
month.

How to make a playlist on youtube 9.9k (per month)

How to download youtube videos on iPhone 49.5k

How to make money on youtube 40k

How to get paid on youtube 12.1k

How to change your profile picture on youtube is 6.6k

How many views on youtube to make money 5.4k

Look at how many people want to download youtube vids to their iPhones just in
the USA. Almost 50,000 per month are Googling that. To be fair, so have I.

What's crazy is that so many people search for simple queries like how to
change your Youtube pic. It should be under settings but it's interesting that
it might be faster to Google for it than figure it out. I found this with
changing my signature in Gmail. I forget where it is so I Google for it rather
than searching through the settings.

Questions people Google about Google Sheets and Docs each month in the USA

How to subtract in google sheets 9.9k

Hot to freeze a row in google sheets 2.9k

How to change margins in google docs 9.9k

How to draw in google docs 6.6k

Questions people Google about Quickbooks each month in the USA

How to use quickbooks 4.4k

How much is quickbooks 1.9k

How to reconcile in quickbooks 1.3k

What sort of questions do people search for on knowledge bases?

Below are some of the most commonly searched for questions on knowledge bases.

How do I change my email address?

How do I reset my password

I can't remember which email address I used to sign up

I want to change the account owner

How do I cancel my subscription?

What's my billing date?

How do upgrade / downgrade my subscription?

Send me my latest invoice

Can you refund me, I've been charged incorrectly

Based on my experience of customer support requests I've seen all of these at
one point or another until we really started to refine that knowledge base,
the onboarding, the UI / UX, the FAQ and more.

Even with those changes we still see the "How do I cancel my subscription
question" or "Can I upgrade / downgrade".

Again, these are partly UX or onboarding issues but even seemingly flawless
interfaces will find customers who struggle. The best evidence of that is when
I realised a customer did not know what "Domain" meant and we had been using
it everywhere. Who knows how many customers we lost because of that.

Key lessons from 4 companies on how to build and maintain a knowledge base


Upscope found that you have to let it evolve naturally at first

  • It has to evolve and can't be too prescriptive in the early days. It needs
    to be built around customer questions and not what you think they might need
    to know. Naturally you'll build a few key articles to simply fill up the empty
    space but beyond that you can save time by building pages on the fly rather
    than scrapping overly long articles you thought they might need.

  • Keeping it updated either requires someone to be responsible for it or the
    only time you update is when a customer needs it and you realise it needs
    updating. Or, worst case, a customer uses the outdated article and wastes
    their own time and then has to tell you it's out of date.

  • It's hard to reconcile whether something should be a short saved live chat
    answer or a full knowledge base article. Having a good document structure with
    a clear action based headline and sub-header helps. This is again something
    that evolves and is hard to plan for but needs to be reviewed.

  • Sticking to language, terminology, paragraph spacing and more is tough
    considering how many articles need to be written by experts in that subject
    area. They may only write a new article once every couple of months so you
    need a design and terminology guide.

  • Use Screenshots and pictures as often as possible. We're now redesigning the
    site and working on adding images to every help article. Blocks of text are
    daunting and it's hard to go back and forth between a help article and the
    main site to follow a set of steps. Pictures help.

Saaslist tells you to be careful with jargon but not remove it entirely

Saaslist's key points reflect some of our own experiences at Upscope. I like
their balanced point of view on jargon because it's a bit of a dilemma when
you're figuring out a common dictionary to use and when to use abbreviations.

  • Avoid the wall of text. Use graphics.

  • Review and update it. Do regular audits to find which pages are visited most
    and least frequently and why.

  • Standardize and be consistent.

  • Avoid industry jargon or at least don't go heavy on it. Customers all of
    different skill levels use it. If you use jargon then link to a glossary.
    Some pages need jargon as they are for people who understand that technical
    language.

See more on Saaslist's points
here.

Uberflip suggests codifying customer personas and asking responsibility

questions

The second point in the list below tickled by brain because at Upscope, the
enterprise customer who uses the Upscope API, has an entirely different flow
to how they sign up, test and then trial the product compared to other
customers.

We've added docs to our knowledge base that cover most use cases but are those
docs organised for an enterprise customer journey? Not yet. We need to get on
it.

  • Uberflip first decided who would be responsible for creating, editing, and
    updating the content, then how often it would be updated and reviewed, and
    what format they would use.

  • Codify your customer personas. "A prospect going through a sales process
    and an existing customer have very different needs, problems, and
    expectations. To best meet those needs with your knowledge base, you need to
    understand how they engage with your platform and how their needs differ
    based on their use cases
    , plans or account types, and more."

  • Do an audit of your content. "A content audit helped us discover some pretty
    glaring holes in our existing knowledge base content"

  • Ask responsibility questions. Who will be responsible for creating the new
    articles
    ? Who is responsible for editing and verification? Who is
    responsible for making updates as new features are released or existing
    features are updated? Will those updates be made daily, weekly, monthly, or
    quarterly
    ? What format will your content take?

  • Structure fluidly. "At Uberflip, we create Marketing
    Streams

    of content for specific types of users, specific plans, and sometimes even
    specific users themselves."

  • Capture insights. In short, if your analysis says a lot of people are
    discussing one specific article then maybe check that article is doing its
    job. Maybe the video on the article needs more work.

  • Integrate it. "For example, here at Uberflip our knowledge base is
    integrated with HubSpot. This gives our marketing team the ability to actively
    use knowledge base content consumption in our lead nurturing process (changing
    lead scores or sending automated emails) and allows our customer success team
    to have user-level insight into potential areas of opportunity or concern"

  • Add a human touch. Uberflip adds team instagram photos to their marketing
    streams. At Upscope we've added some popular memes to some posts and yes, it
    makes a difference. Articles can be so dull in appearance that a bright meme
    lifts the mood.

See Uberflip's full breakdown
here.

Groovehq describes how to structure individual articles

Below are some great points on how to structure each article and I also like
their point on not spending too much time getting a perfect screenshot dressed
up to look nice. Interesting that they say you should add videos but don't
expect people to watch them. I personally prefer screenshots but a video works
if I'm not familiar with a product and I'm learning about it.

  • Structure each article so that it minimally contains: **1. Problem or topic,
  1. Step by step process, 3. Result**: What should happen after the steps have
    been completed by the customer? 4. Related resources: Similar articles,
    questions, or topics.
  • Avoid distracting them while they're reading the article.

  • The less text, the better.

  • Short paragraphs.

  • Bullet points and lists.

  • Break down details into separate articles rather than cramming them all into
    one.

  • Screenshots. "Between you and me, don’t spend too much time beautifying
    these images. Most people come to your knowledge base for quick, clear
    answers. Pictures provide that even if they’re poor quality."

  • Add videos but assume no one will watch them.

  • Organize based on the customer experience. Knowledge base categories based
    on customer experience or the customer journey.

  • Mine failed searches for improvements.

  • Brainstorm related articles using different customer journeys.

See more on Groovehq's advice
here.

Customer Experience Insight says it's an asset and not an IT project

The following points are from Customer Experience Insight.

  • There's too much content and not enough management. No-one is responsible.

  • Many organizations approach self-service as an IT project. Because they’re
    “projects” and not a “business asset,” they become outdated and unattended
    once they’re implemented.

  • Some companies bury their other channels (like live chat or phone) and push
    people to self-service.

I think we've all seen this last one before. It's fairly frustrating and you
just know they've done their analysis and saved themselves some extra staff by
pushing you towards self-service but when self-service is failing then you
just get frustrated hunting for the other channels.

It would be a faster feedback loop to let the customers find and use the live
chat channel so the team knows the common queries the self-service channel
isn't covering. Well, that's how I think of it. Tough call to make in the
quarterly budget meeting asking to increase staff because of a UX initiative
you pushed for.

See the full customer experience insight article here.

How much time and money are enterprises spending on these knowledge base

implementations?

While our costs at Upscope are relatively low the Intercom system in its
entirety still costs around $12,000 per year and we would be considered and
SMB. Costs in implementation, updates and maintenance are many times that if
you consider per hour staffing costs.

For larger enterprises it's just a whole different level.

"The global research, based on responses from more than 1000 organisations,
found 83 per cent of KM platforms purchased to enhance customer experience
(CX) take longer than six months to implement with 55 per cent taking more
than a year. On average, survey respondents spend $US676,000 on their
projects." Source.

What's likely to happen next in knowledge management


Opportunities for land and expand ground up entrants

Reading the above on implementation of knowledge bases, I can just imagine the
directive coming down from a senior manager to begin creation of knowledge
base followed by a team figuring out criteria, sourcing products, planning and
then implementing it.

The dream, in some ways, would be an organic ground up growth of a knowledge
base that just works and has accountability built in. Many companies now sell
into enterprises ground up by landing and expanding and maybe there's space
for a ground up knowledge base as well.

The long implementation cycle and planning creates friction to deploying a
knowledge base. To avoid this friction, new entrants in the market could
provide services to individuals and small teams which they can get started
with straight away and solve immediate problems.

Templates and formality vs personality and a human touch

While getting terminology correct is important, the format of pages and the
way in which information is displayed might best be owned by customer facing
teams and not by disinterested parties who write it up and then play no
further part.

What if creative people built knowledge base articles but they owned it and
put their name to it and managed it themselves? If you want people to be
responsible for it then what if they own it from start to finish. The customer
will see a real person behind that article showing care for their problem.

Pictures, pictures, pictures

Again and again while researching knowledge bases, and given our own
experiences managing some, we know that screenshots and other pictures are
important. Blocks of text are daunting and show a lack of care for the
customer's immediate needs.

Expect more ways to see screen shots and videos added or see more ways where
the information is actually moved to the page, like AnswerDash. When a person
has a problem on your settings page, why send them to a totally different
website to search for an answer when you can show them there and then on the
page.

Bots and flows

Without a doubt the bots are getting smarter. GPT-3 built by Elon Musk's AI
company is extraordinary. It's writing essays that feel human. It's writing
code!

Combine bots with onboarding software and what do you have?

Imagine asking a question like "How do I change my picture on youtube?" and
the bot takes over.

  1. It highlights the settings button on Youtube.

  2. You click that.

  3. On the next page it highlights the Account section.

  4. You click that.

  5. Then it tells you to change your picture by scrolling to the bottom of the
    page and clicking upload.

This is all happening already but just not on a great scale yet. Intercom have
this feature. We have not yet implemented it largely because it does take time
to set up and get going with. Again, there is friction and if you can figure
out how to remove that then you're doing well.

Sunglasses emoji. Continue reading the blog