36
OPTIMIZE YOUR ON-SITE SEARCH Help your customers find exactly what they’re looking for. JP Sherman Red Hat’s Manager of Search & Findability Tweets @jpsherman

Pubcon 2016 Measuring & Improving On-Site Search

Embed Size (px)

Citation preview

Page 1: Pubcon 2016 Measuring & Improving On-Site Search

OPTIMIZE YOUR ON-SITE SEARCH

Help your customers find exactly what they’re looking for.

JP ShermanRed Hat’s Manager of Search & Findability

Tweets @jpsherman

Page 2: Pubcon 2016 Measuring & Improving On-Site Search

Findability is how your customers find what they’re looking for.

UNDERSTANDING THE CORE ASPECTS OF SEARCH AND FINDABILITY

➔ The Four Pillars of On-Site Search Optimization◆ Measurement◆ SERP Structure◆ SERP UI/UX◆ Result Set Quality

Page 3: Pubcon 2016 Measuring & Improving On-Site Search

BUT FIRST, A SHOW OF HANDS

SEO/ Paid Search

How many people regularly use search data to make decisions?

Onsite Search

Page 4: Pubcon 2016 Measuring & Improving On-Site Search

STUDIES SHOWHow many companies regularly use search data to make decisions?

Between 15% and 20% of large companies dedicate resources to on-site search. Smaller businesses range

from 0% to 4%

Six-month-old infants can only hold one thing in memory.

Having more narcissists on a team is better for generating creative outcomes.

The more relaxed you are when you enter a store, the more money you’ll spend.

The ruder someone acts, the more convinced observers become that he or she is powerful.

Shorter sleep duration is associated with increased susceptibility to the common cold.

King penguins are attracted to the colors on each other’s beaks, including colors humans can’t see.

Page 5: Pubcon 2016 Measuring & Improving On-Site Search

ABOUT YOUR CUSTOMERSEngaged. Invested. Curious. In an Active Decision Phase. Trusting. Convinced.

WHO ARE THEY?WHAT ARE THEY MORE LIKELY TO DO?

Page 6: Pubcon 2016 Measuring & Improving On-Site Search

AND YET CUSTOMERS WHO SEARCH ARE...Engaged. Invested. Curious. In an Active Decision Phase. Trusting. Convinced.

❏ They are 5x - 6x more likely to convert❏ They are more likely to share content

❏ They are more likely to stay on-site longer❏ They are more likely to return

❏ They are more likely to recognize brand

Page 7: Pubcon 2016 Measuring & Improving On-Site Search

DIFFERENTIATING BETWEEN DATA & STORYThe measurements are simple to understand, but it’s the nuance that tells the story

Data is the characters in the story

What these characters (data) do IS the story.

Page 8: Pubcon 2016 Measuring & Improving On-Site Search

UNDERSTANDING INTENT & BEHAVIORThe measurements are simple to understand, but it’s the nuance that tells the story

Let’s Start Where We Always Start

Keywords

Page 9: Pubcon 2016 Measuring & Improving On-Site Search

UNDERSTANDING INTENT & BEHAVIORThe measurements are simple to understand, but it’s the nuance that tells the story

Keywords High Freq/ Low Intent

Low Freq/ High Intent Long Tail Unicorns

Popular but vague keywords

Rare but specific keywords

Specific terms > 3 or 4 words

Queries that happen once in a given time-frame

These are the most common types of queries that happen in on-site search. By understanding the types of queries you have, you can start tailoring the search experience to the type of query.

Page 10: Pubcon 2016 Measuring & Improving On-Site Search

UNDERSTANDING INTENT & BEHAVIORThe measurements are simple to understand, but it’s the nuance that tells the story

Keywords

High Frequency/ Low Intent

Low Frequency/ High Intent

Long Tail

Unicorns

Clicks

Conversions

No Clicks

Position

CTR

Conversion Rate

Consumption Rate

Keyword Types Metrics Calculated Metrics

Negative CTR

Page 11: Pubcon 2016 Measuring & Improving On-Site Search

UNDERSTANDING INTENT & BEHAVIORThe measurements are simple to understand, but it’s the nuance that tells the story

Keywords High Freq/ Low Intent

Example Queries:● Bikes● Software● Smart phones● College● Finland● Books● Laptops● Elections

This is the top of the funnel search behavior.

User Intent Isn’t Specific● Information● Ideas● Knowledge● Type Selection● Familiarity● Definitions● Topic Review

Page 12: Pubcon 2016 Measuring & Improving On-Site Search

UNDERSTANDING INTENT & BEHAVIORThe measurements are simple to understand, but it’s the nuance that tells the story

Keywords Low Freq/ High Intent

Example Queries:● Download latest iTunes● Stream The Clash● Pictures of cats in baskets● Vietnamese food near me● Spicy vegan taco recipes● Supernatural monsters that

drink human blood● Latin name for ring-tailed

lemur● What does coulrophobia

mean?

There is no ambiguity in the searchers’ intent.

User Intent Is Specific● Take an action● View specific thing● Learn specific thing● List based content● Answers to a

specific question

Page 13: Pubcon 2016 Measuring & Improving On-Site Search

UNDERSTANDING INTENT & BEHAVIORThe measurements are simple to understand, but it’s the nuance that tells the story

Keywords Long Tail Long tail keywords are those three and four keyword phrases which are very, very specific to whatever you are selling.

Page 14: Pubcon 2016 Measuring & Improving On-Site Search

UNDERSTANDING INTENT & BEHAVIORThe measurements are simple to understand, but it’s the nuance that tells the story

Keywords Unicorns Unicorns are keywords that appear once in a given time-frame.

● On average, 50% to 60% of queries are unicorns

● Unicorn queries are useful to gain quick wins. For queries that make sense, the identification of a unicorn can lead to content that will immediately fulfill that search.

● One caveat, unicorn terms can frequently be typos, nonsense words or long copied/ pasted text

Page 15: Pubcon 2016 Measuring & Improving On-Site Search

UNDERSTANDING INTENT & BEHAVIORThe measurements are simple to understand, but it’s the nuance that tells the story

Metrics CTR CTR in on-site search measures clicks from SERP

● CTR’s are a great way to measure the overall success of your on-site search

● Sometimes, you might see a CTR on a keyword greater than 100%

● This means that the user has opened up links in new tabs.

Page 16: Pubcon 2016 Measuring & Improving On-Site Search

UNDERSTANDING INTENT & BEHAVIORThe measurements are simple to understand, but it’s the nuance that tells the story

Metrics Conversion Rate Consumption Ratevs

Conversions

A quantifiable action taken. Conversions can be purchases, sign-ups, downloads or something else. Always binary.

Consumptions

A qualitative measurement of content. It’s is a spectrum that measures how much of the content has been consumed.

Page 17: Pubcon 2016 Measuring & Improving On-Site Search

UNDERSTANDING INTENT & BEHAVIORThe measurements are simple to understand, but it’s the nuance that tells the story

Metrics No Results No Clicksvs

No Results

When a query is performed and nothing matches the query

No Clicks

The inverse of click throughs

Page 18: Pubcon 2016 Measuring & Improving On-Site Search

WHEN SERPS FAILSearches will fail, your customers will look for strange things - help them out a bit.

Users will experience a search failure.

Giving users a way to refine, tips to search again, a place for feedback will improve

the search experience.

Users w/out assistance re-searched 33%Users w/ assistance re-searched 59%

Page 19: Pubcon 2016 Measuring & Improving On-Site Search

WHEN SERPS FAILSearches will fail, your customers will look for strange things - help them out a bit.

Metrics

No Clicks

Failure means you can ask for

more information

!

Page 20: Pubcon 2016 Measuring & Improving On-Site Search

WHEN SERPS FAILSearches will fail, your customers will look for strange things - help them out a bit.

Metrics

No Clicks

Page 21: Pubcon 2016 Measuring & Improving On-Site Search

WHEN SERPS FAILSearches will fail, your customers will look for strange things - help them out a bit.

Metrics

No Clicks

Wait, What?

Page 22: Pubcon 2016 Measuring & Improving On-Site Search

AUTOSUGGEST CAN HELP FILTER INTENTThe SERP & the Snippet Can be Magnetic to Clicks When Showing Value

Starting with the Search Experience

Page 23: Pubcon 2016 Measuring & Improving On-Site Search

EVEN WHEN THE USER DOESN’T KNOW “CORRECT” PHRASEThe SERP & the Snippet Can be Magnetic to Clicks When Showing Value

Auto-Suggest

Customer Focused

Immediately gives the user options helping fix any “wonky” searches

Business Focused

You can add keymatches or suggestions based off of the query entered

Page 24: Pubcon 2016 Measuring & Improving On-Site Search

AUTOSUGGEST CAN HELP FILTER INTENTThe SERP & the Snippet Can be Magnetic to Clicks When Showing Value

Autosuggest vs. Non Autosuggest

Having autosuggest can increase CTR &

Conversion

CTR: 42%CR: 3.9%

Page 25: Pubcon 2016 Measuring & Improving On-Site Search

AUTOSUGGEST CAN HELP FILTER INTENTThe SERP & the Snippet Can be Magnetic to Clicks When Showing Value

Autosuggest vs. Non Autosuggest

Having autosuggest can increase CTR &

Conversion

CTR: 57%CR: 5.1%

Page 26: Pubcon 2016 Measuring & Improving On-Site Search

CONTEXTUALIZE INFORMATION IN THE UIIf the search results aren’t perfect, adding contextual information in the UI can help the user

Page 27: Pubcon 2016 Measuring & Improving On-Site Search

KNOWLEDGE GRAPHS CAN CONTEXTUALIZEThe SERP & the Snippet Can be Magnetic to Clicks When Showing Value

Knowledge Graphs

High Volume/ Low Intent

Allows the user to immediately focus on what their intent is when it’s not clear in the query.

Page 28: Pubcon 2016 Measuring & Improving On-Site Search

OH… RIGHT… MOBILE MATTERSThe SERP & the Snippet Can be Magnetic to Clicks When Showing Value

Don’t Forget Mobile

Page 29: Pubcon 2016 Measuring & Improving On-Site Search

DISAMBIGUATION FOR SIMILAR SEARCHESUnderstand Where Your Customers Primarily Interact With You.

Page 30: Pubcon 2016 Measuring & Improving On-Site Search

RIGHT... AGAIN WITH THE MOBILE.Understand Where Your Customers Primarily Interact With You.

Page 31: Pubcon 2016 Measuring & Improving On-Site Search

TEN BLUE LINKS MAY NOT EVEN WORKSome products lend themselves to be more visual, or more comparative.

Page 32: Pubcon 2016 Measuring & Improving On-Site Search

BLENDED SEARCH RESULTSJust display all the things

Page 33: Pubcon 2016 Measuring & Improving On-Site Search

OF COURSE, I’M GOING TO TELL YOU TO TEST ITNaturally… but what do I look at to see if it’s a good test?

CTR

Conversion Rate

Accessed Content

Watch These Metrics

Refinements

Does the experimental group have a higher CTR? Implying Value

Are there more purchases or other conversion events?

Does more content get viewed after a search is performed?

Do people perform less searches after an initial search?

Mobile Is there a significant difference in these metrics from mobile devices?

Page 34: Pubcon 2016 Measuring & Improving On-Site Search

IF YOU REALLY WANT TO BE FANCYIt’s a great way to measure search quality, but requires decent development changes.

To deliver the right content quickly to the right person who is searching your site.

Goal

Measure: Time from Query to Conversion/ Consumption

1st Query

1st Click

Pogostick 1st Refinement

2nd Click

conversion? 2nd Refinement

Page 35: Pubcon 2016 Measuring & Improving On-Site Search

KEY TAKEAWAYS TO IMPROVE ONSITE SEARCHMake a plan, make a change every month, measure it & repeat.

It’s not that hard

Change the Snippet

Change the SERP design

Add Contextual Information

Measure, Measure, Measure

Don’t forget, mobile is unique

Page 36: Pubcon 2016 Measuring & Improving On-Site Search

THANK YOU

Please feel free to talk to me if you happen to run into me.

Find me on Twitter @jpsherman