- Количество слайдов: 68
Web metrics A primer 1. Why measure? 2. Determining goals 3. What to measure? 4. How to measure? 5. Putting it all together (examples)
Web metrics - Caveats • Web metrics is a huge (and growing) field, with new strategies and businesses starting (and dying) every day. • This talk only covers a few, key, publicly available tools and services (Google Analytics, You. Tube, Facebook, Twitter) • Even for some public tools (e. g. Google Analytics), there is no way to cover all the options available • Many professional options available, but not covered or evaluated here • Counting on you to share info on additional tools, services, tips, etc.
Web metrics 1. Why measure? (determines what questions you want to ask) • Curiosity • Improve web-site visitor experience • Increase visibility of institution • Improve “reach” of news releases • Important to decide on goals before spending a lot of time, effort, and possibly money on web monitoring
Web metrics - Why measure? Benefits of measuring • Find out what’s working (and what’s NOT working) for web visitors • Identify and improve key “channels” for reaching media • Impress management / justify programs and staff efforts (cost/benefit analysis) • Help with institutional goals for outreach, marketing, fundraising – increasingly a part of PIO’s role, especially for small organizations.
Web metrics - Why measure? Cost-benefit analysis • Increasingly important with tighter budgets • Costs are relatively easy to predict compared with benefits • Web metrics are a great way of demonstrating benefits • Don’t forget to include cost/time for data analysis in web metrics
Web metrics 2. Defining goals • Simple • Easy to measure • Realistic • Relate to core mission of institution (Strategic plan, etc. ) • Relate to core mission of your department
Web metrics - Defining goals Some general types of web goals • Sharing information/inspiration • Defining or improving institution’s public image or “brand” • Attracting new audiences / increasing overall public awareness • Facilitating 2 -way communication • Inspiring action (e. g. ocean conservation; fund raising)
Web metrics - Defining goals Examples of specific goals • Make sure that all major wire services are aware of our news releases • Increase the size of our news release mailing list by 50% • Increase number of times a month that our research is mentioned in science blogs • Increase # of Twitter followers by 50% over next six months • Increase number of monthly hits on “Open House” web page by 20%
Web metrics 3. What to measure (what CAN be measured) • News coverage by organizations and by web users (blogs, tweets, etc. ) • Number of views of a story, video, or tweet • Number/type/demographics of users • How interested users are in your information • Paths that users follow to access your information • Whether users are going where you want them do and doing what you want • Trends over time
Web metrics - What to measure Deciding what to measure • There are so many different things you can measure on the web that it is most efficient to: – Do research (and periodic updates) on what metrics are available for the web “channels” you use. – Pick a limited set of metrics that are most useful for you (customize if appropriate) – Monitor those metrics consistently & regularly – Keep your own records – Periodically & regularly analyze the data and look for trends over time (prepare reports for superiors)
Web metrics - What to measure Measuring news coverage • Traditional approach: How many publications wrote articles on your release (and how big are the pubs? ) • How many blogs and news aggregator sites reprinted your release verbatim (or with minor variations)? • How many blogs mentioned or linked back to your release? (and how large is their readership? ) • Quality of coverage is subjective, but important because web is an “echo chamber” • How many times did it get tweeted/retweeted? • How many people “Liked” it on Facebook?
Web metrics - What to measure Measuring views of a page/story/video, etc. • Number, timing, and sources of page views are easy to measure for internal web sites using web monitoring tools • Tracking links to external site tougher, but tools available for some (e. g. You. Tube) • Highest page views often due to links from mainstream-media sites and blogs with large numbers of viewers (>5 -10, 000 viewers) • Gee-whiz factor is often key for big views; but most of these will be one-time viewers.
Web metrics - What to measure Key terms used in web-page monitoring (not standardized) • “Hit” – Occurs each time a FILE (any file) is supplied by the web server (only available with “server-log tracking”; more on this later). – More representative of total server traffic than popularity because many FILES may be downloaded as part of a single page (and caching issues). • “Page view” – Occurs each time a particular type of file (e. g. html) is supplied by the server (in “server logging”) or a particular page script runs (in “page tagging”).
Web metrics - What to measure Key terms used in web-page monitoring (continued) • “Visit” – Occurs when a single client downloads a series of page requests within a 30 -minute period. – A visit ends if no requests from a particular client come over a 30 minute period. • “Session” – Like a visit, but ends either after 30 minutes without accessing a local page OR if accesses a page from a different site. – “A session ends when someone goes to another site, or 30 minutes elapse between page views, whichever comes first. ”
Web metrics - What to measure Measuring number / demographics of users • Number of visitors over a specific time period • Number of regular visitors (e. g. You. Tube “subscribers, ” Facebook “likes, ” Twitter “followers, ” rss feed subscribers? ) • Age, sex (for registered users) • Geographic location (sometimes to zip code)
Web metrics - What to measure Key terms used in user monitoring (not standardized) • “Unique visitor” • A key term, based on identifying the computer (not the person) that is accessing a particular web site over a specified time period of record-keeping (typically a day, week, or month). • Determined using IP address in server log or cookie/Flash script; thus, a single person visiting from two different computers will count as two Unique Visitors • Note: If you add up the number unique visitors for each day in a month, they will not equal the total number of unique visitors for that month (because the same person visiting two days in a month is counted twice in the daily counts of unique visitors)
Web metrics - What to measure Key terms used in user monitoring (continued) • “New visitor” – A new visitor is a visitor that has not made any previous visits (over the entire period or record-keeping). • “Repeat visitor” – A repeat visitor that has made at least one previous visit to the site in a specific period of time. – Reliability limited by people whose browsers delete cookies each time they exit (they look new each time). – Note: The total number of unique visitors is not necessarily the same as the new plus repeat visitors because one person can be both new and repeat in a single day.
Web metrics - What to measure Importance of demographics • Who is following you is at least as important is how many. • Demographics available from services where users sign up (Facebook followers; registered You. Tube users) • Demographics of registered users does not necessarily represent demographics of all viewers (especially on You. Tube)
Web metrics - What to measure Measuring how interested users are in your information • Time on site/page/session • Repeat visitors/followers • User interactions (comments, shares, “likes, ” etc. ) • Retweets • Offsite links to content
Web metrics - What to measure Determining paths users take to reach your information • Can help assess useability of site • Find out what’s most (and least) popular on your site • For very busy sites, can be used in real time to balance traffic loads and prevent overwhelming servers. • Used in “funnel analysis” (more on this later)
Web metrics - What to measure Key term used in user-path monitoring • “Click path” – What pages a particular visitor follows during a particular session. – Related to “site overlay” view in Google Analytics showing web pages with number of clicks overlaid on each link (totals for a subset of visitors)
Web metrics - What to measure Determining whether users are going where you want them go and doing what you want them to do • Very much a sales/marketing approach • Definitely applicable to outreach and fundraising, and possibly to media work(? ? ) • Require extra staff time & expertise • “Event” analysis – How many users successfully downloaded the video from your last release? • “Funnel” analysis – What steps did each user have to take to find and download this video? (details later)
Web metrics - What to measure Measuring trends over time • Comparing different web metrics is like comparing apples, oranges, bananas, and cumquats (there is no standard) • Trends over time may be more reliable and accurate than absolute numbers
Web metrics 4. How to measure • News release exposure • Internal web site traffic • Facebook • You. Tube • Twitter
Web metrics - How to measure news release exposure • Traditional methods – clipping services (paper and online), Vocus multimedia monitoring, Eurekalert • On-line searches (e. g. Google News) • Advanced searches (unique words, blogs, Twitter searches, etc. ) • Can measure number of original articles and (increasingly) verbatim reprints of releases • Can sometimes estimate “reach” of “publisher” (e. g. number of blog readers) • Can use various on-line tools for calculating “buzz”
Web metrics - How to measure internal web traffic • Method 1: “Server log files” – Software running on your web server counts every page and file that is sent out to each IP address – Data are stored locally in a format available to you and your server administrators – Not tied to a specific vendor – Downside: Doesn’t count cached pages (pages sent once to user’s site, but stored and re-used) – Downside: May requires staff time, storage space – Downside: Useful for server admins, but less so for marketing/PIO types
Web metrics - How to measure Comparing server logging and page tagging • Example: MBARI web stats Jan-April 2011, based on Google Analytics and freeware program Web Log Expert: Web log Google Analytics 187, 452 Visitors 55, 123 Visits 303, 665 Page views 122, 934 Page views 1. 62 pages/visitor avg 2. 23 Pages/Visit But trends are nearly identical:
Web metrics - How to measure Google Analytics • The most widespread tool for web monitoring (Google claims that well over 50% of the largest web sites use Google analytics) • Easy to use at basic level (and free if you have a gmail account) • Very customizable for the advanced user • Becoming increasingly oriented toward marketing and sales vs simple tracking
Web metrics - How to measure Google Analytics – How to use it (very briefly) • (Open GA for MBARI’s web site) • Dashboard – Overview of “big picture” site metrics (customizable) • Intelligence – Set custom “events” for which you want to be notified (e. g. big rise or drop in traffic) • Visitors – Demographics, “loyalty, ” browsers used, etc. • Traffic Sources – Find out who’s linking to you • Content – Find out where people are going (and drill down to see individual pages) • Site search – Find out how people find you in searches (search terms, etc. )
Web metrics - How to measure Google Analytics – How to use it (continued) • Event tracking – Marketing/sales oriented options for the advanced user • Goals – Find out if people are doing things you want them to do (e. g. successfully completing a form or downloading a file) • Custom Reporting – Allows advanced users to graph/output combinations of stats listed above
Web metrics - How to measure Google Analytics – A few tips • Add annotations of events such as news releases. • Can track down sources of spikes to specific web sites (select a SINGLE day) • If you do see a source that drives traffic (and is reputable), try contacting them to get them on your email list, or as a Twitter follower • Customize the dashboard to show key stats you want to compare each time you log in. • Others from the audience?
Web metrics - How to measure Other free web tracking tools • There’s a bazillion of them… – Quantcast. com (standard page tagging) – Compete. com (ranking w/other sites) – Sharethis. com (counts people linking via a variety of social networks)
Web metrics - How to measure Metrics for Facebook • Use Facebook “Insights” pages to track: • Changes in the number of people who “Like” your site over time • Demographics (applies to Facebook members only; not all viewers)
Web metrics - How to measure Metrics for Facebook • Example of “Likes” tracking
Web metrics - How to measure Metrics for You. Tube • Can measure: – Number of views – Demographics (only covers You. Tube members who are logged in)
Web metrics - How to measure Metrics for Twitter • There a bazillion services out there, but I don’t have a specific one to recommend. • Does anyone have experience with them? (audience comment - bit. ly seems to be popular)
Web metrics 5. Putting it all together (ideas and examples) • MBARI experiences • Experiences from other active users of social media • Tips and techniques from marketing types • Caveats
Web metrics - Putting it all together MBARI news releases – effects on direct web traffic • Visitors to news release page spikes within one or two days of release, then tapers for a couple of weeks: • For example, Rappemonads release started at 165 unique visitors /day then tapered to 15 -30/day over next week; 5 -10/day after that. • Older 2010 releases get only 3 -7 unique visitors/day • Biggest hits are from releases featuring weird animal photos: For example, the barreleye release averages 125 unique visitors / day; these pages are very spikey depending on when random bloggers discover the page.
Web metrics - Putting it all together News release monitoring example: rappemonads (a new type of algae) • Release didn’t get much mainstream media coverage, but it did get wide pickup in news aggregation sites and blogs • Because the name of the algae had never appeared in the literature before, a simple Google search (not in News) turned up dozens of sites that used the release more or less verbatim • Based on slight differences in wording, I was able to track the “flow” of information from my email release, our web site (probably rss feed), and Eurekalert • Some blogs show where they got the text, as well as number of views, retweets, etc. • The amount of secondary coverage is impressive
Web metrics - Putting it all together News release monitoring example: tracking a key term
Web metrics - Putting it all together News release monitoring example: tracking a key term (continued)
Web metrics - Putting it all together MBARI Web-site monitoring example • Goals: – Increase general awareness of MBARI research – Provide detailed information about MBARI research for the general public and press. • Implementation: – New articles or photos posted about once every week or sometimes 2 weeks – One staff person spending 8+ hours/week
Web metrics - Putting it all together MBARI Web-site monitoring example • Use both server logging and GA (see previous comparison chart) • News site receives relatively low overall numbers of viewers (a few hundred a day) • Because of low overall visitor numbers to our news site, any bump of 50 -100 visitors/day can make a big impact on our overall traffic. • Bumps may be due to class assignments and outside links from large aggregator or news sites • Low retention because they just want to see a particular image or video on the site
Web metrics - Putting it all together MBARI Web-site monitoring example
Web metrics - Putting it all together MBARI evaluation of potential for Facebook exposure • There is a relationship between frequency of postings and number of “Likes. ” However, it may not be cause and effect, but covariance with other variables, such as general outreach effort. Institution Avg Updates/m onth "Likes" MBA 25 95, 600 WHOI 14 3300 SIO 10 2200 MLML 6 180 Harbor Branch 6 680 VIMS 5 566 Duke marine lab 5 528 Long Lab 4 600 20 877 MBARI (4/11)
Web metrics - Putting it all together MBARI Facebook monitoring example • Goals: – Increase general awareness of MBARI – Drive traffic to our main web site – Share info about general marine topics (not just MBARI) • Implementation: – Set up page Feb 8, 2011 – Posted about 20 -25 times a month so far – One staff person spending 1 -2 hour/week*
Web metrics - Putting it all together MBARI Facebook monitoring example • Tracking “Likes”: Early exponential increase now leveling off • Very event-driven increase (e. g. push from Aquarium) • Suggests we need to make more effort to get more “Likes” if we want to get more “likes. ”
Web metrics - Putting it all together MBARI Facebook monitoring example • Results from first four months: • Exponential growth curve of “Likes” has flattened out already (mostly driven by Aquarium publicity) • But Number of impressions per posting is still increasing (was 150 -250 at end of 1 st month; is now 2, 000 -2, 500) • This is a lot of exposure compared with number of visitors to news articles on web site. • We seem to have over 1, 800 regular visitors who will click on any new item (even ”dry” journal articles) posted on our Facebook page
Web metrics - Putting it all together MBARI Facebook monitoring example • Results of MBA link Suggests possibilities for good synergy between MBARI and Aquarium social networking (we provide content, they provide “eyeballs”) • Our followers are relatively engaged, and comment pretty frequently. (0. 1 to 0. 3 feedback rate) • Very different demographics from You. Tube: – 2/3 of our facebook followers are female! (of these most are 25 -54, w/peak at 25 -34) – Of the males, same age range, but peak at 35 -44) – So far, mostly people in California (plus 5% non-US)
Web metrics - Putting it all together MBARI Facebook monitoring example
Web metrics - Putting it all together MBARI Facebook monitoring example: Using GA to gauage effect on web visitors
Web metrics - Putting it all together MBARI Facebook monitoring example: Effect on web visitors • Overall, the volume of traffic to main web site has NOT changed significantly since last year (Jan-Feb 2010 vs Jan-Feb 2011), despite initiation of Facebook and Twitter programs. • We saw a 30% increase in NEWS web-site traffic, but only due to short bumps due to blog listings – not repeat or sustained visitors (see GA graph) • We did see a tripling of the number of people who were referred to our web site from Facebook, but still only accounts for an average of 10 visitors/day
Web metrics - Putting it all together MBARI You. Tube channel • Goals (not very measurable): – Share cool videos with media and public – Increase general awareness of MBARI research – Share info about general marine topics (not just MBARI) – Drive traffic to our main web site (to find out more) • Implementation: – Set up account 3 years ago – Updated about 1 to 2 times a month – One staff person spending 10 -20 hours a month
Web metrics - Putting it all together MBARI You. Tube monitoring • One BIG hit: Barreleye video by far the most watched – 3. 6 million views (2 years after first posting, this one video still accounts for 70% of our You. Tube channel traffic) • Over the past 2 months, our You. Tube site saw an average of about 1, 250 unique views/day (1, 521 unique views/day over past 12 months) (this is twice as many views as any section (not just page) of our web site) • Since 2009, we have acquired about 1, 900 subscribers (who seem to check out almost any new video we post). • We are steadily increasing our subscriber base at an average net rate of 1 to 2 a day (new subscribers minus those dropping subscriptions).
Web metrics - Putting it all together MBARI You. Tube demographics • Last year, almost ¾ of our You. Tube viewers were males (mostly 35 -54 yr) Very few were under 18 or over 55. • Only about one quarter of viewers were females, but they were spread throughout the age ranges (there were more males than females in 13 -17 yr group). • In the last six months, the proportion of female viewers has increased to 35%, in response to Oct 2010 Halloween video and Jan 2011 Valentine videos.
Web metrics - Putting it all together MBARI You. Tube engagement and effect on web-site traffic • Our weird animal videos get very high levels of responses from viewers, at rates 2 to 10 times higher than “normal” (1 to 2 per 100 views vs 0. 1 to 0. 3). • Most of the comments are on the order of “What a weird #$%^ fish!” • We rarely see You. Tube hit videos correlating with bumps to our web site • The exception are spikes in web-site visits following posting of high-visibility news-release videos (spikes may come from general release publicity)
Web metrics - Putting it all together Web monitoring example: Exploratorium and Google Analytics • Provides Google Analytics info to NSF in support of NSFsponsored web projects (Ice stories). • NSF also funded an outside consultant to survey users and convene focus groups to review the effectiveness of the website. • Uses Google Adwords to advertise program for free (nonprofits may apply through Google Grants) • Likes using GA overlay to see where people are going. • Suggests monitoring what words people are searching for on the site; then create content about those terms if you don’t already have it or make it easier to find these terms.
Web metrics - Putting it all together Web monitoring example: KQED Quest (science program) • TWITTER: Uses Hootsuite—a free tool to track Twitter analytics; also has a built-in URL shortener. Dashboard shows who is following QUEST and who QUEST is following. (QUEST has been on Twitter since April 2009; currently has 3, 600 followers, does 15 -30 tweets a day, and gets up to 100 clicks per tweet. ) • FACEBOOK: QUEST launched their FB site in Jan. 2010; currently has 2, 400 fans. Used ad campaigns within FB to gain fans at a cost of about $. 57 per fan. • FB “Insights” metrics shows demographics. It looks like women between 35 -45 years old in Pleasanton are the biggest group following QUEST. (members only; like MBARI – mostly women) • Consider what a FB follower is worth to you. According to Paul Rogers, “The fact that people have an interest in you has an intrinsic value, even if you don’t know what that is yet. ”
Web metrics - Putting it all together Example: Using demographic information • Demographic information is a blunt tool if you’re trying to reach science writers, reporters, NGOs, other PIOs. • For groups such as this, a targeted approach to cultivating an audience may be better: • It pays to go through your follower (or email) lists every now and then, and look for key individuals. • Conversely, identify key individuals and invite them to become followers (or get on your email list)
Web metrics - Putting it all together Where demographics is useful in targeting (and not) • Members of media (target) • Grade-school students (demographics) • College students (demographics, demographics +. edu domain tracking) • Decision makers / resource managers (target) • Local community members (demographics) • Other researchers and institutions (target) • NGOs and activists (target) • Businesses / industry (target)
Web metrics - Putting it all together Examples of programs generating measurable user interaction • “Ask a scientist” column • Allow users to post comments, photos, sightings, personal experiences, etc. on moderated discussion board • Conduct on-line surveys (more for interaction than statistically useful data) • Hold on-line contests (e. g. name a fish)
Web metrics - Putting it all together Example of goal-driven user tracking: “Funnel Analysis” • Use user tracking to determine the steps that people take in the process of doing something you want them to do (to buy your product or click on your important web page). • Try to figure out the percentage of people that are willing to move through each step (the “conversion rate”)
Web metrics - Putting it all together Example of steps in“Funnel Analysis” • 1) Person Googles “dolphins” • 2) Person clicks on the Google search result showing your photo of a smiling dolphin, and lands on your “Dolphins are our friends” blog entry. • 3) Person clicks through to article on your web site on dolphin research. • 4) Person clicks on prominent button saying “Save our dolphins” • 5) Person signs up to receive email alerts (and funding pleas) about saving dolphin research • 6) After several emails, person writes check to your institution to help fund dolphin research.
Web metrics - Putting it all together Following up on a “Funnel Analysis” • Determine where the bottlenecks are (no pun intended). (Maybe your article on dolphin research is dry and boring, or the funding button is tiny and hidden at the bottom of the page. ) • Try to enhance that process at bottlenecks and maximize the conversion rate at each step.
Web metrics Thank you for listening! • More questions? • Discussion?