diff --git a/app.py b/app.py
old mode 100644
new mode 100755
diff --git a/schema.sql b/schema.sql
index 99b5d0d..8edb002 100644
--- a/schema.sql
+++ b/schema.sql
@@ -3,7 +3,7 @@ CREATE DATABASE `sXXXXX__tif`;
 DROP TABLE IF EXISTS `cache`;
 CREATE TABLE `cache` (
   `cache_id` INT(10) UNSIGNED NOT NULL,
-  `cache_views` INT UNSIGNED NOT NULL,
-  `cache_time` TIMESTAMP NOT NULL,
+  `cache_views` FLOAT NOT NULL,
+  `cache_time` TIMESTAMP NOT NULL DEFAULT '0000-00-00 00:00:00',
   PRIMARY KEY (`cache_id`)
 ) ENGINE=InnoDB;
diff --git a/templates/index.mako b/templates/index.mako
index b8bb765..b5043bf 100644
--- a/templates/index.mako
+++ b/templates/index.mako
@@ -24,7 +24,7 @@
         <table>
             <tr>
                 <td>TIF</td>
-                <td>${result["tif"]}</td>
+                <td>${"{0:.2f}".format(result["tif"])}</td>
             </tr>
             <tr>
                 <td>Transclusions</td>
diff --git a/tif/calc.py b/tif/calc.py
index cc47dfb..66596d6 100644
--- a/tif/calc.py
+++ b/tif/calc.py
@@ -1,7 +1,12 @@
 # -*- coding: utf-8  -*-
 
-from datetime import datetime
+from datetime import datetime, timedelta
+from gzip import GzipFile
+from json import loads
 from os.path import expanduser
+from StringIO import StringIO
+from urllib import quote
+from urllib2 import URLError
 
 from earwigbot.bot import Bot
 from oursql import connect
@@ -15,16 +20,80 @@ def _get_db(bot):
     args["autoreconnect"] = True
     return connect(**args)
 
-def _compute_stats(page, db):
-    with db.cursor() as cursor:
-        query = """SELECT COUNT(*) FROM templatelinks WHERE tl_title = ?
-                   AND tl_namespace = 10 AND tl_from_namespace = 0"""
-        cursor.execute(query, (page.title.replace(" ", "_"),))
-        transclusions = cursor.fetchall()[0][0]
+def _count_transclusions(cursor, title):
+    query = """SELECT COUNT(*)
+        FROM templatelinks
+        WHERE tl_title = ? AND tl_namespace = 10 AND tl_from_namespace = 0"""
+    cursor.execute(query, (title,))
+    return cursor.fetchall()[0][0]
+
+def _count_views(cursor, title, dbname):
+    query = """SELECT SUM(cache_views), MIN(cache_time)
+        FROM templatelinks
+        INNER JOIN {0}.cache ON tl_from = cache_id
+        WHERE tl_title = ? AND tl_namespace = 10 AND tl_from_namespace = 0"""
+    cursor.execute(query.format(dbname), (title,))
+    return cursor.fetchall()[0]
+
+def _get_avg_views(site, article):
+    url = ("https://wikimedia.org/api/rest_v1/metrics/pageviews/per-article/"
+           "{0}.{1}/all-access/user/{2}/daily/{3}/{4}")
+    days = 30
+    slug = quote(article.replace(" ", "_"), safe="")
+    start = datetime.utcnow().strftime("%Y%M%D")
+    end = (datetime.utcnow() - timedelta(days=days)).strftime("%Y%M%D")
+    query = url.format(site.lang, site.project, slug, start, end)
+
+    try:
+        response = site._opener.open(query)  # We're terrible
+    except URLError:
+        return None
+
+    result = response.read()
+    if response.headers.get("Content-Encoding") == "gzip":
+        stream = StringIO(result)
+        gzipper = GzipFile(fileobj=stream)
+        result = gzipper.read()
+
+    try:
+        res = loads(result)
+    except ValueError:
+        return None
 
-        # TODO
-        tif = 0.0
-        cache_time = None
+    if "items" not in res:
+        return None
+    return sum(item["views"] for item in res["items"]) / float(days)
+
+def _update_views(cursor, site, title, dbname):
+    cache_life = "7 DAY"
+    query1 = """SELECT tl_from
+        FROM templatelinks
+        LEFT JOIN {0}.cache ON tl_from = cache_id
+        WHERE tl_title = ? AND tl_namespace = 10 AND tl_from_namespace = 0
+            AND cache_id IS NULL
+            OR DATE_SUB(NOW(), INTERVAL {1}) > cache_time"""
+    query2 = """INSERT INTO {0}.cache (cache_id, cache_views, cache_time)
+            VALUES (?, ?, NOW()) ON DUPLICATE KEY
+            UPDATE cache_views = ?, cache_time = NOW()""".format(dbname)
+
+    cursor.execute(query1.format(dbname, cache_life), (title,))
+    while True:
+        titles = cursor.fetchmany(1024)
+        if not titles:
+            break
+
+        viewcounts = [(t, _get_avg_views(site, t)) for t in titles]
+        parambatch = [(t, v, v) for (t, v) in viewcounts if v is not None]
+        cursor.executemany(query2, parambatch)
+
+def _compute_stats(bot, db, page):
+    dbname = bot.config.wiki["_tifSQL"]["db"]
+    title = page.title.replace(" ", "_")
+
+    with db.cursor() as cursor:
+        transclusions = _count_transclusions(cursor, title)
+        _update_views(cursor, page.site, title, dbname)
+        tif, cache_time = _count_views(cursor, title, dbname)
 
     return tif, transclusions, cache_time
 
@@ -48,7 +117,7 @@ def calculate_tif(title):
         result["error"] = "no page"
         return result
 
-    tif, transclusions, cache_time = _compute_stats(page, db)
+    tif, transclusions, cache_time = _compute_stats(bot, db, page)
 
     result["tif"] = tif
     result["transclusions"] = transclusions