Hero CTA moved from position 1 to position 4 in the layout — visits held steady, but the button that used to convert stopped getting clicked.
Before 2,841
After 2,693
Change −5.2%
Window Apr 14 – May 3
Layout change
/Features / Experiments
Made a change to your site?
Find out if it worked.
Every experiment runs on a fixed window you pick — 7 to 30 days. If Wiley already tracks the metric involved, you get a real before/after number when it closes. If it doesn't, you're the instrument — Wiley still holds you to the hypothesis and reports back honestly either way.
How experiments actually work
1
Something starts it
Click "Run an experiment" and Wiley's AI looks at your real traffic, rankings, and site issues to propose one — or launch straight from a Connection Wiley already flagged. Either way you land on a proposal: a hypothesis, something to watch, and a duration to pick.
→
2
Wiley decides what it can measure
If the metric is something Wiley already tracks — page views on a specific page, a goal event — it snapshots a real baseline the moment the experiment goes active. If it isn't (true of most Connection-launched experiments, since they're tied to whatever the Connection flagged rather than a clean metric), the clock still starts — there's just no number behind it yet.
→
3
You get a report when the window closes
Every experiment runs for the fixed duration you picked up front — 7, 14, 21, or 30 days, no early cutoff. When it ends, Wiley reports a real before/after number if it had one to track, or an honest reflection if it didn't. Either way, it says plainly whether the hypothesis held, partially held, or if there's just not enough there to say.
Two ways to start one
One you ask for, one Wiley already teed up — both end in the same kind of experiment, run the same way.
On-demand
Ask Wiley to propose one
Click "Run an experiment" in your dashboard — it uses a shared weekly credit, the same pool as on-demand nudges. Wiley looks at your traffic, rankings, PageSpeed, and crawl findings and proposes a single experiment, favoring one it can track automatically over one you'd have to watch yourself. It lands as pending — accept it to start the clock, decline it, or leave it for up to 30 days.
From a Connection
Launch straight from what Wiley flagged
See a Connection worth testing further? Click "Run an experiment" on the card. Wiley carries over its context and the metric it detected movement in — you describe what you're actually going to do and pick a duration, then it starts immediately. Since these are tied to a specific change rather than a metric Wiley already tracks, they run as observation-based experiments: Wiley won't invent a number it doesn't have, but it will hold the hypothesis to account when the window closes.
Started — the metric to watch came from this Connection. Wiley's watching the clock; you're watching the result.
What experiments tend to test
Whether Wiley proposed it or you launched it from a Connection, this is the range experiments tend to cover. Some of these Wiley can measure automatically; most can't.
A quick note on measurement: Wiley automatically tracks page views on a specific page and goal events — that's it. Ranking movement, structured-data uptake, and page-speed shifts are all real signals worth checking (Wiley tracks rankings and PageSpeed separately, just not as part of an experiment's before/after), but for those, an experiment works more like a discipline device — a fixed window to hold yourself to a hypothesis, with Wiley's closing report helping you reflect honestly on what you actually saw.
Rewriting a page
Updated a blog post, rewrote a services page, added a new section — anything that meaningfully changes what a visitor sees or what search engines read.
Adding structured data
Pasted in a JSON-LD block that Wiley suggested, or added schema markup yourself. There's no click-through number Wiley auto-tracks for schema, so this usually runs qualitative — but rich-result or ranking changes are typically visible fast if you check yourself.
Targeting a new keyword
Rewrote a title or heading to target a different phrase. This runs qualitative — Wiley tracks your rankings separately (see Connections), but an experiment's own before/after doesn't include rank movement, so check there and use the report window as a prompt to reflect on whether the shift helped.
Freshening up a stale page
Updated a post that hasn't been touched in over a year — added a new section, refreshed the date, made it relevant again. If Wiley's watching page views on that path, you'll get a real number; otherwise it's one to reflect on.
Fixing broken links
Fixed internal links that were returning errors. Worth tracking if it's tied to a page Wiley already measures page views on — otherwise it's one to watch and report back on yourself.
Improving page speed
Made changes to load time — optimized images, removed a heavy script, switched to a faster host. Wiley already tracks PageSpeed separately, so a real change should show up there even though this kind of experiment usually runs qualitative.
What an experiment report looks like
Wiley doesn't just show you numbers — it explains what moved, what didn't, and what to do next. If it was tracking something automatically, you get a real number. If it wasn't, you get an honest reflection instead — never a number Wiley didn't actually collect.
From Wiley Experiment complete: /services page rewrite — 14 days in
You started tracking page views on /services 14 days ago, right after the rewrite.
Traffic: 340 views in the two weeks before the change, 452 views in the two weeks since — up 33%. That's real, if modest, movement.
It's not dramatic enough on its own to call the rewrite a clear win, but it's a genuine lift worth keeping. If you want a cleaner read next time, structured data on this page is still untested — nothing Wiley tracks automatically there, but worth a look.
From Wiley
On-demand requests and Boosts
Every experiment runs on its own fixed clock — Wiley reports back automatically when the window closes, whether or not it had a number to track. But sometimes you want to start one, or ask about one, right now instead of waiting.
That's what the on-demand "Run an experiment" button is for — one click, using a shared weekly credit (the same pool as on-demand nudges). Wiley proposes one from your real data and you accept, decline, or hold onto it for later.
Boost is separate: a monthly credit for a back-and-forth conversation with Wiley — ask about anything on your site, including whether an active experiment is worth watching more closely. A Boost conversation can also end in a proposed experiment if that's the right next step.
Credits, at a glance
On-demand request
"Run an experiment" or "Get a nudge" — one click, shared weekly credit.
Boost
A monthly credit for a longer conversation — can end in a proposed experiment.
Dashboard history
All experiment results stay in your dashboard — not just the nudge that reported them.
Common questions
What counts as an experiment in Wiley?
Anything you're deliberately testing — a change launched from a Connection, something Wiley proposed after looking at your data, or one you asked for. Every experiment gets a hypothesis, something to watch, and a fixed window (7, 14, 21, or 30 days) before Wiley reports back.
How does Wiley measure whether an experiment worked?
It depends on what's being tested. If the metric is one Wiley already tracks automatically — page views on a specific page, or a goal event — it snapshots a real number when the experiment starts and another when it ends, and reports the actual before/after. If it isn't (true of most experiments launched from a Connection, since they're tied to whatever specific change the Connection flagged rather than a clean metric), there's no number to snapshot — Wiley still reports back at the end, but as an honest reflection, not a data comparison. It won't invent a number it doesn't have.
Can I run multiple experiments at the same time?
Up to three active at once. A fourth has to wait until you complete one, decline one, or let one expire. A proposal you haven't started yet doesn't count against that cap, but it does expire after 30 days if you never act on it.
Does Wiley require the tracking script to run experiments?
Only for the experiments Wiley can measure automatically — page views on a specific page and goal events both come from the tracking script's data. Anything else, script or not, runs qualitative: you're the one observing the result, and Wiley's report is a prompt to reflect honestly, not a number.
Start tracking what you change
and whether it works.
Experiments included on all paid plans. 7-day free trial, cancel anytime.