Omniture for Large Sites: Governance Matters. If you run a big site and you bought Adobe SiteCatalyst because someone said it will answer everything, this is for you. The tool is powerful. The mess starts when there is no clear way to run it. At scale the biggest problem is not the tag or the report. It is the rules, owners, and habits that shape what gets tracked and how it changes over time. That is governance, and it decides whether your data earns trust or becomes noise.
Problem framing
On smaller sites you can survive with a few eVars, some friendly sProps, and a mental map of what they mean. On a large site with many teams, releases every week, and a long list of agencies touching tags, that falls apart. Naming drifts. Expiries are wrong. Allocations fight each other. Campaign codes multiply. You wake up with five report suites telling five different stories and no one can sign off a revenue number. The tool did not fail. The way you run it did.
Good governance is not a giant binder. It is a set of simple rules you always follow. A shared variable map. A change log. A data layer contract. Owners for each report suite. A weekly check in to review changes. Sounds boring. Saves launches.
Three cases from the trenches
Case one: the news network with drifting props. A publisher ran eight sections on one domain. Each section had its own dev team. They shared a report suite but used different names for the same ideas. One team used sProp5 for author, another used sProp5 for topic. Trending across the site was broken. The fix was not a new plug in. We created a simple taxonomy for content types, authors, and sections. We locked variable ownership, moved author to sProp3 across the board, and mirrored it to an eVar for pathing with a short expiry. Processing Rules enforced defaults when the page was missing the value. A two page variable map and a thirty minute review each week turned a noisy dashboard into a useful one.
Case two: the commerce group with campaign soup. A retailer with three brands had campaign tracking codes like email spring, eml sprg, and EML 2012 04. Attribution was guesswork. We stood up a single campaign code dictionary and used SAINT classifications to add Channel, Subchannel, Creative, Country, and Owner. In SiteCatalyst 15 we switched to Marketing Channels so the same rules scored traffic the same way across brands. We set channel override rules, last touch and first touch windows, and cut a monthly export to finance. The tip that won trust was boring too. We set a rule that no code ships without a ticket and a link to the code in the dictionary. After two sprints, CPA reports stopped arguing with each other.
Case three: the global site with tag chaos. A multinational had a tag for everything. Video player, survey, a chat tool, three ad networks, and two analytics tools. Page weight was painful and privacy counsel was nervous. We did an audit, moved all vendors into a single tag manager, and built a publish schedule with a hard gate. Only one window each week for tag changes unless it was a severity one bug. We added a simple data layer so teams stopped scraping the DOM. We filtered internal IPs, turned on bot filtering, and added a synthetic test to check that s_vi and s_fid cookies were set before a deploy went live. Uptime improved and the privacy doc finally matched the real world.
Objections and replies
We are too busy shipping. If you ship without a gate you ship bugs you cannot see. A five minute check in saves days of arguing about why orders do not match. Make it part of the release checklist like unit tests.
A tag manager will fix it. A container helps, but it does not decide what eVar7 means or when Campaign expires. Tools move faster when rules are clear. Start with the variable map, then pick your container.
Analysts can fix it in Excel. Short term yes. Long term no. If you keep fixing in spreadsheets, different teams will fix in different ways. Your north star goes dim. Fix the source. Use SAINT and Processing Rules so cleanup sticks.
Governance slows testing. The right rules speed tests. When you decide in advance how to name experiments, which event counts a conversion, and where to store the variant, your test launches faster and your post mortem is boring in a good way.
Regions need freedom. They do. Give them freedom inside a frame. Keep a global set of reserved variables and a small pool of local ones. Keep common expiries and events. Let regions define their own classifications inside that space.
Action steps you can take this week
- Name an owner for each report suite. Put the names in a simple doc that everyone can see.
- Publish a one page variable map. For each prop and eVar list name, allowed values, expiry, allocation, and owner. No fluff.
- Set Marketing Channels. Agree on rules for Paid Search, Natural Search, Email, Display, Social, Referral, and Direct. Lock the windows and stick to them.
- Create a campaign code dictionary with SAINT. One code per effort, shared across teams. Add Owner and End Date.
- Stand up a change log. A simple spreadsheet works. Date, change, who, why, link to ticket, and rollback plan.
- Write a data layer spec. Page name, content type, product id, user status, campaign code. Keep it small, keep it stable.
- Automate a smoke test that loads a key page and checks for s_vi or s_fid, report suite id, and a couple of expected variables. Run it before deploys.
- Meet for twenty minutes each week. Review upcoming changes, new campaigns, and any broken trends. No slides. Just decisions.
- Loop in privacy early. Confirm no PII goes into props, eVars, or traffic variables. Respect Do Not Track and cookie consent where required.
None of this needs a big project. You can start small, and your data will feel cleaner within a sprint. The trick is to treat Omniture governance like you treat code. Version it. Review it. Own it. SiteCatalyst can support a huge site if the rules are clear and the people care about them.
If you are staring at a dashboard that argues with your finance sheet, start with the map. If your campaign report looks like alphabet soup, start with the dictionary. If your tags change every day, start with the gate. Pick one, fix it, and keep going. Your teams will spend less time debating numbers and more time making changes that matter.
Bottom line: the tech is solid. The win comes from governance that scales. If you want a template for the variable map and the change log, ping me. Happy to share what works.