How to run a search strategy workshop
Goals first. Demo last. A session you can steal for a mixed room of exec, marketing, IT and CX.
Key takeaway
Six parts. Goals first. Demo last.
Three goals, and one they will actually protect.
Walk out with a goal sheet, a baseline, a decision and a 90-day plan. A feature list is not a strategy.
I was in as the search person. New customer. Strategy session. Around the table: an executive, digital marketing, a couple of data analysts, IT, customer experience. Someone had already opened a demo on the laptop. A person from marketing asked, quite nicely, when we’d look at the product.
I closed the lid.
It sounds rude. I don’t mean it that way. If I start with the path, marketing goes looking for campaign pins, IT asks where it would live, CX wants fewer tickets, the exec wants a decision they can take upstairs, and the analysts want a dashboard for a thing nobody has named. All of that is fair. It still isn’t a strategy.
I’m a solution architect at a search vendor. In that first meeting I want them to say what would be different if search worked, before I show them how.
Pick three. Star one.
I don’t ask “what are your search goals?” with a blank slide. People freeze, or they write twelve. I put a short list up and ask them to pick three, then star one. The starred one is the one we will not trade away this year.
They choose from this.
Visitors find what they came for (fewer empty results, fewer “I couldn’t find it” calls)
Search conversion, or finished tasks, beat the rest of the site
Campaigns show up in the box the week they go live
Someone in the business can change ranking, synonyms, pins, without a ticket
Support volume from “can’t find it” drops
The box still works on a busy weekend
We can see what people typed last week, and do something with it
A campaign or a catalogue drop does not mean starting search from zero
Three is workable, one is just a slogan, and five puts you back on a wish list. If they star “just add AI” or “pick a vendor by Friday”, I don’t argue. It tells me they haven’t got a search goal yet. We leave the laptop shut until they swap that for one of the eight.
A diagnosis, then a direction, then the work
The kernel of a strategy contains three elements: a diagnosis, a guiding policy, and coherent action.
Richard Rumelt, Good Strategy/Bad Strategy
A product demo is not a diagnosis. A roadmap with no star is just work for nobody in particular.
I keep the rest of the session in that shape. Six parts, same order, even when someone checks the time.
Goals. Done. Three, and one with a star.
Evidence. What did people type last month? How many of those came back empty? What turned into a ticket or a bounce? Who last opened that list? No queries means the next 90 days are about getting the queries. We are not picking a platform yet.
Constraints. What’s actually fixed for the next year. Go-live date. Peak weekend. Budget. Skills in the room. Who is allowed to change a synonym. Whether anyone can even get the query report.
Approach. This is when I show paths, and I only show three.
Keep the current box and run it properly: an owner, reports, synonyms.
Keep the current box and add a search product for the jobs it cannot do.
Replace the box.
They pick one for this phase. A later phase is fine. All three at once is not.
Who owns it after go-live. A week after launch, who can change a result, read the empty searches, and get a campaign term in without raising a ticket? Merchandiser, content, IT, the agency, or nobody. If it’s nobody, we need an owner before we need another feature.
The plan. Ninety days to stop the obvious bleeding. This year to move the starred goal. The rest (AI on top, the nicer modules) waits unless year one actually held. Every line has to point at a goal. If it doesn’t, it came off my slide and we cut it.
Five jobs, one brief (until they star something)
This room fails because five jobs arrive pretending they are one request.
The exec needs something that survives procurement. Marketing needs the campaign in the box. Analysts need a report they can stand behind. IT needs to operate it. CX needs the call that never happens. Skip the star and each of them leaves happy. Then you build a compromise nobody can run.
The owner question is where that gets honest. Search usually dies after launch, not in the workshop. Someone has to run the box the week after we leave, or this was just a nice meeting.
What I want on the table when we stand up
Not a sixty-page strategy deck. Four things.
A goal sheet: three goals, one starred, in their words.
A baseline: actual numbers, or a gap written down (we have no query report).
A decision: the approach, a named owner, and what we are not doing this year.
A roadmap: ninety days and twelve months. Each line has a date, a person, and what “done” looks like.
The measures live on the roadmap. I don’t bury them at the back. Empty results heading under 5%. Search conversion against the rest of the site. Campaign terms in the top queries the week they go live. Tickets down. A report someone in the business can open without asking IT.
If those four are there, they have a search strategy. If they only have a roadmap, they have a build plan for a box.
If the starred goal still needs a product after the diagnosis, we can open it. Opening it first would have been a sales meeting with extra chairs.
Someone always looks at the closed laptop like I’ve hidden the interesting bit. The interesting bit is the board.
The hour before the demo is the one I keep fighting for.
