What happened
What this says
As at Fri 9 Oct 2026The search returns partial matches alongside exact matches, forcing branch teams to sift through multiple results to find the right account. Connor investigates and finds the search tokenizer is splitting company names and returning any token match without exact-match ranking.
Problem: Search tokenizer splits company names into tokens and returns results for any matching token without prioritizing exact matches, making search inefficient.
Handling: Deepa reported the issue affecting daily workflow; branch teams must sift through false hits. Connor reproduced it in testing and identified the root cause: the algorithm needs to weight exact matches higher. Fix is ready to deploy.
Watch: Poor search ranking degrades usability and drives teams back to manual processes.
Details
Updated 9 Oct 26Trail
- created · Deepa Chauhan 23 Feb 2026 09:05Hi Connor, we're having trouble with the account search. When the branch teams search for an account by name, they're getting back results that don't match exactly what they typed. Searching for a specific company name returns not just that account but several others with similar names. The search should give exact matches first or filter more strictly. We're getting too many false hits and it's slowing the teams down. Can you look at this? Thanks, Deepa.
- Public reply · Connor Webb 23 Feb 2026 12:11Hi Deepa, thanks for the report. I can see how that would be frustrating. I'll need to dig into the search configuration to see what's happening. Can you give me an example of one search term and what results you're seeing? That'll help me pinpoint the issue. Cheers, Connor.
- Public reply · Deepa Chauhan 25 Feb 2026 11:57Sure. I tried this morning and when searching for a specific company name, the results include that account plus two or three others with similar names. Each time it's adding partial matches when it should be showing exact matches first. The branch teams need to sift through all the similar names to find the right one.
- Internal note · Connor Webb 25 Feb 2026 12:13Reproduced in test environment. The search tokenizer is splitting company names and returning results for any matching token, not enforcing exact-name-first ranking. The CRM data has many similar names so this breaks workflow. Fix: adjust search algorithm to weight exact matches higher. Will test and deploy.
- Status change · 16 Mar 2026 23:59