Person

Six Weeks Inside a 40-Person Local

AC Adrienne Cole

In February a building trades local in Baltimore County called us because dues reconciliation was taking 19 days every cycle and their treasurer was ready to quit. They had a quote in hand for a membership platform that would have cost 84,000 dollars over three years. We spent six weeks with them and they did not buy the platform. This is the whole engagement written up with the client's permission, including the two things we got wrong. The numbers are theirs and they checked them before this went up.

01 What we found in the first two weeks

Two people were doing the same data entry, neither knew about the other, and the reason traced back to a report that was retired in 2016.

The duplicate nobody had noticed

One payment file was being keyed into two systems by two people who sat on different floors, costing roughly nine hours a cycle.

The bank file came in weekly. The office manager keyed it into the accounting package. On another floor, the membership clerk keyed the same file into the membership database, because until 2016 a nightly report had pushed those figures across and when the report was discontinued nobody told her she could stop. Nine years. About nine hours a cycle between them, plus every mismatch between the two sets of numbers had to be chased down by the treasurer, which was the actual source of the 19 days. Neither of them was doing anything wrong. Both were doing exactly what they had been trained to do. This is the normal shape of the problem, and it is why we insist on sitting with the people at the keyboard rather than interviewing the director.

The current state map

One page, 31 steps, four of them waits longer than a day. We put it on the wall and the room went quiet.

We map on paper first, in the room, with the people who do the work correcting us as we draw. Thirty-one steps from bank file to reconciled ledger. Four waits longer than a day, three of them waiting on one person to approve something they later told us they did not think they were supposed to be approving. The map fits on one page deliberately. Once it needs two pages people stop reading it and start receiving it, and a document you receive is not a document you argue with. We want the argument. The argument is where the corrections come from. We left the paper version taped to the wall of their break room for the whole engagement. Two of the best changes on the final list came from members who read it on a coffee break and told us it was wrong.
02 The six changes

We cap the change list at six. Not because six is magic, but because a list of twenty gets filed and a list of six gets done.

Why the list is short

Long recommendation lists are how consultants transfer responsibility back to the client while appearing thorough.

We could have written thirty findings. We have written thirty findings before, at previous jobs, and I know what happens to them. Three get done, the rest sit in the shared drive, and in two years somebody pays for the same analysis again. So the rule is six, each with a named owner inside the organisation and a date. Not a phase, not a workstream. A person and a date. For this local the six were: stop the duplicate entry, fix the accounting export so it produces the membership file directly, move approval of routine payments below 500 dollars to the office manager, delete two fields nobody had populated since 2019, put the arrears list on a Monday schedule instead of end of month, and write down the reconciliation procedure, which had never existed in writing. All six were done by week six. Five were done by the existing staff with us watching.

The two we got wrong

We proposed a dashboard nobody wanted, and we badly underestimated how long the export fix would take their vendor.

First mistake. In week three I proposed a small reporting dashboard for the executive board. It was a good dashboard. Nobody asked for it, the board reviews numbers once a quarter on paper, and I built it because building things is more fun than not building things. We dropped it in week four and I did not charge for the time. Second mistake, more serious. The export fix required a change from their accounting vendor. I scoped that as two weeks. It took eleven, which is well past the end of our engagement, and for nine of those weeks the office manager was still keying by hand. It worked in the end but the local carried that gap alone. We now ask about vendor change request timelines in the first week and we assume the answer we are given is optimistic by half.

The blank version of the change list we use, one row per change, owner and date required. Free to copy, no signup.

🔗Change list templatefairmeasure.coop↗
03 Where it landed

Nineteen days to four, no new software, total cost 6,400 dollars against an 84,000 dollar three-year quote they did not sign.

Six months on

We called them in August. Still four days, procedure document still in use, and the treasurer has not quit.

We call every client six months after the engagement ends, because a result that does not survive a quarter is not a result. As of August the cycle is holding at four days, occasionally five when a holiday falls badly. The written reconciliation procedure has been updated twice by their own staff, which is the single best sign we have ever gotten. The membership clerk who spent nine years on duplicate entry now runs the arrears outreach, which she says she prefers and which has brought in about 11,000 dollars in back dues. They have not bought the platform. They may need something eventually, and if they do they will now be able to describe exactly what it has to do, which is worth more than the software.

Our full rate card and the sliding scale criteria, published because members should not have to ask what analysis costs.

🔗How we pricefairmeasure.coop↗