Best strategies for boosting Salesforce user adoption in large teams

The adoption chart is green. Logins for the quarter cleared the target the programme set in the kickoff. The pipeline review still starts with an export. Opportunities have sat in the same stage since last quarter, with a next step that says "follow up" and no date. A case comes back to the writer because the resolution is "done", and the next person cannot tell what was done. On a site, a technician texts the office the closure code, because the mobile page does not have it. The office types it in on Friday, into the wrong field. The login chart does not move when any of this happens. It already moved on Monday, when people opened the tab.

Large teams adopt a page that makes the job faster. They ignore a page that asks them to care more about the same clutter. Measure the work. Then change the screen. Then train. A login target reverses that order and produces a chart of people who opened a tab.

#Measure the process, not the visit

Pick two facts the team already agrees are the job. Do not invent a third to make a dashboard look complete. Stage hygiene: an opportunity has a next step written in words a colleague can act on, and a date, and it does not sit in one stage for a quarter because nobody wanted to be the person who moved it. Case quality: a subject a stranger could understand, the product, and a resolution the next person can act on without calling the writer back. If the team argues about what those mean, settle the argument before you publish a number. An unsettled definition will be gamed the same way a login target is gamed, just with more typing.

Publish those two weekly to the manager who owns the team. Name the manager. A global adoption mailbox will not change Tuesday. The manager who runs the pipeline meeting will, if the number is in the meeting and the export is not. Show the records, not a percentage of "compliant users." Ten opportunities with no date is a conversation. A score of 72 is a debate about the score.

Login counts fail because they pay the wrong action. A person can log in, change nothing, and help the chart. A person can open the tab, click around, and leave the stage untouched. A person can do the work in a spreadsheet and never appear, which means your best record-keeper is invisible and your least careful person is a success. If you need a system measure, count records updated with the fields the process requires, in the week the work happened. Sessions started are not that count. Last-login reports are not that count. They answer whether the password works.

Watch the gaming that replaces the login game once you switch measures. A next step of "follow up" on every opportunity, dated today and pushed forward each morning by a flow, will make stage hygiene look perfect. Read ten next steps. If they are the same sentence, the measure has been automated into uselessness, and the flow is the defect. A resolution of "fixed" on every case is the same trick. The definition has to be readable by the next person, and the check is to ask that person, once a week, whether they could act.

#The page changes, or the course will not

Remove the fields nobody fills. You can see them: the column that is blank on the last fifty records, the picklist with a value called Other that holds the real answer, the third section below the fold that was added for a project in a year nobody remembers. Put the required fields where the work starts. The next step and the date sit at the top of the opportunity, not under a related list of historical activities. The product and the resolution sit where a case is written, not on a tab people open after they have already saved.

Required means the save blocks when they are empty, on the path the person actually uses. A red asterisk on a page layout blocks the screen. It does not block an integration, a data load, or an action that writes the record another way. If those paths exist, a validation rule or a before-save check has to carry the same demand, or the page will look strict while the empty records keep arriving from the side door. Say which path you are blocking. Then train on that page, with the team's own records, not a demo account whose fields were filled in advance.

A course on last year's layout asks people to memorise a screen that still fights them. They will pass the quiz, because the quiz asks what the slide said, and they will go back to the workaround, because the workaround is faster. Do not schedule the course to make the kickoff date. Move the fields first. If you cannot move them before the training week, move the training week. The calendar is cheaper to change than a room full of people learning the wrong page.

Salesforce describes the employee assistant in one sentence: "Agentforce (formerly Einstein Copilot) supports employees in the flow of their work with specific tasks by searching for data, creating action plans, and executing them to increase workplace efficiency." Read that as a product description, not as a result you have measured. An assistant that fills a cluttered layout will write into the wrong fields faster. Efficiency against a bad page is more bad data per hour. Fix the layout before you add anything that completes it for the user. If the next step is in the third column, the assistant will put something in the third column, confidently, and the stage hygiene number will improve while the pipeline gets worse.

#Field teams are a different count

Salesforce describes the field product this way: "Agentforce Field Service and Operations (formerly known as Field Service) gives you a powerful, highly customizable, mobile-friendly field service hub in Salesforce." For people on site, adoption is whether the work order was updated on the device before they left. A login at a desk on Friday does not tell you that. The Friday login tells you they had signal and a password at a desk. The job was on Tuesday, in a place where the page either worked or it did not.

If the mobile page does not carry the fields the job needs, they will text the office, and the office will type it in later, wrongly, or into a notes field the report does not read. That text is the measure failing. Count the texts, or sit with the dispatcher and count the "can you update this for me" messages, for one week. The number is the adoption figure for that team. It will not match the login chart, and the login chart is the one that is wrong.

Sit in a van, or on a site, and watch one job from arrival to the moment they would leave. The fields they skip are the fields to remove or to move. The fields they ask the office for are the fields missing from the mobile page. Do not add those fields to a desktop layout and call the mobile problem solved. They are not at the desktop. A long page that requires scrolling with gloves on, or a picklist that needs a search they cannot spell, will be skipped even when it is "on the device." Shorter, and in the order of the job, is the design. Then watch a second job. If the update happens before they leave, you have a page. If it does not, you have another field to move, and a course will not paper over it.

#Managers go first

The person who reviews the pipeline or the queue has to use the same page. If they keep a side spreadsheet, the team will feed the spreadsheet, because that is what the meeting rewards. You can require every field in the org and still lose to a manager who says "just send me the sheet." Sit with five managers before you train their teams. If they export before the Monday meeting, you have an adoption problem no login target will fix. Change the meeting so the page is on the screen and the export is not accepted. Or stop calling the export temporary. It is the system of record, and the CRM is the archive. Pretending otherwise wastes the training budget and insults the people who already knew which file the meeting uses.

This is also how a global rollout lies to itself. A country can be "live" on a slide while the country manager still closes the month from the old file. The cutover is not finished until that file is the page, which is why the report the manager actually opens belongs on the list in a rollout sequenced by country. Adoption work that ignores that report will celebrate logins in a country that has not moved.

Do this for one team until the two measures move, for four weeks, with the page already changed and the manager already in the meeting without the export. Then the next team. Copy the page changes that worked. Do not copy the kickoff video. A launch that asks every region to care more, on the same day, with the same video, is how adoption programmes get a kickoff and no second month. The second month is the point. If the two measures did not move, the page is still wrong, or the meeting still rewards the sheet. Fix one of those. Do not add a reminder email.

Cheap, before the course is booked: deleting fields, moving the next step up, telling a manager the export will not be accepted, watching one job on site. These are days, not a programme.

Expensive, after a year of login targets: a team that has learned to open the tab and keep the real work elsewhere, and a manager who no longer believes a number from the CRM. Rebuilding that trust starts by throwing the login target away in public and putting the ten bad opportunities on the table. People will believe the next measure if it matches what they see in the meeting. They will not believe a third chart that sits beside the export.