A WhatsApp photo of a dial, from somewhere
Your client has dozens of sites, many on meters that are not read automatically. Each month, your team asks site managers for reads. Some reply with a photo of the meter, with no site name. One sends a photo of the wrong meter. One types the read into an email, and a digit is missing. Several do not reply at all, so those sites are estimated again.
Your analyst works through the photos, matches them to sites by guesswork, types the reads into a spreadsheet and submits them to suppliers, one portal at a time.
Why reads are so hard to collect
Meter reads depend on people at the site, who have no reason to prioritise them.
- Site staff change, and new staff do not know where the meter is.
- Requests go to a general inbox or a manager who passes them on.
- Photos arrive without a site or meter reference.
- Reads are not checked against previous reads before submission.
- Submission is done manually through supplier portals or email.
Estimated bills and wasted analyst time
Missing reads lead to estimated bills, which may be far from actual use and later lead to catch-up bills that clients do not expect. Wrong reads cause the same problem in a different way.
The analyst time spent chasing, matching and typing reads is considerable, and it is repeated every month for every client.
It also weakens the rest of your service. Bill validation against estimated reads tells you little, and reports built on estimates can mislead the client about how a site is actually using energy. The quality of a read collected by a busy shop manager on a Tuesday ends up limiting the quality of everything your analysts produce from it.
A read collection flow for site staff
We build a simple flow that site staff can complete in a minute.
- Each meter has a link or QR code, which can be stuck by the meter, that opens a read page already labelled with the site and meter.
- Site staff get a reminder by text or email on the day reads are due.
- They photograph the meter, and the read is extracted from the image with a vision model. They confirm or correct it.
- The read is checked against the meter's number of dials and previous reads. Anything unusual, such as a lower read than last time, asks the person to check again and is flagged for your team.
- Clean reads are submitted to suppliers where an API or accepted file format exists, or queued for a person to submit through the portal.
- A dashboard shows which sites have submitted, which are late and which were flagged.
| Step | Email and photos | Read collection flow |
|---|---|---|
| Request | Email to the site manager | Reminder to named site staff |
| Which meter | Guessed from the photo | Labelled by the link |
| Read value | Typed by the analyst | Extracted and confirmed on site |
| Checks | None | Against dials and past reads |
| Missing reads | Noticed at submission | Dashboard of late sites |
Where supplier portals do not accept automated submission, the flow prepares everything so the manual step is quick.
Reads that arrive labelled and checked
On read day, site staff get a reminder, scan the code by the meter and take a photo. The read is confirmed on the spot. By the end of the day, most sites are in, labelled and checked, and your analyst deals only with the few that were flagged or are late.
Over time, fewer bills are estimated, and those that are can be traced to the specific sites that did not submit.
Are meter reads a monthly chase?
- Site staff send meter photos without labels.
- Reads are often late or missing.
- Wrong reads have been submitted.
- Analysts type reads into spreadsheets.
- Estimated bills are common across client sites.