Read temporary power meter from a photo: WhatsApp to Google Drive and Sheets (with OCR)

Quick answer

You can let operatives WhatsApp a clear photo of the temporary power meter. An automation saves the image into Google Drive, reads the digits with OCR, and logs the reading into a Google Sheet for billing and audits. If confidence is low or the number drops below the last reading, the bot asks for confirmation.

• Capture: WhatsApp Business API (Twilio, MessageBird or 360dialog)
• Store: Google Drive (job/site folder)
• OCR: Google Cloud Vision Text Detection
• Log: Google Sheets row with timestamp, site code, meter serial, register (e.g., 1.8.0), kWh, link to photo

Who this is for

• Small contractors, temp power users, facilities teams and subcontractors who need monthly or weekly meter readings without sending someone to site with a clipboard.
• Anyone building a Building Safety Act evidence trail and wanting dated photos plus a clean log.

How it works

• Operative sends a meter photo to your WhatsApp Business number, optionally with a caption like: SITE=E14 Docklands | METER=K123456 | REG=1.8.0.
• Your WhatsApp provider triggers a webhook in Make.com or Zapier.
• The image is stored in Google Drive using a consistent path, then Google Cloud Vision extracts the numbers.
• A Google Sheet logs the reading with the photo link. If the OCR confidence is low, the bot replies on WhatsApp asking the user to confirm or resend a clearer photo.

Simple flow

• WhatsApp photo
• Google Drive upload
• OCR reading
• Append to Google Sheets

Step-by-step setup in Make.com

Prerequisites

• WhatsApp Business API number via Twilio or 360dialog
• Google Workspace account
• Google Cloud project with Vision API enabled
• A Google Drive root folder, e.g. Drive/Construction Power
• A Google Sheet called Meter Readings Log

Steps

1. Trigger: WhatsApp New Media Message

• Provider: Twilio or 360dialog
• Output fields: media URL, sender, caption, timestamp

2. Download media

• HTTP Get a File from media URL

3. Upload to Google Drive

• Path: Drive/Construction Power/{{SITE}}/{{METER}}/{{YYYY-MM}}/
• Filename: READING_{{YYYYMMDD-HHMMSS}}_{{phone}}.jpg
• Save fileId to pass to later steps

4. OCR with Google Cloud Vision

• Feature: TEXT_DETECTION
• Parse output for the register and the numeric value (4–8 digits). Keep the confidence score

5. Build data

• Fields: timestamp_utc, site_code, meter_serial, register_code, reading_kwh, ocr_confidence, whatsapp_number, image_drive_link (fileId URL)
• If caption is missing data, try to detect site/meter from folder path or a small reference list

6. Append to Google Sheets

• Sheet: Meter Readings Log
• Add a header row once: Timestamp, Site, Meter Serial, Register, Reading kWh, Confidence, WhatsApp, Drive Link

7. Validate and reply on WhatsApp

• Look up the last reading for that meter in the Sheet
• If reading < last or confidence < 0.7, send a WhatsApp reply: “I read 012345 for meter K123456 (1.8.0). Reply Y to confirm or send a clearer photo.”
• If user replies Y, mark reading_source=user_confirmed

8. Notify exceptions (optional)

• Send Slack or email if reading jumps more than your threshold (e.g., > 2x last period)

Code-first option (Cloud Run)

If you prefer a small codebase for more control, host a webhook on Cloud Run.

What the service does:

• Receives the WhatsApp webhook JSON, downloads the photo
• Uploads to Drive, calls Vision, parses text
• Checks against last reading in the Sheet
• Writes the new row and replies to the user for low-confidence cases

Minimal pseudocode outline

app.post('/whatsapp-webhook', async (req, res) => {
  const { mediaUrl, from, caption, timestamp } = parseWhatsApp(req.body);
  const image = await fetch(mediaUrl).then(r => r.arrayBuffer());
  const fileId = await drive.upload(image, makePathFrom(caption));
  const vision = await cloudVision.text(fileId);
  const parsed = parseMeterText(vision.text);
  const last = await sheets.getLastReading(parsed.meterSerial, parsed.register);
  const needsConfirm = parsed.conf < 0.7 || parsed.value < last;
  await sheets.append({ timestamp, ...parsed, from, fileId });
  if (needsConfirm) await whatsapp.reply(from, `I read ${parsed.value} for ${parsed.meterSerial} (${parsed.register}). Reply Y to confirm.`);
  res.sendStatus(200);
});

Data model and folder structure

Recommended Google Sheet columns

• Timestamp (UTC and local)
• Site code
• Meter serial
• Register code (1.8.0 import total; 1.8.1 day; 1.8.2 night; 2.8.0 export)
• Reading kWh
• Reading source (ocr, user_confirmed, manual)
• OCR confidence (0–1)
• WhatsApp number
• Drive link
• Notes

Drive structure

• Drive/Construction Power/SITE_CODE/METER_SERIAL/YYYY-MM/
• Keep raw photos immutable. Use Drive permissions, not public links

Quality checks for UK meters

Reduce misreads and get consistent results on UK sites:

• Teach operatives to select the correct register before taking the photo
• Shoot square-on, no glare, fill the frame with the LCD or odometer
• For cycling displays, wait for the exact register screen
• For dial meters, add a text message with the reading as a backup
• Ask for a WhatsApp location share to tag the reading to the site

Useful guidance

• OBIS codes explained: 1.8.0 import, 2.8.0 export, 1.8.1 day, 1.8.2 night
• Time-stamps: phone EXIF plus WhatsApp timestamp; keep both in the log
• Proof for clients: link the Sheet and Drive folder in your monthly application for payment

Costs and limits

• WhatsApp Business API: typically pence per conversation depending on provider
• Make.com or Zapier: plan-based operations pricing
• Google Cloud Vision: pay per image; low cost at small volumes
• Google Drive/Sheets: included with Workspace

Troubleshooting

• Photo is blurry or glare: reply asking for a new shot, remind user to turn off flash
• OCR picks the wrong number: add a rule to ignore numbers with decimal points if your meter should be whole kWh
• Reading lower than last month: flag for review, possible meter rollover or wrong register
• Multiple registers: add a short menu in WhatsApp or use keywords in the caption like REG=1.8.0

FAQ

Is this acceptable for invoicing clients?

Yes, many clients accept dated photo evidence and a consistent log, especially for temporary builders’ supplies. Check your contract requirements and keep the original images versioned in Drive.

What about data protection?

Use a dedicated WhatsApp Business number, restrict Drive access to your team, and include this workflow in your staff privacy notice. Avoid public links.

Do I need a developer?

No. Make.com can handle the full flow. If you have higher volumes or complex validation, a small Cloud Run service gives more control.